a simple guide to the analysis of social science quantitative data
DESCRIPTION
One of the complexities for many undergraduate students and for first time researchers is ‘How to blend their socialization with the systematic rigours of scientific inquiry?’ For some, the socialization process would have embedded in them hunches, faith, family authority and even ‘hearsay’ as acceptable modes of establishing the existence of certain phenomena. These are not principles or approaches rooted in academic theorizing or critical thinking. Despite insurmountable scientific evidence that have been gathered by empiricism, the falsification of some perspectives that students hold are difficulty to change as they still want to hold ‘true’ to the previous ways of gaining knowledge. Even though time may be clearly showing those issues are obsolete or even ‘mythological’, students will always adhere to information that they had garnered in their early socialization. The difficulty in objectivism is not the ‘truths’ that it claims to provide and/or how we must relate to these realities, it is ‘how do young researchers abandon their preferred socialization to research findings? Furthermore, the difficulty of humans and even more so upcoming scholars is how to validate their socialization with research findings in the presence of empiricism. Within the aforementioned background, social researchers must understand that ethic must govern the reporting of their findings, irrespective of the results and their value systems. Ethical principles, in the social or natural research, are not ‘good’ because of their inherent construction, but that they are protectors of the subjects (participants) from the researcher(s) who may think the study’s contribution is paramount to any harm that the interviewees may suffer from conducting the study. Then, there is the issue of confidentiality, which sometimes might be conflicting to the personal situations faced by the researcher. I will be simplistic to suggest that who takes precedence is based on the code of conduct that guides that profession. Hence, undergraduate students should be brought into the general awareness that findings must be reported without any form of alteration. This then give rise to ‘how do we systematically investigate social phenomena?’TRANSCRIPT
A Simple Guide to the Analysis of Quantitative Data
An Introduction with hypotheses, illustrations and references
By
Paul Andrew Bourne
A Simple Guide to the Analysis of Quantitative Data: An Introduction with hypotheses, illustrations and references
By
Paul Andrew BourneHealth Research Scientist, the University of the West Indies,
Mona Campus
Department of Community Health and PsychiatryFaculty of Medical SciencesThe University of the West Indies, Mona Campus, Kingston, Jamaica
2
© Paul Andrew Bourne 2009
A Simple Guide to the Analysis of Quantitative Data: An Introduction with hypotheses, illustrations and references
The copyright of this text is vested in Paul Andrew Bourne and the Department of Community Health and Psychiatry is the publisher, no chapter may be reproduced wholly or in part without the expressed permission in writing of both author and publisher.
All rights reserved. Published April, 2009
Department of Community Health and PsychiatryFaculty of Medical SciencesThe University of the West Indies, Mona Campus, Kingston, Jamaica.
National Library of Jamaica Cataloguing in Publication Data
A catalogue record for this book is available from the National Library of Jamaica
ISBN 978-976-41-0231-1 (pbk)
Covers were designed and photograph taken by Paul Andrew Bourne
3
Table of ContentsPage
Preface 8Menu bar – Contents of the Menu bar in SPSS 11
Function - Purposes of the different things on the menu bar12Mathematical symbols (numeric operations), in SPSS 13Listing of Other Symbols
14The whereabouts of some SPSS functions, or commands
16Disclaimer 19Coding Missing Data 20Computing Date of Birth
21List of Figures 26List of Tables 29How do I obtain access to the SPSS PROGRAM? 351. INTRODUCTION ……………………………………………………………........ 43
1.1.0a: steps in the analysis of hypothesis…………………………………… 451.1.1a Operational definitions of a variable………………………………… 471.1.1b Typologies of variable ………………..………………………………. 491.1.1 Levels of measurement………..………………………………………... 501.1.3 Conceptualizing descriptive and inferential statistics ……………….. 59
2. DESCRIPTIVE STATISTICS ANALYZED ….……………………………........ 622.1.1 Interpreting data based on their levels of measurement………..……. 642.1.2 Treating missing (i.e. non-response) cases…………………….………. 84
3. HYPOTHESES: INTRODUCTION …………………………….………………. 873.1.1 Definitions of Hypotheses………………..……..………………………. 883.1.2: Typologies of Hypothesis……………………………………………… 893.1.3: Directional and non-Directional Hypotheses………………………….. 903.1.4 Outliers (i.e. skewness)…………………………….……………………. 913.1.5 Statistical approaches for treating skewness…………….……………… 93
4. Hypothesis 1…[using Cross tabulations and Spearman ranked ordered correlation]……………………………………………………….. 96
A1. Physical and social factors and instructional resources will directly influence the academic performance of students who will write the Advanced Level Accounting Examination;
A2. Physical and social factors and instructional resources positively influence the academic performance of students who write the Advanced level Accounting examination and that the relationship varies according to gender;
4
B1. Pass successes in Mathematics, Principles of Accounts and English Language at the Ordinary/CXC General level will positively influence success on the Advanced level Accounting examination;
B2. Pass successes in Mathematics, Principles of Accounts and English Language at the Ordinary.
5. Hypothesis 2…………[using Crosstabulations]..…………………………….. 152
There is a relationship between religiosity, academic performance, age and marijuana smoking of Post-primary schools students and does this relationship varies based on gender.
6. Hypothesis 3……….…..…[Paired Sample t-test]…….……………………… 164
There is a statistical difference between the pre-Test and the post-Test scores.
7. Hypothesis 4….………[using Pearson Product Moment Correlation]…..…........184
Ho: There is no statistical relationship between expenditure on social programmes (public expenditure on education and health) and levels of development in a country; and H1: There is a statistical association between expenditure on social programmes (i.e. public expenditure on education and health) and levels of development in a country
8. Hypothesis 5….. ………[using Logistic Regression]…………………………........199
The health care seeking behaviour of Jamaicans is a function of educational level, poverty, union status, illnesses, duration of illnesses, gender, per capita consumption, ownership of health insurance policy, and injuries. [ Health Care Seeking Behaviour = f( educational levels, poverty, union status, illnesses, duration of illnesses, gender, per capita consumption, ownership of health insurance policy, injuries)]
9. Hypothesis 6….. ……[using Linear Regression] ….………………………….. 207
There is a negative correlation between access to tertiary level education and poverty controlled for sex, age, area of residence, household size, and educational level
of parents
10. Hypothesis 7….. ……[using Pearson Product Moment Correlation Coefficient and Crosstabulations]……………………….......................
223
There is an association between the introduction of the Inventory Readiness Test and the Performance of Students in Grade 1
5
11. Hypothesis 8….…………[using Spearman rho]………………………………....232
The people who perceived themselves to be in the upper class and middle class are more so than those in the lower (or working) class do strongly believe that acts of incivility are only caused by persons in garrison communities
12. Hypothesis 9………………………………………………………………........ 235
Various cross tabulations
13. Hypothesis 10………[using Pearson and Crosstabulations]………………........249
There is no statistical difference between the typology of workers in the construction industry and how they view 10-most top productivity outcomes
14. Hypothesis 11….…[using Crosstabulations and Linear Regression]……........265
Determinants of the academic performance of students
15. Hypothesis 12….……[using Spearman ranked ordered correlation]…........278
People who perceived themselves to be within the lower social status (i.e. class) are more likely to be in-civil than those of the upper classes.
16. Data Transformation…………………………………………………........ 281
Recoding 291Dummying variables 309Summing similar variables 331Data reduction 340
Glossary……………..….. ………………………………………………………........ 350
Reference…..………….…………………………………………………………........ 352
6
Appendices…………..….. ………………………………………………………........ 356Appendix 1- Labeling non-responses 356Appendix 2- Statistical errors in data 357Appendix 3- Research Design 359Appendix 4- Example of Analysis Plan 366Appendix 5- Assumptions in regression 367Appendix 6- Steps in running a bivariate cross tabulation 368Appendix 7- Steps in running a trivariate cross tabulation 380Appendix 8- What is placed in a cross tabulations table, using the above SPSS output 394Appendix 9- How to run a Regression in SPSS 395Appendix 10- Running Regression in SPSS 396Appendix 11a- Interpreting strength of associations 407Appendix 11b - Interpreting strength of association 408Appendix 12- Selecting cases 409Appendix 13- ‘UNDO’ selecting cases 417Appendix 14- Weighting cases 420
Appendix 15- ‘Undo’ weighting cases 429Appendix 15- Statistical symbolisms 440Appendix 16 – Converting from ‘string’ to ‘numeric’ data –
Apparatus One – Converting from string data to numeric data 443
Apparatus Two – Converting from alphabetic and numeric data to all ‘numeric data 447
Appendix 17- Steps in running Spearman rho 454
Appendix 18- Steps in running Pearson’s Product Moment Correlation 459
Appendix 19-Sample sizes and their appropriate sampling error 464
Appendix 20 – Calculating sample size from sampling error(s) 465
Appendix 21 – Sample sizes and their sampling errors 467
Appendix 22 - Sample sizes and their sampling errors 468
Appendix 23 – If conditions 469
Appendix 24 – The meaning of ρ value 477
Appendix 25 – Explaining Kurtosis and Skewness 478
Appendix 26 – Sampled Research Papers 479-560
7
PREFACE
One of the complexities for many undergraduate students and for first time researchers is ‘How to blend their socialization with the systematic rigours of scientific inquiry?’ For some, the socialization process would have embedded in them hunches, faith, family authority and even ‘hearsay’ as acceptable modes of establishing the existence of certain phenomena. These are not principles or approaches rooted in academic theorizing or critical thinking. Despite insurmountable scientific evidence that have been gathered by empiricism, the falsification of some perspectives that students hold are difficulty to change as they still want to hold ‘true’ to the previous ways of gaining knowledge. Even though time may be clearly showing those issues are obsolete or even ‘mythological’, students will always adhere to information that they had garnered in their early socialization. The difficulty in objectivism is not the ‘truths’ that it claims to provide and/or how we must relate to these realities, it is ‘how do young researchers abandon their preferred socialization to research findings? Furthermore, the difficulty of humans and even more so upcoming scholars is how to validate their socialization with research findings in the presence of empiricism.
Within the aforementioned background, social researchers must understand that ethic must govern the reporting of their findings, irrespective of the results and their value systems. Ethical principles, in the social or natural research, are not ‘good’ because of their inherent construction, but that they are protectors of the subjects (participants) from the researcher(s) who may think the study’s contribution is paramount to any harm that the interviewees may suffer from conducting the study. Then, there is the issue of confidentiality, which sometimes might be conflicting to the personal situations faced by the researcher. I will be simplistic to suggest that who takes precedence is based on the code of conduct that guides that profession. Hence, undergraduate students should be brought into the general awareness that findings must be reported without any form of alteration. This then give rise to ‘how do we systematically investigate social phenomena?’
The aged old discourse of the correctness of quantitative versus qualitative research will not be explored in this work as such a debate is obsolete and by rehashing this here is a pointless dialogue. Nevertheless, this textbook will forward illustrations of how to analyze quantitative data without including any qualitative interpretation techniques. I believe that the problems faced by students as how to interpret statistical data (ie quantitative data), must be addressed as the complexities are many and can be overcome in a short time with assistance.
My rationale for using ‘hypotheses’ as the premise upon which to build an analysis is embedded in the logicity of how to explore social or natural happenings. I know that hypothesis testing is not the only approach to examining current germane realities, but that it is one way which uses more ‘pure’ science techniques than other approaches.
Hypothesis testing is simply not about null hypothesis, Ho (no statistical relationships), or alternative hypothesis, Ha, it is a systematic approach to the investigation of observable phenomenon. In attempting to make undergraduate students recognize the rich annals of
8
hypothesis testing and how they are paramount to the discovery of social fact, I will recommend that we begin by reading Thomas S. Kuhn (the Scientific Revolution), Emile Durkheim (study on suicide), W.E.B. DuBois (study on the Philadelphian Negro) and the works of Garth Lipps that clearly depict the knowledge base garnered from their usage.
In writing this book, I tried not to assume that readers have grasped the intricacies of quantitative data analysis as such I have provided the apparatus and the solutions that are needed in analyzing data from stated hypotheses. The purpose for this approach is for junior researchers to thoroughly understand the materials while recognizing the importance of hypothesis testing in scientific inquiry.
Paul Andrew Bourne, Dip Ed, BSc, MSc, PhD Health Research Scientist
Department of Community Health and PsychiatryFaculty of Medical Sciences
The University of the West IndiesMona-Jamaica.
9
ACKNOWLEDGEMENT
This textbook would not have materialized without the assistance of a number of people (scholars, associates, and students) who took the time from their busy schedule to guide, proofread and make invaluable suggestions to the initial manuscript. Some of the individuals who have offered themselves include Drs. Ikhalfani Solan, Samuel McDaniel and Lawrence Nicholson who proofread the manuscript and made suggestions as to its appropriateness, simplicities and reach to those it intend to serve. Furthermore, Mr. Maxwell S. Williams is very responsible for fermenting the idea in my mind for a book of this nature. Special thanks must be extended to Mr. Douglas Clarke, an associate, who directed my thoughts in time of frustration and bewilderment, and on occasions gave me insight on the material and how it could be made better for the students.
In addition, I would like to extend my heartiest appreciation to Professor Anthony Harriott and Dr. Lawrence Powell both of the department of Government, UWI, Mona-Jamaica, who are my mentors and have provided me with the guidance, scope for the material and who also offered their expert advice on the initial manuscript.
Also, I would like to take this opportunity to acknowledge all the students of Introduction to Political Science (GT24M) of the class 2006/07 who used the introductory manuscript and made their suggestions for its improvement, in particular Ms. Nina Mighty.
10
Menú Bar
Content:
A social researcher should not only be cognizant of statistical techniques and modalities of performing his/her discipline, but he/she needs to have a comprehensive grasp of the various functions within the ‘menu’ of the SPSS program. Where and what are constituted within the ‘menu bar’; and what are the contents’ functions?
Box 1: Menu Function
‘Menu bar’ contains the following:
- File- Edit- View- Data- Transform- Analyze- Graph- Utilities- Add-ons- Window- Help
The functions of the various contents of the ‘menu bar’ are explored overleaf
11
Menu Bar
Functions : Purposes of the different things on the menu bar
File – This icon deals with the different functions associated with files such as (i) opening .., (ii) reading …, (iii) saving …, (iv) existing.
Edit – This icon stores functions such as – (i) copying, (ii) pasting, (iii) finding, and (iv) replacing.
View – Within this lie functions that are screen related.
Data – This icon operates several functions such as – (i) defining, (ii) configuring, (iii) entering data, (iv) sorting, (v) merging files, (vi) selecting and weighting cases, and (vii) aggregating files.
Transform – Transformation is concerned with previously entered data including (i) recoding, (ii) computing, (iii) reordering, and (vi) addressing missing cases.
Analyze – This houses all forms of data analysis apparatus, with a simply click of the Analyze command.
Graph – Creation of graphs or charts can begin with a click on Graphs command
Utilities – This deals with sophisticated ways of making complex data operations easier, as well as just simply viewing the description of the entered data
12
MATHEMATICAL SYMBOLS (NUMERIC OPERATIONS), in SPSS
NUMERIC OPERATIONS FUNCTIONS
+ Add- Subtract* Multiply/ Divide
** Raise to a power( ) Order of operations< Less than> Greater than
<= Less than or equal to>= Greater than or equal to= Equal
~ = Not equal to& and: both relations must be trueI Or: either relation may be true~ Negation: true between false, false
become trueBox 2: Mathematical symbols and their Meanings
13
LISTING OF OTHER SYMBOLS
SYMBOLS MEANINGS
YRMODA (i.e. yr. month, day)a
Date of birth (e.g. 1968, 12, 05)Y intercept
b Coefficient of slope (or regression)f frequencyn Sample sizeN Population R Coefficient of correlation,
Spearman’sr Coefficient of correlation , PearsonSy Standard error of estimate
W ot Wt Weightµ Mu or population meanβ Beta coefficient
3 or χ Measure of skewness∑ summationσ Standard deviationχ2 Chi-Square or chi square, this is the
value use to test for goodness of fitCC Coefficient of Contingency fa Frequency of class interval above
modal groupfb Frequency of class interval below
modal groupX A single value or variable_R
Adjusted r, which is the coefficient of correlation corrected for the number of cases
_ _ X or Y
RNDSYSMISMISSING
Type I Error
Type II Error
Φr2
Arithmetic mean of X or Y
Round off to the nearest integerThis denotes system-missing valuesAll missing valuesClaiming that events are related (or means are different when they are notThis assumes that events (or means are not different) when they arePhi coefficientThe proportion of variation in the
14
dependent variable explained by the independent variable(s)
LISTING OF OTHER SYMBOLS
SYMBOLS MEANINGS
P(A)
P(A/B)
Probability of event A
Probability of event A given that event B has happened
CV Coefficient of variation
SE
O
X
Y
df
t
R2
Standard error
Observed frequency
Independent (explanatory, predictor) variable in regression
Dependent (outcome, response, criterion) variable in regression
Degree of freedom
Symbol for the t ratio (the critical ratio that follows a t distribution
Squared multiple correlation in multiple regression
15
FURTHER INFORMATION ON TYPE I and TYPE II Error
Finding from your SurveyYou found that the null hypothesis is:
The Real world The null hypothesis is
really……..
True False
True No Problem Type 2 Error
False Type 1 Error No Problem
THE WHEREABOUTS OF SOME SPSS FUNCTIONS
Functions or Commands Whereabouts, in SPSS (the process in arriving at various commands)
Mean,Mode,Median,Standard deviation,Skewness, or kurtosis,RangeMinimum or maximum
Analyze Descriptive statistics
Frequency
Statistics
Analyze
16
Chi-square Descriptive statistics crosstabs
Pearson’s Moment Correlation Analyze
Correlate bivariate
Spearman’s rhoAnalyze
Correlate Bivariate
(ensure that you deselect Pearson’s, and select Spearman’s rho)
Linear RegressionAnalyze
Regression Linear
Logistic RegressionAnalyze
Regression Binary
Discriminant AnalysisAnalyze
Classify Discriminant
Mann-Whitney U TestAnalyze
Nonparametric Test 2 Independent Samples
Independent –Sample t-test Analyze Compare means
Independent Samples T-Test
Wilcoxon matched-pars test orWilcoxon signed-rank test
Analyze Nonparametric Test
2 Independent Samples
t-testAnalyze
Compare means
Paired-samples t-testAnalyze
Compare means Paired-samples T-test
One-sample t-testAnalyze
Compare means One-samples T-test
One-way analysis of varianceAnalyze
Compare means One-way ANOVA
17
Factor AnalysisAnalyze
Data reduction Factor
Descriptive (for a single metric variable)
Analyze Descriptive statistics
Descriptive
Graphs Pie chart Bar charts Histogram
Graphs (select the appropriate type)
Scatter plotsGraphs
Scatter…
Weighting casesData
Weight cases…. Select weight cases by
Selecting casesGraphs
Select cases… If all conditions are satisfied
Select If
Replacing missing valuesTransform
Missing cases values…
Box 3: The whereabouts of some SPSS Functions
18
Disclaimer
I am a trained Demographer, and as such, I have undertaken extensive review of
various aspects to the SPSS program. However, I would like to make this unequivocally clear
that this does not represent SPSS (Statistical Product and Service Solutions, formerly Statistical
Package for the Social Sciences) brand. Thus, this text is not sponsored or approved by SPSS,
and so any errors that are forthcoming are not the responsibility of the brand name.
Continuing, the SPSS is a registered trademark, of SPSS Inc. In the event that you need more
pertinent information on the SPSS program or other related products, this may be forwarded to:
SPSS UK Ltd., First Floor, St. Andrews House, West Street, Working GU211EB, United
Kingdom.
19
Coding Missing Data
The coding of data for survey research is not limited to response, as we need to code missing
data. For example, several codes indicate missing values and the researcher should know them
and the context in which they are applicable in the coding process. No answer in a survey
indicates something apart from the respondent’s refusal to answer or did not remember to
answer. The fundamental issue here is that there is no information for the respondent, as the
information is missing.
Table : Missing Data codes for Survey Research
Question Refused answer Didn’t know answer No answer recorded
Less than 6 categories 7 8 9
More than 7 and less
than 3 digits
97 98 99
More than 3 digits 997 998 999
Note
Less than 6 categories – when a question is asked of a respondent, the option (or response) may
be many. In this case, if the option to the question is 6 items or less, refusal can be 7, didn’t
know 8 or no answer 9.
Some researchers do not make a distinction between the missing categories, and 999 are used
in all cases of missing values (or 99).
20
Computing Date of Birth – If you are only given year of birth Step 1
Step 1:
First, select transform, and then compute
21
Step 2
On selecting ‘compute variable’ it will provide this dialogue box
22
Step 3
In the ‘target variable’, write the word which the researcher wants to use to represents the idea
23
Step 4
If the SPSS program is more than 12.0 (ie 13 – 17), the next process is to select all in ‘function group’ dialogue box
In order to convert year of birth to actual ‘age’, select ‘Xdate.Year’ and sometimes this is bYear.
24
Step 5
Having selected XYear, use this arrow to take it into the ‘Numeric Expression’ dialogue box
Replace the ‘?’ mark with variable in the dataset
25
LISTING OF FIGURES AND TABLES
Listing of Figures
Figure 1.1.1: Flow Chart: How to Analyze Quantitative Data?
Figure 1.1.2: Properties of a Variable.
Figure 1.1.3: Illustration of Dichotomous Variables
Figure 1.1.4: Ranking of the Levels of Measurement
Figure 1.1.5: Levels of Measurement
Figure 2.1.0: Steps in Analyzing Non-Metric Data
Figure 2.1.1: Respondents’ Gender
Figure 2.1.2: Respondents’ Gender
Figure 2.1.3: Social Class of Respondents
Figure 2.1.4: Social Class of Respondents
Figure 2.1.5: Steps in Analyzing Metric Data
Figure 2.1.6: ‘Running’ SPSS for a Metric Variable
Figure 2.1.7: ‘Running’ SPSS for a Metric Variable
Figure 2.1.8: ‘Running’ SPSS for a Metric Variable
Figure 2.1.9: ‘Running’ SPSS for a Metric Variable
Figure 2.1.10: ‘Running’ SPSS for a Metric Variable
Figure 2.1.11: ‘Running’ SPSS for a Metric Variable
Figure 2.1.12: ‘Running’ SPSS for a Metric Variable
Figure 2.1.13: ‘Running’ SPSS for a Metric Variable
Figure 2.1.14: ‘Running’ SPSS for a Metric Variable
Figure 2.1.15: ‘Running’ SPSS for a Metric Variable
26
Figure 2.1.16: ‘Running’ SPSS for a Metric Variable
Figure 4.1.1: Age - Descriptive Statistics
Figure 4.1.2: Gender of Respondents
Figure 4.1.3: Respondent’s parent educational level
Figure 4.1.4: Parental/Guardian Composition for Respondents
Figure 4.1.5: Home Ownership of Respondent’s Parent/Guardian
Figure 4.1.6: Respondents’ Affected by Mental and/or Physical Illnesses
Figure 4.1.7: Suffering from mental illnesses
Figure 4.1.8: Affected by at least one Physical Illnesses
Figure 4.1.9: Dietary Consumption for Respondents
Figure 6.1.2: Typology of Previous School
Figure 6.1.3: Skewness of Examination i (i.e. Test i)
Figure 6.1.4: Skewness of Examination ii (i.e. Test ii)
Figure 6.1.5: Perception of Ability
Figure 6.1.6: Self-perception
Figure 6.1.7: Perception of task
Figure 6.1.8: Perception of utility
Figure 6.1.9: Class environment influence on performance
Figure 6.1.10: Perception of Ability
Figure 6.1.11: Self-perception
Figure 6.1.12: Self-perception
Figure 6.1.13: Perception of task
Figure 6.1.14: Perception of Utility
27
Figure 6.1.15: Class Environment influence on Performance
Figure 7.1.1: Frequency distribution of total expenditure on health as % of GDP
Figure 7.1.2: Frequency distribution of total expenditure on education as % of GNP
Figure 7.1.3: Frequency distribution of the Human Development Index
Figure 7.1.4: Running SPSS for social expenditure on social programme
Figure 7.1.5: Running bivariate correlation for social expenditure on social programme
Figure 7.1.6: Running bivariate correlation for social expenditure on social programme
Figure13.1.1: Categories that describe Respondents’ Position
Figure13.1.2: Company’s Annual Work Volume
Figure13.1.3: Company’s Labour Force – ‘on an averAge per year’
Figure13.1.4: Respondents’ main Area of Construction Work
Figure13.1.5: Percentage of work ‘self-performed’ in contrast to ‘sub-contracted’
Figure13.1.6: Percentage of work ‘self-performed’ in contrast to ‘sub-contracted’
Figure 13.1.7: Years of Experience in Construction Industry
Figure13.1.8: Geographical Area of Employment
Figure13.1.9: Duration of service with current employer
Figure13.1.10: Productivity changes over the past five years
Figure 14.1.1: Characteristic of Sampled Population
Figure 14.1.2: Employment Status of Respondents
28
Listing of Tables
Table 1.1.1: Synonyms for the different Levels of measurement
Table 1.1.2: Appropriateness of Graphs, from different Levels of measurement
Table 1.1.3: Levels of measurement1 with examples and other characteristics
Table1.1.4: Levels of measurement, and measure of central tendencies and measure of variability
Table1.1.5: combinations of Levels of measurement, and types of statistical Test which are application
Table 1.1.6a: Statistical Tests and their Levels of Measurement
Table 1.1.6b:
Table 2.1.1a: Gender of Respondents
Table 2.1.1b: General happiness
Table 2.1.2: Social Status
Table 2.1.3: Descriptive Statistics on the Age of the Respondents
Table 2.1.4:“From the following list, please choose what the most important characteristic of democracy …are for you”
Table 4.1.1: Respondents’ Age
Table 4.1.2 (a) Univariate Analysis of the explanatory Variables
Table 4.1.2(b): Univariate Analysis of explanatory
Table 4.1.2 (c): Univariate Analysis of explanatory
Table 4.1.3: Bivariate Relationships between academic performance and subjective Social Class (n=99)
1
29
Table 4.1.4: Bivariate Relationships between comparative academic performance and subjective Social Class (n=108)
Table 4.1.5: Bivariate Relationships between academic performance and physical exercise (n= 111)
Table 4.1.6 (i): Bivariate Relationships between academic performance and instructional
materials (n=113)
Table 4.1.6 (ii) Relationship between academic performance and materials among students who will be writing the A’ Level Accounting Examination, 2004
Table 4.1.7: Bivariate Relationships between academic performance and Class attendance (n= 106)
Table 4.1.8: Bivariate Relationship between academic performance and attendance
Table 4.1.9: Bivariate Relationships between academic performance and breakfast consumption, (n=114)
Table 4.1.10: Relationship between academic performances and breakfasts consumption among A’ Level Accounting students, controlling for Gender
Table 4.1.11: Bivariate Relationships between academic performance and migraine (n=116)
Table 4.1.12: Bivariate Relationships between academic performance and mental illnesses, (n=116)
Table 4.1.13: Bivariate Relationships between academic performance and physical illnesses, (n=116)
Table 4.1.14: Bivariate Relationships between academic performance and illnesses (n=116)
Table 4.1.15. Bivariate Relationships between current academic performance and past performance in CXC/GCE English language Examination, (n= 112)
Table 4.1.16: Bivariate Relationships between academic performance and past performance in CXC/GCE English language Examination, controlling for Gender
Table 4.1.17: Bivariate Relationships between academic performance and past performance in CXC/GCE Mathematics Examination n=
Table 4.1.18 (i): Bivariate Relationships between academic performance and past performance in CXC/GCE principles of accounts Examination (n= 114)
30
Table 4.1.19 (ii): Bivariate Relationships between academic performance and past performance in CXC/GCEPOA Examination, controlling for Gender
Table 4.1.20: Bivariate Relationships between academic performance and Self-Concept (n= 112)
Table 4.1.21: Bivariate Relationships between academic performance and Dietary Requirements (n=116)
Table 4.1.22: Summary of Tables
Table 5.1.1: Frequency and percent Distributions of explanatory model Variables
Table 5.1.2: Relationship between Religiosity and Marijuana Smoking (n=7,869)
Table 5.1.3: Relationship between Religiosity and Marijuana Smoking controlled for Gender
Table 5.1.4: Relationship between Age and marijuana smoking (n=7,948)
Table 5.1.5: Relationship between marijuana smoking and Age of Respondents, controlled for sex
Table 5.1.6: Relationship between academic performances and marijuana smoking, (n=7,808)
Table 5.1.7: Relationship between academic performances and marijuana smoking, controlled for Gender
Table 5.1.8: Summary of Tables
Table 6.1.1: Age Profile of respondent
Table 6.1.2: Examination Scores
Table 6.1.3(a): Class Distribution by Gender
Table 6.1.3(b): Class Distribution by Age Cohorts
Table 6.1.3(c): Pre-Test Score by Typology of Group
Table 6.1.3(c): Pre-Test Score by Typology of Group
Table 6.1.4: Comparison of Examination I and Examination II
Table 6.1.5: Comparison a Cross the Group by Tests
31
Table 6.1.6: Analysis of Factors influence on Test ii Scores
Table 6.1.7: Cross-Tabulation of Test ii Scores and Factors
Table 6.1.8: Bivariate Relationship between student’s Factors and Test ii Scores
Table 7.1.1: Descriptive Statistics - total expenditure on public health (as Percentage of GNP HRD, 1994)
Table 7.1.2: Descriptive Statistics of expenditure on public education (as Percentage of GNP, Hrd, 1994)
Table 7.1.3: Descriptive Statistics of Human Development (proxy for development)
Table 7.1.4: Bivariate Relationships between dependent and independent Variables
Table 7.1.5: Summary of Hypotheses Analysis
Table8.1.1: Age Profile of Respondents (n = 16,619)
Table 8.1.2: Logged Age Profile of Respondents (n = 16,619)
Table 8.1.3: Household Size (all individuals) of Respondents
Table 8.1.4: Union Status of the sampled Population (n=16,619)
Table 8.1.5: Other Univariate Variables of the Explanatory Model
Table 8.1.6: Variables in the Logistic Equation
Table 8.1.7: Classification Table
Table 8.1.1: Univariate Analyses
Table 8.1.2: Frequency Distribution of Educational Level by Quintile
Table 8.1.3: Frequency Distribution of Jamaica’s Population by Quintile and Gender
Table 8.1.4: Frequency Distribution of Educational Level by Quintile
Table 8.1.5: Frequency Distribution of Pop. Quintile by Household Size
Table 8.1.6: Bivariate Analysis of access to Tertiary Edu. and Poverty Status
Table 8.1.7: Bivariate Analysis of access to Tertiary Edu. and Geographic Locality of Residents
Table 8.1.8: Bivariate Analysis of geographic locality of residents and poverty Status
32
Table 8.1.9: Bivariate Relationship between access to tertiary level education by Gender
Table 8.1.10: Bivariate Relationship between Access to Tertiary Level Education by Gender controlled for Poverty Status
Table 8.1.11: Regression Model Summary
Table 10.1.1: Univariate Analysis of Parental Information
Table 10.1.2: Descriptive on Parental Involvement
Table 10.1.3: Univariate Analysis of Teacher’s Information
Table 10.1.4: Univariate Analysis of ECERS-R Profile
Table 10.1.5: Bivariate Analysis of Self-reported Learning Environment and Mastery on Inventory Test
Table 10.1.6: Relationship between Educational Involvement, Psychosocial and Environment involvement and Inventory Test
Table 10.1.6: Relationship between Educational Involvement, Psychosocial and Environment Involvement and Inventory Test
Table 10.1.8: School Type by Inventory Readiness Score
Table 11.1.1: Incivility and Subjective Social Status
Table 12.1.2: Have you or someone in your family known of an act of Corruption in the last 12 months?
Table 12.1.3: Gender of Respondent
Table 12.1.4: In what Parish do you live?
Table 12.1.5: Suppose that you, or someone close to you, have been a victim of a crime. What would you do...?
Table 12.1.6: What is your highest level of Education?
Table 12.1.7: In terms of Work, which of these best describes your Present situation?
Table 12.1.8: Which best represents your Present position in Jamaica Society?
Table 12.1.9: Age on your last Birthday?
Table 12.1.10: Age categorization of Respondents
33
Table 12.1.11: Suppose that you, or someone close to you, have been a victim of a crime. what would you do... by Gender of respondent Cross Tabulation
Table 12.1.12: If involved in a dispute with neighbour and repeated discussions have not made a difference, would you...? by Gender of respondent Cross Tabulation
Table 12.1.13: Do you believe that corruption is a serious problem in Jamaica? by Gender of respondent Cross Tabulation
Table 12.1.14: have you or someone in your family known of an act of corruption in the last 12 months? by Gender of respondent Cross Tabulation
Table 14.1.1: Marital Status of Respondents
Table 14.1.2: Marital Status of Respondents by Gender
Table 14.1.3: Marital Status by Gender by Age cohort
Table 14.1.4: Marital Status by Gender by Age Cohort
Table 14.1.5 Educational Level by Gender by Age Cohorts
Table 14.1.6: Income Distribution of Respondents
Table 14.1.7: Parental Attitude Toward School
Table 14.1.8: Parent Involving Self
Table 14.1.9: School Involving Parent
Table 14.1.8: Regression Model Summary
Table 15.1.1: Correlations
Table 15.1.2: Cross Tabulation between incivility and social status
34
How do I obtain access to the SPSS PROGRAM?
Step One:
In order to access the SPSS program, the student should select ‘START’ to the
bottom left hand corner of the computer monitor. This is followed by selecting
‘All programs’ (see below).
Select ‘START’ and then ‘All Program
35
Step Two:
The next step to the select ‘SPSS for widows’. Having chosen ‘SPSS for
widows’ to the right of that appears a dialogue box with the following options –
SPSS for widows; SPSS 12.0 (or 13.0…or, 15.0); SPSS Map Geo-dictionary
Manager Ink; and last with SPSS Manager.
Select ‘SPSS for widows’
36
Step Three:
Having done step two, the student will select SPSS 12.0 (or 13.0, or 14.0 or 15.0) for
Widows as this is the program with which he/she will be working.
Select SPSS 12.0 (or 13.0, or 14.0 or 15.0) for Widows
37
Step Four:
On selecting ‘SPSS for widows’ in step 3, the below dialogue box appears. The
next step is the select ‘OK’, which result in what appears in step five.
Select‘OK’
38
Step Five:
What should I now do? The student should then select the ‘inner red box’ with the ‘X’.
Select the ‘inner red box’ with the X’.
39
Step Six:
This is what the SPSS spreadsheet looks like (see Figure below).
40
Step Seven:
What is the difference here? Look to the bottom left-hand cover the spreadsheet
and you will see two terms – (1) ‘Data View’ and (2) ‘Variable View’. Data
View accommodates the entering of the data having established the template in
the ‘Variable View’. Thus, the variable view allows for the entering of data (i.e.
responses from the questionnaires) in the ‘Data View’. Ergo, the student must
ensure that he/she has established the template, before any typing can be done in
the ‘Data View.
Data View
Observe what the ‘Data View’ widow looks like
41
Variable View
Observe what the ‘Variable View’ widow looks like
42
CHAPTER 1
1.1.0a: INTRODUCTION
This book is in response to an associate’s request for the provision of some material that would
adequately provide simple illustrations of ‘How to analyze quantitative data in the Social
Sciences from actual hypotheses’. He contended that all the current available textbooks,
despite providing some degree of analysis on quantitative data, failed to provide actual
illustrations of cases, in which hypotheses are given and a comprehensive assessment made to
answer issues surrounding appropriate univariate, bivariate and/or multivariate processes of
analysis. Hence, I began a quest to pursued textbooks that presently exist in ‘Research Methods
in Social Sciences’, ‘Research Methods in Political Sciences’, “Introductory Statistics’,
‘Statistical Methods’, ‘Multivariate Statistics’, and ‘Course materials on Research Methods’
which revealed that a vortex existed in this regard.
Hence, I have consulted a plethora of academic sources in order to formulate this text.
In wanting to comprehensively fulfill my friend’s request, I have used a number of dataset that
I have analyzed over the past 6 years, along with the provision of key terminologies which are
applicable to understanding the various hypotheses.
I am cognizant that a need exist to provide some information in ‘Simple Quantitative
Data Analysis’ but this text is in keeping with the demand to make available materials for
aiding the interpretation of ‘quantitative data’, and is not intended to unveil any new materials
in the discipline. The rationale behind this textbook is embedded in simple reality that many
undergraduate students are faced with the complex task of ‘how to choose the most appropriate
43
statistical test’ and this becomes problematic for them as the issue of wanting to complete an
assignment, and knowing that it is properly done, will plague the pupil. The answer to this
question lies in the fundamental issues of - (1) the nature of the variables (continuous or
discrete), and (2) what is the purpose of the analysis – is to mere description, or to provide
statistical inference and/or (3) if any of the independent variables are covariates2. Nevertheless,
the materials provided here are a range of research projects, which will give new information
on particular topics from the hypothesis to the univariate analysis and the bivariate or
multivariate analyses.
2 “If the effects of some independent variables are assessed after the effects of other independent variables are statistically removed…” (Tabachnick and Fidell 2001, 17)
44
1.1.0b: STEPS IN ANALYZING A HYPOTHESIS
One of the challenges faced by a social researcher is how to succinctly conceptualize (i.e.
define) his/her variables, which will also be operationalized (measured) for the purpose of the
study. Having written a hypothesis, the researcher should identify the number of variables
which are present, from which we are to identify the dependent from the independent variables.
Following this he/she should recognize the level of measurement to which each variable
belongs, then the which statistical test is appropriate based on the level of measurement
combination of the variables. The figure below is a flow chart depicting the steps in analyzing
data when given a hypothesis.
The production of this text is in response to the provision of a simple book which
would address the concerns of undergraduate students who must analyze a hypothesis. Among
the issues raise in this book are (1) the systematic steps involved in the completion of
analyzing a hypothesis, (2) definitions of a hypothesis, (3) typologies of hypothesis, (4)
conceptualization of a variable, (4) types of variables, (5) levels of measurement, (6)
illustration of how to perform SPSS operations on the description of different levels of
measurement and inferential statistics, (7) Type I and II errors, (8) arguments on the treatment
of missing variables as well as outliers, (9) how to transform selected quantitative data, (10)
and other pertinent matters.
The primary reason behind the use of many of the illustrations, conceptualizations and
peripheral issues rest squarely on the fact the reader should grasp a thorough understanding of
how the entire process is done, and the rationale for the used method.
45
FIGURE 1.1.1: FLOW CHART: HOW TO ANALYZE QUANTITATIVE DATA?
This entire text is ‘how to analyze quantitative data from hypothesis’, but based on Figure
1.1.1, it may appear that a research process begins from a hypothesis, but this is not the case.
Despite that, I am emphasizing interpreting hypothesis, which is the base for this monograph
starting from an actual hypothesis. Thus, before I provide you with operational definitions of
STEP ONEWrite your Hypothesis
STEP ONEWrite your Hypothesis STEP TWO
Identify the variables from the hypothesis
STEP TWOIdentify the
variables from the hypothesis
STEP THREEDefine and
operationalize each variable selected from the hypothesis
STEP THREEDefine and
operationalize each variable selected from the hypothesis
STEP FOUR
Decide on the level of
measurement for each variable
STEP FOUR
Decide on the level of
measurement for each variable
STEP FIVE
Decide which variable is DV,
and IV
STEP FIVE
Decide which variable is DV,
and IVSTEP SIXCheck for skewness,
and/or outliers in metric variables
STEP SIXCheck for skewness,
and/or outliers in metric variables
STEP SEVENDo descriptive statistics for
chosen variables selected
STEP SEVENDo descriptive statistics for
chosen variables selected
STEP EIGHTIf statistical association, causality or
predictability is need, continue, if
not stop!
STEP EIGHTIf statistical association, causality or
predictability is need, continue, if
not stop!
STEP NINEIf statistical Inference is
needed, look at the
combination DV and IV(s)
STEP NINEIf statistical Inference is
needed, look at the
combination DV and IV(s)
STEP TENChoose the appropriate
statistical test based on the
combination of DV and IVS, and
STEP TENChoose the appropriate
statistical test based on the
combination of DV and IVS, and
STEP TENHaving used
the test, analyze the data carefully, based on the statistical test
STEP TENHaving used
the test, analyze the data carefully, based on the statistical test
ANALYZINGQUANTITATIVE
DATA
ANALYZINGQUANTITATIVE
DATA
46
variables, I will provide some contextualization of ‘what is a variable?’ then the steps will be
worked out.
47
1.1.1a: DEFINITIONS OF A VARIABLE
Undergraduates and first time researchers should be aware that quantitative data analysis are primarily based on (1) empirical literature, (2) typologies of variables within the hypothesis, (3) conceptualization and operationalization of the variables, (4) the level of measurement for each variables. It should be noted that defining a variable is simply not just the collation a group of words together, because we feel a mind to as each variable requires two critical characteristics in order that it is done properly (see Figure 1.1.2).
FIGURE 1.1.2: PROPERTIES OF A VARIABLE.
In order to provide a comprehensive outlook of a variable, I will use the definitions of a
various scholars so as to give a clear understanding of what it is.
“Variables are empirical indicators of the concepts we are researching. Variables, as their name implies, have the ability to take on two or more values...The categories of each variable must have two requirements. They should be both exhaustive and mutually exclusive. By exhaustive, we mean that the categories of each variable must be comprehensive enough that it is possible to categorize every observation” (Babbie, Halley, and Zaino 2003, 11).
“.. Exclusive refers to the fact that every observation should fit into only one category “(Babbie, Halley and Zaino 2003, 12)
“A variable is therefore something which can change and can be measured.” (Boxill, Chambers and Wint 1997, 22)
PROPERITIES OF A VARIABLEPROPERITIES OF A VARIABLE
MUTUAL EXCLUSIVITIYMUTUAL EXCLUSIVITIYEXHAUSTIVNESSEXHAUSTIVNESS
48
“The definition of a variable, then, is any attribute or characteristic of people, places, or events that takes on different values.” (Furlong, Lovelace, Lovelace 2000, 42)
“A variable is a characteristic or property of an individual population unit” (McClave, Benson and Sincich 2001, 5)
“Variable. A concept or its empirical measure that can take on multiple values” (Neuman 2003, 547).
“Variables are, therefore, the quantification of events, people, and places in order to measure observations which are categorical (i.e. nominal and ordinal data) and non-categorical (i.e. metric) in an attempt to be informed about the observation in reality. Each variable must fill two basic conditions – (i) Exhaustiveness – the variable must be so defined that all tenets are captured as its is comprehensive enough include all the observations, and (ii) mutually exclusivity – the variable should be so defined that it applies to one event and one event only – (i.e. Every observation should fit into only one category) (Bourne 2007).
One of the difficulties of social research is not the identification of a variable or
variables in the study but it’s the conceptualization and oftentimes the operationalization of
chosen construct. Thus, whereas the conceptualization (i.e. the definition) of the variable may
(or may not) be complex, it is the ‘how do you measure such a concept (i.e. variable) which
oftentimes possesses the problem for researchers. Why this must be done properly bearing in
mind the attributes of a variable, it is this operational definition, which you will be testing in
the study (see Typologies of Variables, below). Thus, the testing of hypothesis is embedded
within variables and empiricism from which is used to guide present studies. Hypothesis
testing is a technique that is frequently employed by demographers, statisticians, economists,
psychologists, to name new practitioners, who are concerned about the testing of theories, and
the verification of reality truths, and the modifications of social realities within particular time,
space and settings. With this being said, researchers must ensure that a variable is properly
defined in an effort to ensure that the stated phenomenon is so defined and measured.
49
1.1.1b TYPOLOGIES of VARIABLE (examples, using Figure 1.1.2, above)
Health care seeking behaviour: is defined as people visiting a health practitioner or health
consultant such as doctor, nurse, pharmacist or healer for care and/ or advice.
Levels of education: This is denominated into the number of years of formal schooling that
one has completed.
Union status – It is a social arrangement between or among individuals. This arrangement
may include ‘conjugal’ or a social state for an individual.
Gender: A sociological state of being male or female.
Per capita income: This is used a proxy for income of the individual by analyzing the
consumption pattern.
Ownership of Health insurance: Individuals who possess of an insurance polic/y (ies).
Injuries: A state of being physically hurt. The examples here are incidences of disability,
impairments, chronic or acute cuts and bruises.
Illness: A state of unwellness.
Age: The number of years lived up to the last birthday.
Household size - The numbers of individuals, who share at least one common meal, use
common sanitary convenience and live within the same dwelling.
Now that the premise has been formed, in regard to the definition of a variable, the next
step in the process is the category in which all the variables belong. Thus, the researcher needs
50
to know the level of measurement for each variable - nominal; ordinal; interval, or ration (see
1.1.2a).
1.1.2a: LEVELS OF MEASUREMENT3: Examples and definitions
Nominal - The naming of events, peoples, institutions, and places, which are coded numerical by the researcher because the variable has no normal numerical attributes. This variable may be either (i) dichotomous, or (ii) non-dichotomous.
Dichotomous variable – The categorization of a variable, which has only two sub-groupings - for example, gender – male and female; capital punishment – permissive and restrictive; religious involvement – involved and not involved.
Non-dichotomous variable – The naming of events which span more than two sub-categories (example Counties in Jamaica – Cornwall, Middlesex and Surrey; Party Identification – Democrat, Independent, Republican; Ethnicity – Caucasian, Blacks, Chinese, Indians; Departments in the Faculty of Social Sciences – Management Studies, Economics, Sociology, Psychology and Social Work, Government; Political Parties in Jamaica – Peoples’ National Party (PNP), Jamaica Labour Party (JLP), and the National Democratic Movement (NDM); Universities in Jamaica – University of the West Indies; University of Technology, Jamaica; Northern Caribbean University; University College of the Caribbean; et cetera)
Ordinal - Rank-categorical variables: Variables which name categories, which by their very nature indicates a position, or arrange the attributes in some rank ordering (The examples here are as follows i) Level of Educational Institutions – Primary/Preparatory, All-Age, Secondary/High, Tertiary; ii) Attitude toward gun control – strongly oppose, oppose, favour, strongly favour; iii) Social status – upper--upper, upper-middle, middle-middle, lower-middle, lower class; iv) Academic achievement – A, B, C, D, F.
Intervalor ratio These variables share all the characteristics of a nominal and an ordinal variable
along with an equal distance between each category and a ‘true’ zero value – (for example – age; weight; height; temperature; fertility; votes in an election, mortality; population; population growth; migration rates, .
3 Stanley S. Stevens is created for the development of the typologies of scales – level of measurement – (i) nominal, (ii) ordinal, (iii) interval and (iv) ratio. (see Steven 1946, 1948, 1968; Downie and Heath 1970)
51
Now that the definitions and illustrations have been provided for the levels of measurement,
the student should understand the position of these measures (see 1.1.2b).
52
Figure 1.1.3: Illustration of dichotomous variables
Dichotomy (or
Dichotomous variable
Dichotomy (or
Dichotomous variable
Typologies of Book
Typologies of Book
GenderGender ScienceScience
Fictional Fictional Non-
FictionalNon-
FictionalMaleMale FemaleFemale PurePure AppliedApplied
InductionInduction DeductionDeduction
Parametric statistics
Parametric statistics
Non-parametric statistics
Non-parametric statistics
AliveAlive DeadDead
BurialBurial Non-burialNon-burial
DecomposedDecomposedNon-
decomposedNon-
decomposed
use primarydata
use primarydata
use secondarydata
use secondarydata
Religious service
Religious service
Non-religious service
Non-religious service
53
1.1.2b: RANKING LEVELS OF MEASUREMENT
Figure 1.1.4: Ranking of the levels of measurement
The very nature of levels of measurement allows for (or do not allow for) data manipulation. If
the level of measurement is nominal (for example fiction and non-fiction books), then the
researcher does not have a choice in the reconstruction of this variable to a level which is
below it. If the level of measurement, however, is ordinal (for example no formal education,
primary, secondary and tertiary), then one may decide to use a lower level of measure (for
example below secondary and above secondary). The same is possible with an interval
variable. The social scientist may want to use one level down, ordinal, or two levels down,
nominal. This is equally the same of a ratio variable. Thus, the further ones go up the
pyramid, the more scope exists in data transformation.
RATIO
RRR
INTERVAL
ORDINAL
NOMINAL
highest
lowest
54
Table 1.1.1: Synonyms for the different Levels of measurement
Levels of Measurement Other terms
Nominal Categorical; qualitative, discrete4
Ordinal Qualitative, discrete; rank-ordered; categorical
Interval/Ratio Numerical, continuous5, quantitative; scale; metric, cardinal
Table 1.1.2: Appropriateness of Graphs for different levels of measurement
Levels of Measurement Graphs
Bar chart Pie chart Histogram Line Graph
Nominal √ √ __ __
√ √ __ __Ordinal
__ __ √ √Interval/Ratio (or metric)
4 Discrete variable – take on a finite and usually small number of values, and there is no smooth transition from one value or category to the next – gender, social class, types of community, undergraduate courses5 Continuous variables are measured on a scale that changes values smoothly rather than in steps
55
Table 1.1.3: Levels of measurement6 with Examples and Other Characteristics
Levels of Measurement
Nominal Ordinal Interval Ratio
Examples Gender Social class Temperature AgeReligion Preference Shoe size HeightPolitical Parties Level of education Life span WeightRace/Ethnicity Gender equity Reaction timePolitical Ideologies levels of fatigue Income; Score on an Exam.
Noise level Fertility; Population of a country Job satisfaction Population growth; crime rates
Mathematical properties Identity Identity Identity Identity ____ Magnitude
Magnitude Magnitude ____ _____ Equal Interval Equal interval ____ _____ _____ True zero
Mathematical Operation(s) None Ranking Addition; Addition; Subtraction Subtraction; Division; Multiplication
Compiled: Paul A. Bourne, 2007; a modification of Furlong, Lovelace and Lovelace 2000, 74
6 “Levels of measurement concern the essential nature of a variable, and it is important to know this because it determines what one can do with a variable (Burham, Gilland, Grant and Layton-Henry 2004, 114)
56
Table1.1.4: Levels of measurement, Measure of Central Tendency and Measure of Variability
Levels of Measurement Measure of central tendencies Measure of variability
Mean Mode Median Mean deviation Standard deviation
Nominal NA √ NA NA NA
Ordinal NA √ √ NA NA
Interval/Ratio7 √ √ √ √ √
NA denotes Not Applicable
7 Ratio variable is the highest level of measurement, with nominal being first (i.e. lowest); ordinal, second; and interval, third.
57
Table1.1.5: Combinations of Levels of measurement, and types of Statistical test which are applicable8
Levels of Measurement Statistical Test
Dependent Independent VariableNominal Nominal Chi-square
Nominal Ordinal Chi-square; Mann-Whitney
Nominal Interval/ratio Binomial distribution; ANOVA; Logistic Regression; Kruskal-Wallis
Discriminant Analysis
Ordinal Nominal Chi-square
Ordinal Ordinal Chi-square; Spearman rho;
Ordinal Interval/ratio Kruskal-Wallis H; ANOVA
Interval/ratio Nominal ANOVA;
Interval/ratio OrdinalInterval/ratio Interval/ratio Pearson r, Multiple Regression
Independent-sample t test
Table 1.1.5 depicts how a dependent variable, which for example is nominal, which when combined with an independent variable,
Nominal, uses a particular statistical test.
8 One of the fundamental issues within analyzing quantitative data is not merely to combine then interpret data, but it is to use each variable appropriately. This is further explained below.
58
STATISTICAL TESTS AND THEIR LEVELS OF MEASUREMENT
Test IndependentVariable
Dependentvariable
Chi-Square (χ2) Nominal, Ordinal Nominal, OrdinalMann-Whitney U test
Dichotomous Nominal, Ordinal
Kruskal-Wallis H test
Non-dichotomous,Ordinal
Ordinal, or skewed9
MetricPearson’s r Normally distributed10
Metric Normally distributed
MetricLinear Regress Normally distributed
Metric, dummyNormally distributed
MetricIndependent SamplesT-test
Dichotomous Normally distributedMetric
AVONA Nominal, Ordinal (non-dichotomous11)
Normally distributedMetric
Logistic regression Metric, dummy Dichotomous (skewedvalues or otherwise
Discriminant analysis
Metric, dummy Dichotomous (normally distributed value)
Notes to Table 1.1.6b
Chi-Square (χ2) Used to test for associations between two variables Mann-Whitney U test Used to determine differences between two groupsKruskal-Wallis H test Used to determine differences between three or more groupsPearson’s r Used to determine strength and direction of a relationship
between two valuesLinear Regression Used to determine strength and direction of a relationship
between two or more valuesIndependent SamplesT-test Used to determine difference between two groupsAVONA Used to determine difference between three or more groupsLogistic regression Used to predict relationship between many valuesDiscriminant analysis Used to predict relationship between many values
9 Skewness indicates that there is a ‘pileup’ of cases to the left or right tail of the distribution10 Normality is observed, whenever, the values of skewness and kurtosis are zero11 Non-dichotomous (i.e. polytomous) which denotes having many (i.e. several) categories
LEVELS OF MEASURMENT AND THEIR MEASURING ASSOCIATION
Figure 1.1.5: Levels of measurement Lambda (ג) – This is a measure of statistical relationship between the uses of two nominal
variables
Phi (Φ) – This is a measure of association between the use of two dichotomous variables (i.e. dichotomous dependent and dichotomous independent) –
[Φ = √[ χ2/N]
Cramer’s V (V) – This is a measure of association between the use of two nominal variables (i.e. in the event that there is dichotomous dependent and
dichotomous independent) – V = √[ χ2/N(k – 1)] is identical to phi.
Gamma (γ) – This is used to measure the statistical association between ordinal by ordinal variable
Contingency coefficient (cc) – Is used for association in which the matrix is more than 2 X 2 (i.e. 2 for dependent and 2 for the independent – for example 2X3;
3X2; 3X3 …) - √ [χ2/ χ2 + N]
Pearson’s r – This is used for non-skewed metric variables - n∑xy - ∑x.∑y √ [n∑x2 – (∑x) 2 - [n∑y2 – (∑y) 2
LEVELS OF MEASUREMENT
LEVELS OF MEASUREMENT
NOMINALNOMINAL ORDINALORDINAL INTERVAL/RATIOINTERVAL/RATIO
LambdaLambda
Cramer’s VCramer’s V
Contingency coefficientsContingency coefficients
GammaGamma
Somer’s DSomer’s D
Kendall ‘s tau-BKendall ‘s tau-B
Pearson’s rPearson’s r
PhiPhi Kendall’s tau-cKendall’s tau-c
60
1.1.3: CONCEPTUALIZING DESCRIPTIVE AND INFERENTIAL STATISTICS
Research is not done in isolation from the reality of the wider society. Thus, the social
researcher needs to understand whether his/her study is descriptive and/or inferential as it
guides the selection of certain statistical tools. Furthermore, an understanding of two
constructs dictate the extent to which the analyst will employ as there is a clear
demarcation between descriptive and inferential statistics. In order to grasp this
distinction, I will provide a number of authors’ perspectives on each terminology.
“Descriptive statistics describe samples of subjects in terms of variables or combination
of variables” (Tabachnick and Fidell 2001, 7)
“Numerical descriptive measures are commonly used to convey a mental image of
pictures, objects, tables and other phenomenon. The two most common numerical
descriptive measures are: measures of central tendencies and measures of variability
(McDaniel 1999, 29; see also Watson, Billingsley, Croft and Huntsberger 1993, 71)
“Techniques such as graphs, charts, frequency distributions, and averages may be used
for description and these have much practical use” (Yamane 2973, 2; see also Blaikie
2003, 29; Crawshaw and Chambers 1994, Chapter 1)
61
“Descriptive statistics – statistics which help in organizing and describing data, including
showing relationships between variables” (Boxill, Chamber and Wind 1997, 149)
“We’ll see that there are two areas of statistics: descriptive statistics, which focuses on
developing graphical and numeral summaries that describes some…phenomenon, and
inferential statistics, which uses these numeral summaries to assist in making…
decisions” (McClave, Benson, Sinchich 2001, 1)
“Descriptive statistics utilizes numerical and graphical methods to look for patterns in a
data set, to summarize the information revealed in a data set, and to present the
information in a convenient form” (McClave, Benson and Sincich 2001, 2)
“Inferential statistics utilizes sample data to make estimates, decisions, predictions, or
other generalizations about a larger set of data” (McClave, Benson and Sincich 2001, 2)
“The phrase statistical inference will appear often in this book. By this we mean, we
want to “infer” or learn something about the real world by analyzing a sample of data.
The ways in which statistical inference are carried out include: estimating…parameters;
predicting…outcomes, and testing…hypothesis …” (Hill, Griffiths and Judge 2001, 9).
Inferential statistics is not only about ‘causal’ relationships; King, Keohane and
Verba argue that it is categorized into two broad areas: (1) descriptive, and (2) causal
inference. Thus, descriptive inference speaks to the description of a population from
62
what is made possible, the sample size. According to Burham, Gilland, Grant and
Layton-Henry (2004) state that:
Causal inferences differ from descriptive ones in one very significant way: they take a ‘leap’ not only in terms of description, but in terms of some specific causal process [i.e. predictability of the variables]” (Burham, Gilland, Grand and Layton-Henry 2004, 148).
In order that this textbook can be helping and simple, I will provide operational
definitions of concepts as well as illustration of particular terminologies along with
appropriateness of statistical techniques based on the typologies of variable and the level
of measurement (see in Tables 1.1.1 – 1.1.6, below).
63
CHAPTER 2
2.1.0: DESCRIPTIVE STATISTICS
The interpretation of quantitative data commences with an overview (i.e. background
information on survey or study – this is normally demographic information) of the
general dataset in an attempt to provide a contextual setting of the research (descriptive
statistics, see above), upon which any association may be established (inferential
statistics, see above). Hence, this chapter provides the reader with the analysis of
univariate data (descriptive statistics), with appropriate illustration of how various levels
of measurement may be interpreted, and/or diagrams chosen based on their suitability.
A variable may be non-metric (i.e. nominal or ordinal) or metric (i.e. scale,
interval/ratio). It is based on this premise that particular descriptive statistics are provide.
In keeping with this background, I will begin this process with non-metric, then metric
data. The first part of this chapter will provide a thorough outline of how nominal and/or
ordinal variables are analyzed. Then, the second aspect will analyze metric variables.
64
Figure 2.1.0: Steps in Analyzing Non-metric data
STEP ONEEnsure that the variable is non-
metric (e.g. Gender, general
happiness)
STEP ONEEnsure that the variable is non-
metric (e.g. Gender, general
happiness)STEP TWO
Select Analyze
STEP TWO
Select Analyze
STEP THREESelect descriptive
statistics
STEP THREESelect descriptive
statistics
STEP FOUR
select frequency
STEP FOUR
select frequency
STEP FIVE
select the non-metric variable
STEP FIVE
select the non-metric variable
STEP SIX
select statistics at the end
STEP SIX
select statistics at the end
STEP SEVENselect mode or
mode and median (based on if the
variable is nominal or ordinal respective
STEP SEVENselect mode or
mode and median (based on if the
variable is nominal or ordinal respective
STEP EIGHT
select Chart
STEP EIGHT
select Chart
STEP NINE
Choose bar or pie graphs
STEP NINE
Choose bar or pie graphs
STEP TEN
select paste or ok
STEP TEN
select paste or ok
STEP TEN
Analyze the output (use Table
2.1.1a)
STEP TEN
Analyze the output (use Table
2.1.1a)
HOW TO DODESCRIPTIVE
STATISTICS FOR A NO-METRIC VARIABLE?
HOW TO DODESCRIPTIVE
STATISTICS FOR A NO-METRIC VARIABLE?
65
2.1.1a: INTERPRETING NON-METRIC (or Categorical) DATA
NOMINAL VARIABLE (when there are not missing cases)
Table 2.1.1a: Gender of respondents
Frequency Percent Valid Percent
Male 150 69.4 69.4Gender:
Female 66 30.6 30.6
Total 216 100.0 100.0
Identifying Non-missing Cases: When there are no differences between the percent
column and those of the valid percent column, then there are no missing cases.
How is the table analyzed? Of the sampled population (n=21612), 69.4% were males
compared to 30.6% females.
12 The total number of persons interviewed for the study. It is advisable that valid percents are used in descriptive statistics as there may be some instances then missing cases are present with the dataset, which makes the percent figure different from those of the valid percent (Table 2.1.1b).
66
NOMINAL VARIABLE: Establishment of when missing cases
Table 2.1.1b: General Happiness
Frequency Percent Valid Percent
Very happy 467 30.8 31.1GeneralHappiness:
Pretty happy 872 57.5 58.0
Not too happy 165 10.9 11.0
Missing Cases 13 0.9 -
Total 1,517 100.0 100.0
Identifying Missing Cases: In seeking to ascertain missing data (which indicates that some of the respondents did no answer the specified question), there is a disparity between the values for percent and those in valid percent. In this case, 13 of 1,517 respondents did not answer question on ‘general happiness’. In cases where there is a difference between the two aforementioned categories (i.e. percent and valid percent), the student should remember to use the valid percent. The rationale behind the use of the valid percent is simple, the research is about those persons who have answered and they are captured in the valid percent column. Hence, it is recommended that the student use the valid percent column at all time in analyzing quantitative data.
Interpretation: Of the sampled population (n=1,517), the response rate is 99.1%
(n=1,504)13. Of the valid responses (n=1,504), 31.1% (n=467) indicated that they were
‘very happy’, with 58.0% (n=872) reported being ‘pretty happy’, compared to 11.0%
(n=165) who said ‘not too happy’.
13 Because missing cases are within the dataset (13 or 0.9%), there is a difference between percent and valid percent. Thus, care should be taken when analyzing data. This is overcome when the valid percents are used.
67
Owing to the typology of the variable (i.e. nominal), this may be presented graphical by
either a pie graph or a bar graph.
Pie graph
Male, 69.4, 69%
Female, 30.6, 31%
Figure 2.1.1: Respondents’ gender
OR
Bar graph
0
10
20
30
40
50
60
70
Male Female
Figure 2.1.2: Respondents’ gender
68
ORDINAL VARIABLE
Table 2.1.2: Subjective (or self-reported) Social Class
Frequency Percent Valid Percent
Social class:Lower 100 46.3 46.3
Middle 104 48.1 48.1
Upper 12 5.6 50.6
Total 216 100.0 100.0
Interpreting the Data in Table 2.1.2:
When the respondents were asked to select what best describe their social standing, of the sampled population (n=216), 46.3% reported lower (working) class, 48.1% revealed middle class compared to 5.6% who said upper middle class. Based on the typology of variable (i.e. ordinal), the graphical options are (i) pie graph and/or (2) bar graph.
Note: In cases where there is no difference between the percent column and that of valid percent, researchers infrequently use both columns. The column which is normally used is valid percent as this provides the information of those persons who have actually responded to the specified question. Instead of using ‘valid percent’ the choice term is ‘percent’.
69
46.348.1
5.6
05
101520253035404550
Lower class Middle class Upper middleclass
Figure 2.1.3: Social class of respondents
Or
Upper middle
class, 5.6 Lower class, 46.3
Middle class, 48.1
Figure 2.1.4: Social class of respondents
70
2.1.1b: STEPS IN INTERPRETING METRIC VARIABLE: METRIC (i.e. scale or interval/ratio)
Figure 2.1.5: Steps in Analyzing Metric data
STEP ONEKnow the
metric variable (Age)
STEP ONEKnow the
metric variable (Age) STEP TWO
Select Analyze
STEP TWO
Select Analyze
STEP THREESelect descriptive
statistics
STEP THREESelect descriptive
statistics
STEP FOUR
select frequency
STEP FOUR
select frequency
STEP FIVE
select the metric
variable
STEP FIVE
select the metric
variable STEP SIX
select statistics at
the end
STEP SIX
select statistics at
the end
STEP SEVENselect mean,
standard deviation, skewness
STEP SEVENselect mean,
standard deviation, skewness
STEP EIGHT
select Chart
STEP EIGHT
select Chart
STEP NINE
Choose histogram with normal curve
STEP NINE
Choose histogram with normal curve
STEP TEN
select paste or ok
STEP TEN
select paste or ok
STEP TEN
Analyze the output (use Table 2.1.3)
STEP TEN
Analyze the output (use Table 2.1.3)
HOW TO DODESCRIPTIVE
STATISTICS FOR A METRIC VARIABLE?
HOW TO DODESCRIPTIVE
STATISTICS FOR A METRIC VARIABLE?
71
INTERPRETING METRIC DATA: METRIC (i.e. scale or interval/ratio) VARIABLE
Table 2.1.3: Descriptive statistics on the Age of the RespondentsN Valid 216 Missing 0Mean 20.33Median 20.00Mode 20Std. Deviation 1.692Skewness 2.868Std. Error of Skewness .166
Of the sampled population (n=216), the mean age of the sample was 20 yrs and 4 months (i.e. 4 = 0.33 x 12) ± 1 yr. and 8 months (i.e. 8 = 0.692 x 12), with a skewness of 2.868 yrs. Statistically an acceptable skewness must be less than or equal to 1.0. Hence, this skewness in this sample is unacceptable, as it is an indicator of errors in the reporting of the data by the respondents. With this being the case, the researcher (i.e. statistician) has three options available at his/her disposal. They are (1) to remove the skewness, (2) not use the data – because of the high degree of errors and (3) use the median instead of the mean. It should be noted that all the measure of central tendencies (i.e. the arithmetic mean, arithmetic mode and the arithmetic median) are about the same (i.e. mean – 20.33, mode – 20.0, and median – 20.0). This situation is caused by extreme values in the data set. Hence, in this case, the arithmetic mean is disported by the values (or value) and so it is not advisable this be used to indicate the centre of the distribution. (See below how this is done in SPSS)
The figure below is to enable readers to have a systematic plan of ‘how to arrive
at the SPSS output’ for analyzing a metric variable (for example age of respondents).
Following the figure, I implement the plan in an actual SPSS illustration of how this is
done.
72
Figure 2.1.6: ‘Running’ SPSS for a Metric variable
Step One:ANALYZE
73
Figure 2.1.7: ‘Running’ SPSS for a Metric variable
Step Two:
Descriptive statistics
74
Figure 2.1.8: ‘Running’ SPSS for a Metric variable
Step Three:
select Frequency
75
Figure 2.1.9: ‘Running’ SPSS for a Metric variable
Step Four:Select the metric variable – The metric
variable – in this case is age
76
Figure 2.1.10: ‘Running’ SPSS for a Metric variable
Step Fiveselect the metric variable
from over here to
to here
77
Figure 2.1.11: ‘Running’ SPSS for a Metric variable
to the end of Step Five, you’ll see statisticsselect it
78
Figure 2.1.12: ‘Running’ SPSS for a Metric variable
Step Six:A metric variable requires that you do the following:
mean
select skewness, kurtosis
Choose the following: SD, minimum, range
79
Figure 2.1.13: ‘Running’ SPSS for a Metric variable
Step Seven:
To the end of Step Five, you will see Charts; this means you should select Histogram with normal curve
80
Figure 2.1.14: ‘Running’ SPSS for a Metric variable
Step Eight:
Highlight the argument
Step Nine:select ‘run’, which is this
Key
81
Figure 2.1.15: ‘Running’ SPSS for a Metric variable
Step Ten:
Final Output, which the researcher will now analyze
82
20 40 60 80
Age on your last birthday?
0
20
40
60
80
100
120
Fre
qu
ency
Mean = 34.95Std. Dev. = 13.566N = 1,280
Histogram
Figure 2.1.16: ‘Running’ SPSS for a Metric variable
Step Eleven:
This is pictorial of the distribution of the metric
variable, age
83
2.1.2a: MISSING (i.e. NON-RESPONSE) CASES
Table 2.1.4: “From the following list, please choose what the most important characteristic of democracy …are for you”
Frequency Percent
Open and fair election 314 23.5
An economic system that guarantees a dignified salary 177 13.2
Freedom of speech 321 24.0
Equal treatment for everybody 29522.0
Respect for minority 35 2.6
Majority rules 54 4.0
Parliamentarians who represented their electorates 52 3.9
A competitive party system 47 3.5
Don’t know/No answer 43 3.214
Total 1338 100.0Source: Powell, Bourne and Waller 2007, 11
Of the sampled population (n=1,338), when asked “From the following list, please choose what is four you the most important characteristic of democracy …?”, 23.5% (n=314) ‘open and fair elections’ 13.2% (n=177) remarked ‘An economic system that guarantees a dignified salary’, 24.0% (n=321) said ’Freedom of speech’ , 22.0% (n=295) indicated ‘Equal treatment for everybody by courts of law’, 2.6% (n=35) mentioned ‘Respect for minorities’, 4.0% (n=54) felt ‘Majority rule’, 3.9% (n=52) believed ‘Members of Parliament who represent their electors’, and 3.5% (n=47) informed that ‘A
14 “Don’t know/no answer” is an issue of fundamental importance in survey research. This is called non-response.
84
competitive party system’ compared to 3.2% (n=43) who had no answer – (i.e. ‘Don’t know/No answer), which is referred to as ‘missing values’ or, see note 4.
85
The issue of non-response becomes problematic whenever it is approximately 5%, or
more (see for example George and Mallery 2003, chapter 4; Tabachnick and Fidell 2001,
chapter 4; Thirkettle 1988, 10). Missing data are simply not just about ‘non-response’,
but they may distort the interpretation of data in case of ‘inferential statistics’. In some
instances that they are so influential that they create what is called, Type II error.
According to Thirkettle 1998, “Unless every person to be interviewed is interviewed the
results will not be valid. Non-response must therefore be kept to negligible proportions”
(Thirkettle 1988, 10). Thirkettle’s perspective is idealistic, and this is not supported by
ant of the other scholars to which I have read (see for example Babbie, Halley and Zaino
2003; George and Mallery 2003; Tabachnick and Fidell 2001; Bobko 2001; Willemsen
1974). The issue of what is an unacceptable ‘non-response rate’ is 20%. When this
marker is reached or surpassed, researchers are inclined not to use the variable. Thus, in
the case of Table 2.1.4, a non-response rate of 3.2% is considered to be negligible.
Furthermore, missing data is simply not about ‘non-response’ from the
interviewed but it is the difficulty of generalizability that it may cause, which posses the
problem in data analysis. “Its seriousness depends on the pattern of missing data, how
much is missing, and why it is missing” (Tabachnick and Fidell 2001, 58).
According to Tabachnick and Fidell (2001):
The pattern of missing data is more important than the amount missing. Missing values scattered randomly through a data matrix pose less serious problems. Nonrandomly missing values, on the other hand, are serious no matter how few of them there are because they affect the generalizability of results (Tabachnick and Fidell 2001, 58).
He continues that If only a few data points, say, 5% or less, are missing in a randomly pattern form a large data set, the problems are less serious and almost any procedure for handling missing vales yields similar results (Tabachnick and Fidell 2001, 59).
86
2.1.2b: TREATING MISSING (i.e. NON-RESPONSES) CASES
Unlike a dominant theory which is generally acceptable by many scholars, the construct
of missing data is fluid. Thus, I will be forwarding some of the arguments that exist on
the matter.
Fundamentally, the handling of missing cases primarily rest in the following categorizations. These are – (1) if the cases are less than 5%, (2) number of non-response exceeds 20% and (3) randomly or non-randomly distributed with the dataset. Scholars, such as Thirkettle (1988) ands Tabachnick and Fidell (2003) believe that in the event that the number of such cases are less than or equal to 5%, they are acceptable. On the other hand, in the event when such non-responses are more than or equal to 20%, those variables are totally dropped from the data analysis. Thus, according to Tabachnick and Fidell 2001, chapter 4; George and Mallery 2003, chapter 4, these are the available options in manipulating missing cases:
drop all cases with them; deletion of cases (i.e. this is a default function of SPSS, SAS, and
SYSTAT); impute values for those missing cases-
insert series mean15,16 mean of nearby points, median of nearby points;
using regression – (i) linear trends at point, and (ii) linear interpolation;
expectation maximization (EM)17, 18
using prior knowledge, and multiple imputation
15 “It is best to avoid mean substitution unless the proportion of missing is very small and there are no other options available to you” (Tabachnick and Fidell 2001, 66)16 “Series mean is by far the most frequently used method” (George and Mallery 2003, 50)17 “EM methods offer the simplest and most reasonable approach to imputation of missing data. as long as you have access to SPSS MVA …(Tabachnick and Fidell 2001, 66)
18 “Regression or EM. These methods are the most sophisticated and are generally recommended” (de Vaus 2002, 69)
87
CONCLUSION
The issue of how to ‘treat missing variables’ is as unresolved as the inconclusiveness of a
‘Supreme Being, God’ and as the divergence of views on the same. One scholar
forwards the view that 10% of the data cases can be missing for them to be replaced by
‘mean values’ (Marsh 1988), whereas another group of statisticians Tabachnick and
Fidell (2004) believed that not more than 5% of the cases should be absence, for
replacement by any approach. The latter scholars, however, do not think that a 5%
benchmark in and of itself is an automatic valuation for replacement but that the
researcher should test this by way of cross tabulation. This is done with some other
variable(s) in an attempt to ascertain if any difference exists between the responses and
the non-responses. If on concluding that no-difference is present between the responses
and the non-responses, it is only then that they subscribe to replacement of missing data
within the dataset. Hence, missing data are replaced by one of the appropriate
mathematical technique – ‘series mean’, ‘mean of nearby points’, ‘median of nearby
points’, ‘linear interpolation’, and/or ‘linear trends at points’.
The perspective is not the dominant viewpoint as within the various disciplines,
some scholars are ‘purist’ and so take a fundamental different stance from other who may
relax this somewhat.
One of the difficulties is for social researchers and upcoming practitioners of the
craft are to grasp – their discipline’s delimitations and some of the rationale which are
present therein in an effort to concretize their own position grounded by some
empiricism. In keeping with this tradition, I will present a discourse on the matter; and I
88
will add that scholars should be mindful of what obtains within their craft. It should be
noted that sometimes these premises are ‘best practices’ and in other instances, they are
merely guide and not ‘laws’.
On the other hand, in a dialogue with Professor of Demography at the University
of the West Indies, Mona, C. Uche, PhD., he being a ‘purist’ of the Chicago School,
believe than the arbitrary substitution of non-responses can be a misrepresentation of the
views of the non-respondents, and so he advice researcher do to take that route, even if
the cases are less than 5%.
In a monologue with Professor of Applied Sociology, Patricia Anderson, PhD.,
from the same Chicago School held the view that while it is likely to replace missing data
point for a variable, in the case in Jamaica non-response should be taken as is. She
argued that no answer, in Jamaica, is somewhat different from those who are indicated
choiced responses. Thus, if the researcher substitution ‘missing cases’ with mean value
or any other technique for that rather, he/she runs the risk of misrepresenting the social
reality.
With Marsh, Tabachnick and Fidell, Uche, and Anderson, we may conclude this
discourse has many more time left in its wake. Thus, the ‘treatment of missing values’
must be left up to the researcher within the context of society and any validation of a
chosen perspective.
89
CHAPTER 3
3.1.0: HYPOTHESIS: INTRODUCTION
All research is based on the premise of an investigation of some unknown phenomenon.
Quantitative studies, on the other hand, are not merely to provide information but it is
substantially hinged on the foundation of hypothesis testing, as this allows for some
logical way of thinking. Therefore, this chapter focuses on the continuation of Chapter 2,
while further the research process, which is the use of hypothesis, and the use of
appropriate statistical test in an effort to validate the hypothesis of the research, in
question. One author argues that it is widely accepted that studies should be geared
towards testing hypothesis (Blaikie 2003, 13). He continues that “when research starts
out with one or more hypotheses, they should ideally be derived from a theory of some
kind, preferably expressed in for of a set of propositions” (Blaikie 2003, 14).
The use of hypothesis, in objectivism, is not limited to examination of some past
theories, but without this the realities that social scientists seek to explore become more
so a maze, with no ending in sight. According to Blaikie 2003, “Hypotheses that are
plucked out of thin air, or are just based on hunches, usually makes limited contributions
to the development of knowledge because they are unlikely to connect with the existing
state of knowledge (Blaikie 2003, 14).
Thus, I will begin the definition of the construct, hypothesis. Then I will proceed
with a full interpretation of the results beginning with the germane univariate data (see
90
for example chapter 2) followed by the most suitable associational test (see chapter 1),
given the levels of measurement.
3.1.1: DEFINITIONS OF HYPOTHESIS
“A hypothesis is a preposition of a relationship between two variables: a dependent and an independent” (Babbie, Hally, and Zaino 2003, 12). The dependent variable is influenced by external stimuli (or the independent variable), and the independent variable is actually acting on its own to “cause”, or “lead to” an impact on the dependent. According to Babbie, Hally and Zaino, “A dependent variable is the variable you are trying to explain (Babbie, Hally and Zaino 2003, 13).
Boxill, Chambers and Wint (1997), on the other hand, write that a “Hypothesis – a non-obvious statement which makes an assertion establishing a testable base about a doubtful or unknown statement (Boxill, Chambers and Wint 1997, 150).
With Neuman (2003) stating that a hypothesis is “The statement from a causal explanation or proposition that has a least one independent and one dependent variable, but it has yet to be empirically tested” (Neuman 2003, 536).
Another group of scholars write that a hypothesis is “A statement about the (potential) relationship between the variables a researcher is studying. They are usually testable statements in the form of predictions about relationships between the variables, and are used to guide the design of studies.” (Furlong, Lovelace and Lovelace 2000, G8).
Every hypothesis must have two attributes. These are (1) a dependent variable, and
(2) an independent variable. Thus, embedded within each hypothesis are at least two
variables. So as to make this easily understandable, I will a few examples.
There is an association between breakfast consumption and ones academic
performance – DV (dependent variable) – academic performance; and IV
(independent variable) – breakfast consumption.
Determinants of wellbeing of the Jamaica elderly (such a hypothesis
require the use of multiple regression analysis as they possesses a number
91
of different causal factors. Hence, the DV is wellbeing. And IVs are –
educational attainment; biomedical conditions; age cohorts of the elderly
(young elderly, old-elderly and the oldest-old elderly); union status; area
of residence; social support; employment status; number of people in
household; financial support; environment conditions; income; cost of
health care; exercise;
3.1.2: TYPOLOGIES OF HYPOTHESIS
In social research hypotheses are categorized as either (1) theoretical or (2) statistical.
According to Blaikie (2003) “Statistical hypotheses deal only with the specific problem
of estimating whether a relationship found in a probability sample also exists in the
population” (Blaikie 2003, 178).
This textbook will only use statistical hypotheses. Furthermore, statistical hypotheses are
written as null, Ho19 and alternative, Ha
20. The Ho indicates no statistical association in
the population; whereas the Ha denotes a statistical association in the population between
the dependent and the independent variable (s). Furthermore, a statistical hypothesis may
be either directional or non-directional.
19 In regression analysis, the null hypothesis, Ho: β = 0.20 When using regression analytic technique, the alternative hypothesis, Ha : β ≠ 0
92
3.1.3: DIRECTIONAL AND NON-DIRECTIONAL HYPOTHESES
NON-DIRECTIONAL HYPOTHESES
Non-directional hypotheses exist whenever the researcher has not specified any direction
for the hypothesis: The examples here are as follows:
Politicians are more corrupt than Clergymen;
There is an association between number of hours spent studying and the
examination results had;
Men are less likely to be personal secretaries than women;
curative care, preventative care, social class, educational attainment, and
types of school attended are determinants of well-being
DIRECTIONAL HYPOTHESES
Directional hypotheses exist when the researcher specifies a direction for the hypothesis:
1. Positive relationship – meaning an increase in one variable sees an increase in
other variable(s): -
An increase in ones age is associated with a direct change in more
years of worked experiences;
There is a positive relationship between educational attainment and
income received;
There is a direct relationship between fertility and population
increases.
2. Negative relationship – meaning an increase in one variable result in a reduction
in other variable(s): -
93
An increase in ones age is associated with a reduction in physical
functioning;
There is an inverse relationship between educational attainment
and the fertility of a woman;
There is an inverse relationship between the number of hours the
West Indian crickets spent practice and them failing;
3.1.4a: OUTLIERS
Despite the fact that it is mathematically appropriate to compute the mean for interval and ratio data [i.e. metric or scale data], there are times when the median may be more descriptive measure of central tendency for interval and ratio data because highly irregular values (called outliers) [exist] in the data set [and these] may affect the value of the mean (especially in small sets of scores), but they have no effect on the value of the median” (Furlong, Lovelace and Lovelace 2000, 94-95).
It is on this premise that median is used instead of the mean as a measure of
central tendency. Statistically, the mean is affect by extremely large or small values,
which explains the reason for the skewness that exists in the descriptive statistics for
interval/ratio variables. Thus, care must be taken in using highly skewed data for a
hypothesis. In the event that the researcher intends to use the skewed variable as is,
he/she should ensure that the statistical test is appropriate for this situation (see Chapter
I). Otherwise, the information that is garnered is of no use.
94
In the event that outliers are detected within a variable, the researcher should
explore his/her available options before a decision is taken on any particular event. If
skewness (i.e. an indicator of outliers) is detected, this does not presuppose that mean is
inappropriate as some statisticians argue that an acceptable value is approximately ± 1.
The social research should be cognizant that outliers are not only an issue in
metric variable but may also be present in categorical variables. According to
Tabachnick and Fidell:
Rummel (1970) suggests deleting dichotomous variables with 90-10 splits between categories or more both because the correlation coefficients between these variables and others are truncated and because the scores for the cases in the small category are more influential than those in the category with numerous cases (Tabachnick and Fidell 2001, 67)
3.1.4b: REASONS for OUTLIERS
data recording entry; Instrumentation error - the item entered in the particular
category, may be different from those previously entered.
3.1.4c: IDENTIFICATION of OUTLIERS
mathematically – using skewness; graphical approach.
3.1.4d: TREATMENT of OUTLIERS
If data entry – correct this by using the questionnaire, then redo the analysis;
95
If instrumentation – drop the case(s).
96
3.1.5: STATISTICAL APPROACHES FOR ADDRESSING SKEWNESS
However, if the skewness happens to be more than the absolute value of 1 (i.e. the
numerical value without taking into consideration the sign for the value), the following
should be sought in an attempt to either (i) remove the skewness, or (ii) reduce the
skewness. These options are as follows:
i) Log10 the value;
ii) Loge or ln, the value;
iii) Square root, the variable;
iv) Square, the variable.
In the event that we are unable to reduce or remove skewness, the researcher
should not use the mean as a measure of the ‘average’ as it is affect by outliers21 which
are present within the dataset. In addition, he/she should ensure that the variable in
question, for the purpose of hypothesis testing, is in keeping with a statistical test that is
able to accommodate such a skewness (see Chapter I).
In order to provide a better understanding the construct in this text, I will present
each hypothesis in a new chapter.
21 “An outlier is a case with such an extreme value on one variable ( a univariate outlier) or such a strange combination of scores on two or more variables (multivariate outlier) that they distort statistics (Tabachnick and Fidell 2001, 66)
97
3.1.6: LEVEL OF SIGNIFICANCE and CONFIDENCE INTERVAL
Setting the level of confidence is a critical aspect of hypothesis testing in quantitative
studies. A confidence interval (CI) of 95% means that we may reject the null hypothesis,
Ho, 5% of the time (level of significance = 100% minus CI or CI = 100% minus level of
significance). According to Blaikie,
If we do not want to make this mistake [level of significance), we should set the level as high as possible, say 99.9%, thus running only a 0.01% risk. The problem is that the higher we set the level, the greater is the risk of a type II error [see Appendix II]. Conversely, the lower we set the level [of significance], the greater is the possibility of committing a type I error [see Appendix II] and the possibility of committing a type II error. (Blaikie 2003, 180)
In the attempt to complete research projects and/or assignments, we sometimes
fail to execute all the assumptions that are applicable to a particular variable. Even
though we would like to examine the association and/or causal relationships that exit
between or among different variables (i.e. hypothesis testing), this anxiety should not
overshadow ones adherence to the statistical principles, which are there to guide the
soundness of the interpretation of the figures. Thus, care is needed in ensuring that we
apply mathematical appropriateness prior to the execution of hypothesis testing.
The chapters that will proceed from here onwards will utilize the preceding
chapter and this one. In that, I will commence each chapter with a hypothesis followed
by presentation of the appropriate descriptive and inferential statistics. The social
researcher should not that the hypothesis will be separated into variables; this will allow
me to apply the most suitable inferential tools as was discussed in chapter I and II.
98
I am cognizant that undergraduate students would want a textbook that do their
particular study but this book is not that. This textbook seeks to bridge that vortex, which
is ‘how do I interpret various descriptive and inferential statistics?’ Hence, I have sought
to provide a holistic interpretation of the ‘data analysis’ section of a study, using
hypotheses. Hypothesis testing disaggregates generalizations into simple propositions
that can be verified by empirical, which is rationale for using them to depict the logical
processes in data interpretation.
99
CHAPTER 4
It may appear from you reading thus far that descriptive statistics is presented separately
from inferential statistics in your paper, and that they are disjoint. A research is a whole,
which requires descriptive and sometimes inferential statistics. It should be noted
however that a study may be entirely descriptive (see for example Probing Jamaica’s
Political Culture by Powell, Bourne and Waller 2007) or it may some association,
causality or predictability (i.e. inferential statistics). If project requires inferential
statistics, then a fundamental layer in the data analysis is the descriptive statistics. The
use of the inferential statistics rests squarely with the level of measurement, the
typologies of variable and the set of assumptions which are met by the variables.
Tabachnick and Fidell (2001) aptly summarize this fittingly when they said that:
Use of inferential and descriptive statistics is rarely on either-or proposition. We are usually interested in both describing and making inferences about a data set. We describe the data, find reliable difference or relationships, and estimate population values for the reliable findings. However, there are more restrictions on inferences than there are on description (Tabachnick and Fidell 2001, 8)
In keeping with providing a simple textbook of how to analyze quantitative data,
the previously outlined chapters have sought to give a general framework of what is
expected in the interpretation of social science research. This is only the base; as such, I
will not embark, from henceforth, to provide the readers with worked examples of
different hypotheses, in each chapter, and the inclusion of detailed interpretations of those
hypotheses, from a descriptive to an inferential statistical perspective.
100
HYPOTHESIS 1:
General hypotheses
A1. Physical and social factors and instructional resources will directly influence the
academic performance of students who will write the Advanced Level Accounting
Examination;
A2. Physical and social factors and instructional resources positively influence the
academic performance of students who write the Advanced level Accounting
examination and that the relationship varies according to gender.
B1. Pass successes in Mathematics, Principles of Accounts and English Language at
the Ordinary/CXC General level will positively influence success on the
Advanced level Accounting examination;
B2. Pass successes in Mathematics, Principles of Accounts and English Language at
the Ordinary/CXC General level will positively influence success on the
Advanced level Accounting examination and that these relationships vary based
on gender.
In answering a hypothesis in any research, the student needs to present background
information on the sampled population (or sample). This is referred to as descriptive
statistics. The description of the data is primary based on the level of measurement (see
Table 1.1.1 and Table 1.1.2) as each level of measurement requires a different approach and
statistical description. Thus, in order to examine the aforementioned hypothesis, we will
illustrate the particular description within the context of the level of measurement.
101
How to use SPSS in finding ‘Descriptive Statistics’?
The example here is finding descriptive statistics for ‘AgAge’
102
Step One: Select ‘Analyze’
103
Step Two: Select ‘Descriptive Statistics’
104
Step Four: Go to ‘Frequency’
105
Step Five: Select the ‘Frequency’ Option
By selecting the ‘frequency option’, the dialogue box that appears is as follows
This is the ‘dialogue box’
106
Step Six: Finding the ‘variable name’ for which you seek to carry out the statistical operation
Look in the left-hand side of the dialogue box for the variable in question
107
Step Seven (a): Taking the variable over to the ‘right-hand side’ of the dialogue box
The identified variable on the ‘left-hand side’ of the dialogue should be taken to the right hand side by way of this
arrow.
By selecting (or depressing) on the arrow, the variable crosses to the right hand side
108
Step Seven (b): This is what ‘step seven’ looks like -
109
Step Eight: Select ‘statistics’ in which the ‘descriptive statistics’ are contained in SPSS
By selecting ‘statistics’
Having selected ‘statistiss’ this dialogue box appears
110
Step Nine: Select the ‘appropriate’ descriptive statistics, which is based on the level of measurement
Given that the ‘variable’ is metric, we select the following options –Mean; mode; median; stand deviation, mininum or maximum, and skewness
111
Step Ten: Having chosen the ‘appropriate descriptive
statistics’, select Continue
Having selected ‘continue’, it looks like nothing has happened or back to the initial dialogue box
112
Step Eleven: Select OK.
Select OK.
113
Step Twelve: What appears after ‘Step Eleven?’
A summary of the descriptive statistics appears as well as the metric variable – in this case it is ‘Age of individual’
A summary of the descriptive statistics appears as well as the metric variable – in this case it is ‘Age of individual’
114
Step Thirteen: Producing a pictorial depiction of the ‘metric variable’
If the student needs a graphical displace of the metric variable, he/she must select ‘Graph’ at the end of the
dialogue box
Select G
raph
115
Step Fifteen: Having selected graph, we need to choose the type of ‘graph’
Based on the fact that the variable is a metric one, we should select ‘Histogram’ as well as ‘with normal curve’. The normal curve is a quick display of ‘skewness.Then select ‘continue’
Based on the fact that the variable is a metric one, we should select ‘Histogram’ as well as ‘with normal curve’. The normal curve is a quick display of ‘skewness.Then select ‘continue’
116
Step Sixteen: Select ‘continue’
Select ‘OK’, which produces the graphical display below
117
A graphical display of the ‘choosing graph’
Note: The researcher (or student) should make a table of the appropriate descriptive statistics, see overleaf.
118
ANALYSES & INTERPRETATION OF FINDINGS
SOCIO-DEMOGRAPHIC PROFILE
Table 4.1.1: Respondents’ Age
Particulars (in years)
Mean 17.48
Median 17.0
Standard deviation 1.275
Skewness 2.083
Minimum 16.000
Range 9.000
The findings reported in Table 4.1.1 shows a skewness of 2.083 years for the sampled
respondents. This is a clear indication that the age variable within the data set is highly
skewed, based on the fact that it is beyond ± 1 (see figure 4.1). As such, the researcher
assumed for the purpose of this exercise that this variable cannot be use for any further
analysis, as no method was able to reduce skewness below 1. Hence, with the mean age
of the sampled population being 17 years and approximately 6 ± 1.275 years, based on
the skewness (see Figure 4.1, below), then it follows that a better value to represent the
average is 17.0 years, the median.
119
Figure 4.1.1: AGE DESCRIPTIVE STATISTICS
120
males43%
females57%
Figure 4.1.2: Gender of Respondents22
The sample consists of 136 private and public grammar schools’ students in Kingston and
St. Andrew, Jamaica. Of the 136 respondents, one individual did not respond to most of
the questions asked including his/her age at last birth however, he/she did respond to the
question on major illnesses and on gender. Of the valid sample size (i.e. 136
interviewees), 59 were males and 77 females.
22 SPSS unlike Microsoft Excel does not specialize in graphic presentations of data, which explains a rationale why graphs in the latter are more professional than those produced by the former. Hence, I recommend that we transport the value from the SPSS’s output to Excel.
121
0.00%
5.00%
10.00%
15.00%
20.00%
25.00%
30.00%
35.00%
40.00%
45.00%
Primary/All Age Technical High
Primary/All Age
Junior High
Secondary/Traditional High
Technical High
Vocational
Teritary
Figure 4.1.3: Respondent’s parent educational level
Of sampled population, 42.4 percent of the respondents indicated that their parents had
attained a tertiary level education, with some 40.9 percent a secondary level education
and 6.1 percent a vocational level education and 10.6 percent at least a junior (all-age)
high school level education (see Figure 4.1.3 above).
122
0.00%
5.00%
10.00%
15.00%
20.00%
25.00%
30.00%
35.00%
40.00%
Mother only Father only Mother andFather
Other
Mother only
Father only
Mother and Father
Other
Figure 4.1.4: Parental/guardian composition for respondents
The findings in this research revealed that approximately 38 percent of the sampled
respondents living in a nuclear family structure (with both father and mother), with 36
percent, living with a mother only and 9.6 percent living with their fathers only (see
Figure 4.4).
123
0.00%
10.00%
20.00%
30.00%
40.00%
50.00%
60.00%
70.00%
Owned by family Rented by family
Owned by family
Rented by family
Figure 4.1.5: Home ownership of respondent’s parent/guardian
Most of the respondents indicated that their parents/guardians owned there homes (68.1
percent) with 31.9 percent stated that the family rented the homes that they occupy.
124
0.00%
10.00%
20.00%
30.00%
40.00%
50.00%
60.00%
70.00%
None One At least two
Figure 4.1.6: Respondents’ Affected by Mental and/or Physical illnesses
The results in Figure 4.6 above are not surprising. Since a large majority of the
respondents was not eating properly and furthermore their diet during the days were
predominately carbohydrates (that is, snacks or ‘drunken foods’). Some 31.4 percent of
the sampled population indicated that they had a least one type of mental illness. Of the
31.4 percent of respondents with a particular mental illness, approximately 4 percent had
at least two such types of illnesses (see Table 4.2).
125
0.00%
10.00%
20.00%
30.00%
40.00%
50.00%
60.00%
70.00%
80.00%
Yes No
Figure 4.1.7: Suffering from mental illnesses
Of the various types of mental illnesses that were investigated and responded to by the
sampled population, approximately 23 percent of the students suffered from migraine
(see Table 4.2). Moreover, the Sixth Form programme is an academic one and so
requires the continuous cognitive domain of the students; therefore, researchers even if it
does not influence the students’ academic performance must understand this
psychological issue. This issue is singled out as it the only one with a value in excess of
two percent.
126
Have32%
None68%
Figure 4.1.8: Affected by at least one Physical Illnesses
Some 31.6 percent of the sample size was affected by at least one physical illness (see
Table 4.2). The overwhelming majority of the respondents (14 percent) suffered from
asthma attacks and 2.9 percent from numbness of the hands with 1.5 percent indicated
that they had arthritis and sickle cell.
127
47.00%
47.50%
48.00%
48.50%
49.00%
49.50%
50.00%
50.50%
51.00%
51.50%
Moderate Poor
Figure 4.1.9: Dietary consumption for respondents
Although this research was not concerned with the number of calories that a male or a
female should consume daily, none of the respondents was having all the daily dietary
requirements as stipulated by the Caribbean Food and Nutrition Institute. Approximately
48.5 per cent of the respondents indicated that they were eating poorly and simple
majority reported a moderate consumption of the dietary requirements.
128
TABLE 4.1.2 (a) UNIVARIATE ANALYSIS OF THE EXPLANATORY VARIABLES
Details Frequency (%)
ACADEMIC PERFORMANCE Distinction 44 (37.9) Credit 20 (17.2) Past 46 (31.7) Fail 6 (5.2)Average Academic Performance 57.2 ± 15.423 (SD) ACADEMIC PERFORMANCE (Perception of respondent) Better 49 (39.5) Same 36 (29.0) Worse 39 (31.5)GENDER Male 58 (43) Female 77 (57)PHYSICAL EXERCISE Infrequent 38 (29.2) Moderate 10 (7.7) Frequent 82 (63.1)PSYCHOLOGICAL ILLNESSES None 92 (67.6) At least one 39 (28.7) At least two 5 (3.7)SUBJECTIVE SOCIAL CLASS Lower class 18 (15.3) Middle class 95 (80.5) Upper class 5 (4.2)PHYSICAL ILLNESS None 93 (68.4) At least one 36 (26.5) At least two 7 (5.1)CLASS ATTENDANCE Very poor 9 (8.5) Poor 37 (34.9) Good 49 (46.2) Excellent 11 (10.4)
SD represents standard deviation
23 This indicates 57.2 ± 15.4, mean and SD
129
TABLE 4.1.2(b): UNIVARIATE ANALYSIS OF EXPLANATORY
Details Frequency (%)
MATERIAL RESOURCES Low availability 10 (7.7) Moderate availability 40 (30.8) High availability 80 (61.5)BREAKFAST Frequently 4 (3.0) Moderately 127 (95.5) Infrequently 2 (1.5)Self-rated SELF CONCEPT Negative 61 (46.6) Positive 70 (53.4)AGE GROUP 16 – 17 YRS 77 (57.0) 18 – 19 YRS 52 (38.5) 20 – 25 YRS 6 (4.4)Average Age 17.7 ± 1.0 (SD)
130
Table 4.1.2 (c): UNIVARIATE ANALYSIS OF EXPLANATORY
VARIABLE FREQUENCY AND (PERCENT)
PAST SUCCESSES IN CXC/GCECOURSE: Principles of Accounts Fail 15 (11.2) Grade 1/A 49 (36.6) Grade 2/B 60 (44.8) Grade 3/C 10 (7.5) English Language Fail 8 (6.1) Grade 1/A 43 (32.8) Grade 2/B 50 (38.2) Grade 3/C 30 (22.9) Mathematics Fail 21 (16.2) Grade 1/A 20 (15.4) Grade 2/B 45 (34.6) Grade 3/C 44 (33.8)
From Table 4.2 (a), approximately 94.8 percent of the sample had an academic
performance (based on the GCE grade system) above an E while 5.2 percent of the
sample had failing scores. Academic performance was further classified into four (4)
groups as follows; 1. Distinction (i.e. grades A and B – scores from 70), 2.Credit (i.e. C),
3. Pass (i.e. D and E) and 4. Fail (i.e. scores below 40 per cent). Further, the statistics
(data) revealed that 40.0 percent of the respondents indicated that their academic
performance (test scores - grades ) in Advanced Level Accounting was better this term in
comparison to last term while 28.8 percent said their grades were the same in both terms
in comparison to 31.2 percent who said their scores were worse. This 31.2 percent
indicates a worrying fact that must be diagnosed with immediacy. In that, a marginal
131
number of prospective candidates (i.e.39.5 %) were performing better in comparison to
those who were performing worse (31.5%) (See Table 4 above)
The information in table 4 showed that 3 percent of students were consuming
breakfast on a regular basis while 1.5 percent of the same were having breakfast rarely in
comparison to 95.5 percent of them who were having the same sometimes (i.e.
moderately). Approximately 57.0 percent of the sample was between the age cohorts of
16 to 17 years, while 38.5 percent were between 17 to 19 years in comparison to 4.4
percent above 20 years. Of the sample of Advanced level Accounting students, some
61.5 percent of them had a high availability of instructional resources; some 7.7 percent
had little availability to material resources in comparison to 30.8 percent who had an
averaged availability of instructional resources.
On to the issue of self-concept, 46.6 percent of the sample of students had a low
concept of self, 29.8 percent with a moderate concept and 23.7 percent with a high
concept of themselves. This brings me to another issue, 15.3 of the sample of students
said they were from the lower class, 80.5 percent of them were from the middle class and
4.2 percent from the upper class (see Table 4.2, above).
132
STEPS IN HOW TO ‘RUN’ CROSS TABULATIONS?
One of the difficulties faced by undergraduate students is ‘how to “run”, and “interpret” quantitative data. In order that I provide assistance to this issue, I will begin the process by “running” the data in SPSS, followed by the interpretation of cross tabulations. (Steps in running cross tabulations24).
24 I am aware that some students may require assistance not only in analyzing cross tabulations, but how to ‘run’ the SPSS program. Hence, I have answered your request in this monograph. (See Appendix VI)
STEP TWELVE
Analyze the output
STEP TWELVE
Analyze the output STEP ONE
Assume bivariate
STEP ONE
Assume bivariate
STEP TWO
Select Analyze
STEP TWO
Select Analyze
STEP THREESelect
descriptive statistics
STEP THREESelect
descriptive statistics
STEP FOUR
select crosstabs
STEP FOUR
select crosstabs
STEP FIVE
in row place either DV or
IV
STEP FIVE
in row place either DV or
IV
STEP SIX
in column vice versa to Step
5
STEP SIX
in column vice versa to Step
5
STEP SEVEN
select statistics
STEP SEVEN
select statistics
STEP EIGHT
choose chi-Square,
contingency coefficient and
Phi
STEP EIGHT
choose chi-Square,
contingency coefficient and
Phi
STEP NINE
select cells
STEP NINE
select cells
STEP TEN
in percentage, select – row, column and
total
STEP TEN
in percentage, select – row, column and
total
STEP ELEVEN
select paste or ok
STEP ELEVEN
select paste or ok
HOW TO
RUN CROSS TABULATIONS, in
SPSS?
HOW TO
RUN CROSS TABULATIONS, in
SPSS?
133
HOW TO ‘ANALYSE’ CROSS TABULATIONS – when there is no statistical relationship?
Table 4.1.3: Bivariate relationships between academic performance and subjective social class (in %), N=99
Subjective Social Class
Lower MiddleUpper
Academic Performance
Distinction 40.0 37.0 33.3
Credit 6.7 21.0 0.0
Pass 46.6 37.0 66.7
Fail 6.7 5.0 0.0
Total 15 81 3
χ 2 (4)= 3.147, ρ value = 0.790
From Table 4.1.3, there is no statistical relationship between subjective social class and
academic performance [χ 2 (6)25 = 3.147, p= 0.790 >0.0526] based on the population
sampled. The Chi square analysis27 was contrasted with Spearman’s correlation, at the
two (2) tailed level; and the latter’s Ρ value = 0.883, again indicating that there was no
statistical correlation between subjective social class and academic performance based on
the population sampled. Statistically this could be a Type II error (see Appendix II). (Note – The analysis does not go beyond what is written, if there is not relationship).
Table 4.1.4: Bivariate relationships between comparative academic performance and subjective social class (in %), N=10825 The ‘6’ is the degree of freedom, df, which is calculated as (number of rows minus 1) times (number of columns minus 1)26 In this case the level of significance, 5%, is an arbitrary point that the researcher assumes the outcome will be biased, or The probability of rejecting a true null hypothesis; that is, the possibility of make a Type I Error. In this case there is a Type II error (See Appendix II)27 The social researcher needs to understand that when analyzing Chi Square, one should use the values for the independent variables. If the independent variable is in the column, use the column percentages. However, if the independent variable is in the row, use the row percentage for your analysis.
134
Subjective Social Class
Lower Middle Upper
ComparativeAcademic Performance
Better 31.3 41.4 20.0
Same 37.4 27.6 40.0
Worse 31.3 31.0 40.0
Total 16 87 5
χ 2 (4) = 1.597, ρ value = 0.809
The results (in Table 4.1.4) indicate that there is no statistical relationship [χ 2(4) = 1.597,
ρ value 0.809 >0.05] between subjective social class and past and-or present academic
performance of the sampled population over the Christmas term in comparison to the
Easter term. Even when Spearman’s correlation, at the two-tailed level, was used the P=
0.999 indicating that there was no statistical correlation between the two variables based
on the population sampled.
135
HOW TO ‘ANALYSE’ CROSS TABULATIONS – when there is no statistical relationship?
TABLE 4.1.5: BIVARIATE RELATIONSHIPS BETWEEN ACADEMIC PERFORMANCE AND PHYSICAL EXERCISE (in %), N= 111
Physical Exercise
Infrequently Moderately Frequently
Academic Performance
Distinction 39.4 12.5 41.4
Credit 27.3 12.5 14.3
Pass 33.3 62.5 38.6
Fail 0.0 12.5 5.7
Total 33 8 70
χ 2 (6) = 8.066, ρ value = 0.233
The results (in Table 4.1.5) indicated that there was no statistical relationship between
physical exercise and academic performance [χ2 (6) = 8.66, ρ value = 0.233 > 0.05]
based on the population sampled.
NOTE: Whenever there is no statistical association (or correlation) between variables,
the researcher cannot examine the figure for difference as there is no statistical difference
between or among the values.
136
HOW TO ‘ANALYSE’ CROSS TABULATIONS – when there is a statistical relationship?
Table 4.1.6 (i): Bivariate relationships between academic performance and instructional materials (in %), N=113
Instructional Materials
Infrequently Moderately Frequently
Academic Performance
Distinction 20.0 26.4 45.9
Credit 0.0 11.8 21.6
Pass 40.0 61.8 28.4
Fail 40.0 0.0 4.1
Total 5 34 74
χ 2 (6) = 27.45528, ρ value = 0.00129
Based on Table 4.1.6(i), the results indicated that there was a statistical relationship
between material resources (i.e. instructional materials) and academic performance [χ
2(2) = 27.455, ρ value = 0.001 <0.05] based on the population sampled. The strength of
the relationship is moderate (cc = .44230 or 44.2 % - See Appendix) and this indicated,
there is a positive relationship between resources and better academic performance.
Based on the coefficient of determination, instructional resources explain approximately
28 This is the Chi Square value (27.455), which is found in the Chi Square Test29 This figure, 0.0000 (which should be written as 0.001), is found in the Symmetric Measures Table (it is the Approx sig.) – (see for example Corston and Colman 2000, 37)30 Correlations coefficients, cc, or phi, ф, indicates (1) magnitude of relationship, (2) direction of the association, sign , and (3) strength.
137
20 percent of the proportion of variation in academic performance of the population
sampled.
Of the students who had indicated infrequent use of instructional materials, 20.0
percent received distinction compared to 26.4 percent of those with moderate use of
material resources and 45.9 percent of those with a high availability of instructional
materials. Forty percent of those who indicated low (ie infrequent use) of material
resources failed their last test compared to 0.0 percent of those who indicated moderate
use of instructional materials and 4.1 percent of those who frequent use material
resources.
138
Table 4.1.6 (ii) Relationship between academic performance and materials resource among students who will be writing the A’ Level Accounting examination By Gender (in %), 2004, N=103
Instructional Resources Instructional Resources
Low Moderate High Low Moderate High
Male31 Female32
Distinction 0.0 14.3 59.3 50.0 35.0 38.3
Academicperformance: Credit 0.0 0.0 22.2 0.0 20.0 21.3
Pass 66.7 85.7 14.8 0.0 45.0 36.2
Fail 33.3 0.0 3.7 50.0 0.0 4.3
Total 3 14 27 2 20 37
From Table 4.1.6 (ii) above, the results indicated that there was a statistical significant
relationship between availability of resource materials and academic performance of
males and not for females based on the population sampled. The relationship between
instructional resources and academic performance was only explained by the male
gender. The strength of the relationship was strong (cc = 0.62), meaning that males
performance is positively related to the availability of instructional resources. Based on
the coefficient of determination, 38.6 percent the proportion of variation of the academic
performance among males was explained by material resources based on the population
sampled.
31 χ2 (1) = 27.65, ρ value = 0.001, n= 44
32 χ2 (1) = 12.076, ρ value = 0.060, n= 59
139
Approximately 59 percent of males who had a high availability of resource
materials obtained distinction compared 14 percent of them had moderate number of
resource materials and zero percent had low availability of materials. Twenty two
percent of those who had a high availability of instructional materials at their disposal
received credit on their last Accounting test; zero percent had low and moderate
availability of instructional resources. Approximately 15 percent of those who had a high
availability of resource materials passed their last test; 86 percent of them had moderate
number of instructional materials in comparison to 67 percent with a low availability of
materials. Furthermore, the data revealed that 3.7 percent of those who had a high
availability of instructional materials failed their last Accounting test in comparison to
33.3 percent and 0.0 with low and moderate availability of materials respectively.
140
Table 4.1.7: Bivariate relationships between academic performance and class attendance (in %), N= 90
Class Attendance
Very poor Poor Good Excellent
Academic Performance
Distinction 33.3 31.0 37.0 60.0
Credit 0.0 24.1 19.5 10.0
Pass 50.0 41.4 37.0 30.0
Fail 16.7 3.5 6.5 0.0
Total 6 29 46 10
χ 2 (6) =6.423, ρ value = 0.697
The results (in Table 4.17) indicate that there was no statistical relationship between
class attendance and academic performance (χ 2(9) = 6.423, ρ value = 0.697 >0.05) of the
population sampled. The researcher further investigated this phenomenon and found that
there is a statistical correlation (using Spearman’s correlation) between comparative
academic performance (i.e. students’ performance this term - Easter in comparison to last
term – Christmas) and class attendance (P=0.047). With this finding, the researcher used
Chi-Square Analysis and it showed that there was no statistical correlation between the
two (2) previously mentioned variables based on the population sampled (see Table 4.1.9
(b) overleaf).
141
Table 4.1.9: Bivariate relationships between academic performance By Breakfast consumption (in %), N=114
Breakfast consumption
Frequently Moderate None
Academic Performance
Distinction 0.0 39.8 0.0
Credit 75.0 15.7 0.0
Pass 25.0 38.9 100
Fail 0.0 5.6 0.0
Total 4 108 2
χ 2 (6) =12.878, ρ value = 0.045
Based on Table 4.1.9 above, the results indicate that there is a positive relationship
between breakfast consumption and academic performance (χ 2(6) = 12.878, ρ value
0.045 <0.05). The results indicated that there is a statistical significant relationship
between the two variables previously mentioned based on the population sampled. Being
an in increase of breakfast will see an increase in ones academic performance. It should
be noted that the strength of the relationship is weak (cc = 0.319). Nevertheless, 10.18
percent of the proportion of variation in academic performance was explained by
consuming breakfast (the coefficient of determination).
Approximately 40 percent of those who had breakfast received distinction on their
last Accounting test in comparison to zero in the category of frequently and none.
Seventy five percent of those who frequently had breakfast got credit on the last
Accounting test in comparison to 16 percent who had the same on a moderate basis, and )
142
percent who had none. On the other hand, 25.0 percent of those who did not consume
breakfast on a regular passed the last Accounting test in comparison to 38.9 percent who
had the same on a moderate basis and 100 percent of them saying no breakfast
whatsoever. In regards to breakfast consumption, 5.6 percent of those who had breakfast
on a moderate basis failed their last Accounting test compared to 0 percent who had none
and 0 percent had it on a frequent basis
Table 4.1.10: Relationship between academic performances and breakfasts consumption among A’ Level Accounting students, controlling for gender, N=103
Breakfast consumption Breakfast consumption
Freq Moderate None Freq Moderate None
Male33 Female34
Distinction 0.0 39.5 0.0 0.0 40.0 0.0
Academicperformance: Credit 100.0 11.6 0.0 66.7 18.5 0.0
Pass 0.0 44.2 100.0 33.3 35.4 100.0
Fail 0.0 4.7 0.0 0.0 6.1 0.0
Total 1 43 1 3 65 1
The results (in Table 4.1.10) indicate that there is no statistical relationship between
academic performance and eating breakfast when controlled for gender (χ 2(6) =7.884
and 6.478 for males and females respectively with Ρ value s >0.05. Therefore, gender
does not explain the statistical relationship between eating breakfast and academic
performance. 33 χ2 (1) = 27.65, ρ value = 0.24, n= 45
34 χ2 (1) = 6.478, ρ value = 0.37, n= 69
143
Table 4.1.11: Bivariate relationships between academic performance By Migraine (in %), N=116
Migraine (i.e. Health condition)
No Yes
Academic Performance
Distinction 38.2 37.0
Credit 15.7 22.2
Pass 40.5 37.0
Fail 5.6 3.8
Total 89 27
χ 2 (6) =0.721, ρ value = 0.868
Based on Table 4.1.11 above, the results indicate that there is no statistical relationship
between migraine and academic performance (χ 2(2) = 0.898, p>0.05) of the population
sampled.
144
Table 4.1.12: Bivariate relationships between academic performance and Self-reported mental illnesses, N=113
Self-reported Mental Illness
None One At least two
Academic Performance
Distinction 40.5 24.2 100.0
Credit 15.2 24.2 0.0
Pass 38.0 48.6 0.0
Fail 6.3 3.0 0.0
Total 79 33 4
χ 2 (6) =10.647, ρ value = 0.100
Based on Table 4.1.12 above, the results indicate that there is no statistical relationship
between the experienced mental illnesses and academic performance (χ 2(6) = 10.647, ρ
value >0.05). Even when Spearman’s rho35 correlation, at the two-tailed level, was used
the P (value) = 0.967 that indicates no statistical correlation between the variables of the
population sampled.
35 The rho in Spearman is interpreted similar to that of the r in the Pearson’s Product-Moment Correlation Coefficient (See for example Downie and Heath 1970, 123)
145
Table 4.1.13: Bivariate relationships between academic performance and physical illnesses, (n=116)
Physical IllnessNone One At least two
Academic Performance
Distinction 38.7 34.5 42.8
Credit 17.5 17.2 14.4
Pass 37.5 44.8 42.8
Fail 6.3 3.5 0.0
Total 80 29 7
χ 2 (6) =1.204, ρ value = 0.977
Based on Table 4.1.13 above, the results indicate that there is no statistical relationship
between academic performance and physical illnesses (χ 2(6) = 1.204, p>0.05) based on
the population sampled. Even when Spearman’s correlation, at the two-tailed level, was
used the P (value) = 0.912 that indicates no statistical correlation between the variables
based on the population sampled.
146
Table 4.1.14: Bivariate relationships between academic performance and general illness (n=116)
General IllnessNone At least One
Academic Performance
Distinction 38.7 36.1
Credit 17.5 16.7
Pass 37.5 44.4
Fail 6.3 2.8
Total 80 36
χ 2 (6) = 0.936, ρ value = 0.817
Based on Table 4.1.14 above, the results indicate that there is no statistical relationship
between physical illnesses and academic performance (χ 2(3) = 0.936, p>0.05) of this
population sampled.
147
Table 4.1.15. Bivariate relationships between current academic performance and past performance in CXC/GCE English language examination, (n= 112)
Past performance in CXC English language
GRADE 1/A GRADE 2/B GRADE 3/C FAILAcademic Performance
Distinction 37.1 40.9 36.0 50.0
Credit 22.8 11.4 16.0 25.0
Pass 28.6 45.4 44.0 25.0
Fail 11.4 2.3 4.0 0.0
Total 35 44 25 8
χ 2 (6) = 7.955, ρ value = 0.539
Based on Table 4.1.15, the results indicate that there is no relationship between past
performance in English Language at the Caribbean Examination Council (CXC) or the
Ordinary Level and academic performance at the Advanced level (in Accounting) (χ 2(9)
= 7.955, p>0.05). This result continued even when Spearman’s correlation, at the two-
tailed level, was used with a P (value) = 0.581 indicating no statistical correlation
between past success in English Language at the Ordinary Level or the General
Proficiency level (i.e. CXC) and academic performance in Advanced Level Accounting.
148
Table 4.1.16: Bivariate relationships between academic performance and past performance in CXC/GCE English language examination, controlling for gender
Gender Value df Asymp. Sig. (2-sided)MALE Pearson Chi-
Square 10.752(a) 9 .293
Likelihood Ratio 11.092 9 .269 Linear-by-Linear
Association.812 1 .367
N of Valid Cases43
FEMALE Pearson Chi-Square
3.258(b) 9 .953
Likelihood Ratio 3.353 9 .949 Linear-by-Linear
Association.002 1 .969
N of Valid Cases 69
P (value) > 0.05 for both gender
Table 4.1.16 shows clearly that the academic performance of A’ Level candidates are not
statistical related by past performance in CXC/GCEEnglish language. As irrespective of
the gender of the population sampled the Ρ value was greater than 0.05 (i.e. 0.293 and
0.953 for males and females respectively).
149
Table 4.1.17: Bivariate relationships between academic performance and past performance in CXC/GCE Mathematics examination n= 101
Past Performance in CXC/GCE Mathematics
Poor Moderate Good ExcellentAcademic Performance
Distinction 31.58 55.56 44.74 38.46
Credit 26.32 16.67 10.53 26.92
Pass 36.84 27.78 36.84 26.92
Fail 5.26 0.00 7.89 7.69
Total 19 18 38 26
χ 2 (9) = 7.745, ρ value = 0.560
Based on Table 4.1.17, the results indicate that there is no statistical relationship between
past performance in CXC/GCE Mathematics examination and today’s academic
performance in Advanced level Accounting (χ 2(9) = 7.745, p>0.05). Even when
Spearman’s correlation, at the two-tailed level, was used the P (value) = 0.196 which
represents no correlation between the two variable of the population sampled.
150
Table 4.1.18 (i): Bivariate relationships between academic performance and past performance in CXC/GCE principles of accounts examination (n= 114)
Past Performance in CXC/GCE Mathematics
Poor Moderate Good ExcellentAcademic Performance
Distinction 30.0 52.1 26.5 28.6
Credit 20.0 22.9 12.2 14.3
Pass 40.0 20.8 59.2 42.9
Fail 10.0 4.2 2.0 14.3
Total 10 48 49 7
χ 2 (9) = 17.968, ρ value = 0.036
Based on Table 4.1.18 (i), the results indicated that there was a statistical relationship
between past performance in Principles of Accounts (POA) at the CXC/GCE level and
present academic performance at the A’Level (χ 2(9) = 17.968, p<0.05). The results
indicated that better a grade in POA at the Ordinary level is directly related to better
performance in A’Level Accounting based on the population sampled. The strength of
the relationship is moderate (cc = .4). Approximately 14 percent of the proportion of
variation in academic performance is explained by passed performance in POA at the
Ordinary level coefficient of determination).
Based on Table 4.1.18, of the self-reported past performance in CXC/GCE
Mathematics, of those who indicated a moderate grade, 52.1% of them claimed that they
have been receiving distinction in A’Level Accounting (ie class work) compared to 30%
who had received a poor grade in CXC/GCE Mathematics, 26.5% of good CXC/GCE
151
grade in Mathematics and 28.6% who mentioned an excellent grade in Mathematics.
Only 10.0% of those who claimed a poor grade in CXC/GCE Mathematics were failing
A’Level Accounting class work compared to 4.2% of those with moderate, 2.0% with
good and 14.3% of an excellent Mathematics score from CXC/GCE Mathematics.
Embedded in this finding is the contribution of some mathematical skills in good
performance in A’Level Accounting. Excellent mathematical skills are not need to score
distinctions in A’Level Accounting, but it aids in current performance on A’Level
Accounting.
152
Table 4.1.20: Bivariate relationships between academic performance and self-concept (n= 112)
Self-reported Self-concept
Low Moderate HighAcademic Performance
Distinction 37.5 46.7 34.6
Credit 23.2 16.7 7.7
Pass 33.9 36.7 50.0
Fail 5.4 0.0 7.7
Total 56 30 16
χ 2 (9) = 6.307, ρ value = 0.390
Based on Table 4.1.20 above, the results indicate that there is no statistical relationship
between the self-concept of the A’ Level students and their academic performance (χ 2(6)
= 6.307, p>0.05) of the population sampled. Spearman’s correlation, at the two-tailed
level, concurred [P (value) was 0.541] with the Chi-Squared results above that there was
no statistical correlation between ones concept of self and academic performance.
Furthermore, even when the researcher looked at self-concept as being positive or
negative, there was no statistical significance between it and academic performance [χ 2
(2) = 2.672, P (value)>0.05] of the population sampled.
153
Table 4.1.21: Bivariate relationships between academic performance and dietary requirements (n=116)
Dietary Requirements
Poor Moderate Good ExcellentAcademic Performance
Distinction 35.8 39.7 NA NA
Credit 17.0 7.5 NA NA
Pass 41.5 38.1 NA NA
Fail 5.7 4.8 NA NA
Total 53 63 0 0
χ 2 (9) = 0.245, ρ value = 0.970
From Table 4.1.21 above, the results indicate that there was no statistical relationship
between dietary requirements and students’ academic performance (χ 2(9) = 0.245,
p>0.05) of the population sampled.
154
TABLE 4.1.22: SUMMARY OF TABLES
VARIABLES – Sampled population (χ 2(2) )
Rejected Null Hypotheses:
ACADEMIC PERFORMANCE and MATERIAL RESOURCES 114 (0.001) ACADEMIC PERFORMANCE and BREAKFAST 114 (0.045)
ACADEMIC PERFORMANCE and PAST SUCCESS IN CXC/GCEPOA 114 (0.036)
COMPARATIVE ACADEMIC PERFORMANCE and INSTRUCTIONAL RESOURCES 103 (0.054)
Fail to Reject Null hypotheses:
ACADEMIC PERFORMANCE and dietary requirements 116 (0.970)
ACADEMIC PERFORMANCE and Self concept 112 (0.390)
ACADEMIC PERFORMANCE and Mathematics 112 (0.560)
ACADEMIC PERFORMANCE and English Language 112 (0.539)
ACADEMIC PERFORMANCE and Physical Illness 116 (0.817)
ACADEMIC PERFORMANCE and Mental Illness 116 (0.603)
ACADEMIC PERFORMANCE and Migraine 116 (0.868)
ACADEMIC PERFORMANCE and Class Attendance 106 (0.697)
ACADEMIC PERFORMANCE and Physical Exercise 110 (0.233)
ACADEMIC PERFORMANCE and Subjective Social Class 108 (0.790)
COMPARATIVE ACADEMIC PERFORMANCE and Subjective Social Class 99 (0.790)
CHAPTER 5
155
HYPOTHESIS 2:
General hypothesis
There is a relationship between religiosity, academic performance, age and marijuana
smoking of Post-primary schools students and does this relationship varies based on
gender.
TABLE 5.1.1: FREQUENCY AND PERCENT DISTRIBUTIONS OF EXPLANATORY MODEL VARIABLES
VARIABLE FREQUENCY AND PERCENT
MARIJUANA SMOKING Non-Usage 7,356 (92.5%) Usage 593 (7.5%)
RELIGIOSITY Low 351 (4.4%) Moderate 1,365 (78.3%) High 6,197 (78.3%)
AGE Less Than & Equal 15 Years 4,452 (55.7%) Greater Than & Equal 16 Years 3,543 (44.3%)
ACADEMIC PERFORMANCE Below Average 645 (8.2%) Average 690 (8.8%) Above Average 6,510 (83.0%)
GENDER Male 3,558 (44.5%) Female 4,437 (55.5%)
The sample consisted of 7,996 post-primary school Jamaican students.
Approximately 7.5 percent (N= 593) of the sample was marijuana smokers compared
156
with 92.5 percent (N= 7,356) who were not. From Table 3 (above), 78.3 percent (N=
6,197) of the sample was highly religious individuals compared with 4.4 percent (N=
351) were of low religiosity and 17.3 percent (N=1,365) of moderate religiosity.
Furthermore, the findings revealed that approximately 55.7 percent (N= 4,452) of the
sample was below or equal to 15 years of age while 44.3 percent (N= 3,543) were above
or equal to 16 years of age. Of the sample of post-primary school students, some 83.0
percent (N= 6,510) of them got grades beyond 70 percent compared with 8.2 percent
(N=645) whose grades were below 50 percent while 8.8 percent (N= 690) got average
grades. The grades were compiled from data between June and September 1996. In
addition, males constituted approximately 45 percent (N= 3,558) of the sample compared
with 55 percent (N= 4,437) females (See Table 5.1.1).
BIVARIATE RELATIONSHIPS
Table 5.1.2: RELATIONSHIP BETWEEN RELIGIOSITY AND MARIJUANA
SMOKING (N=7,869)
RELIGIOSITY
MARIJUANA
SMOKING
Number and Percent Number and Percent Number and Percent
Low Moderate High
Non-Usage 294 (84.2%) 1,213(89.2%) 5,780(93.8%)
Usage 55 (15.8%) 147(10.8%) 380(6.2%)
2= 72.313, Ρ value <0.05
Based on the Table 5.1.2, the results indicated that there is a relationship between
religiosity and marijuana smoking (2(2) = 72.313, p<0.05). From the findings there was
a significant relationship between the two variables previously mentioned.
Approximately 84 percent (N= 294) of respondents who were of low religiosity were
157
non-smokers compared with 89 percent (N= 1,213) of moderate religiosity and 94
percent (N= 5,780) had high religiosity. Also, approximately 6 percent (N=380) of
respondents who indicated high religiosity were marijuana smokers compared to 11
percent (N=147) with moderate religiosity while 16 percent (N=55) who had low
religiosity. From the findings, students of low religiosity have a higher probability of
smoking “weed” in comparison to high believer cohort. The strength of the relationship
is very weak (Phi = 0.09542); although, 0.645 percent (i.e. coefficient of determination)
of the proportion of variation in marijuana smoking was explained by religiosity.
158
Table 5.1.3: RELATIONSHIP BETWEEN RELIGIOSITY AND MARIJUANA
SMOKING CONTROLLED FOR GENDER
RELIGIOSITY
MARIJUANA
SMOKING
Number and
Percent
Number and
Percent
Number and
Percent
Low Moderate High
Non-Usage Male 152(78.4%)
Female 142(91.6%)
Male 673(84.7%)
Female 540(95.6%)
Male 2,231(90.1%)
Female 3,549 (96.3%)
Usage Male 42(21.6%)
Female 13(8.4%)
Male 122(15.3%)
Female 25(4.4%)
Male 244(9.9%)
Female 136(3.7%)
Table 5.1.3 results indicated that there was a statistical significant relationship
between religiosity and marijuana smoking irrespective of the sampled gender. From the
findings, the data for the males revealed a 2(2) = 36.708 with a Ρ value of 0.001
compared with 2(2) = 9.032 with a Ρ value of 0.0109 for the females. Furthermore, 21.6
percent (N=42) of males who smoked ganja either no religiosity or a low religiosity
compared with 8.4 percent (N=13) for the females. Of the smokers who had a high belief
religion, 9.9 percent were males compared with only 3.7 percent who were females.
With regard to the non-smokers, of those who have a high religiosity 90.1 percent (N=
2,231) were males compared with 96.3 percent (N=3,549) who were females. Of the
non-smokers with a low religiosity, there were significantly more females (91.6 %)
compared with males (78.4%). Even though there was a statistical relationship between
159
religiosity and marijuana smoking and that gender did not alter this association, the
strength of the relationship for male is very weak (cc = 0.1024) and this was equally so
for females (cc = 0.04524). The relationship between the stated variables was even
weaker for females (4.4%) compared with that of males (10.24%) with a coefficient of
determination (i.e. this explains the proportion of variation of the smoking marijuana due
to religiosity) of 0.8876 percent for males and 0.0901 for females. The interpretation
here is, 8.876 percent of the variation in “weed” smoking is explained by maleness
compared with 9.01 which is explained by femaleness.
160
Table 5.1.4: RELATIONSHIP BETWEEN AGE AND MARIJUANA SMOKING
(N=7,948)
AGE OF RESPONDENTS
MARIJUANA
SMOKING
Number and Percent Number and Percent
≤ 15 years ≥ 16 years
Non-Usage 4,143(93.6%) 3,213(91.3%)
Usage 285(6.4%) 307(8.7%)
Ρ value < 0.05
The results indicated that there is a relationship between the age of the sampled
respondents and marijuana smoking (2(2) = 14.8567, Ρ value = 0.001). Based on Table
5.1.4, the findings indicated that there is a significant relationship between the two
variables previously mentioned but the strength of this relationship is very weak (Phi =
0.04323).
Approximately 94 percent (N= 4,143) of respondents who were less than or equal
to 15 years old were non-smokers compared with 91 percent (N=3,213) of those 16 years
and older. On the other hand, approximately 6 percent (N=285) of respondents 15 years
and less were smokers in comparison to 9 percent (N=307) 16 years and older. From
Table 6, 0.19 percent of the proportion of variation in marijuana smoking was explained
by the age of the sampled population (i.e. coefficient of determination).
161
Table 5.1.5: RELATIONSHIP BETWEEN MARIJUANA SMOKING AND AGE OF
RESPONDENTS, CONTROLLED FOR SEX
AGE OF RESPONDENTS
MARIJUANA
SMOKING
Number and Percent Number and Percent
Ρ value s
Less Than & Equal to 15
Years
Greater Than & Equal 16
Years
Non-Usage Male 1788 (89.7%)
Female 2355(96.8%)
Male 1320(86.2%)
Female 1893(95.2%)
0.001
0.009
Usage Male 206 (10.3%)
Female 79 (3.2%)
Male 212(13.8%)
Female 95(4.8%)
0.001
0.009
From Table 5.1.5, despite the sampled population gender, the results indicated
that there was a statistical significant relationship between age of the respondents and
‘weed’ smoking 2(1) = 14.8567, Ρ value = 0.001 and 2(1) = 10.19793, Ρ value = 0.001
for males and females respectively). The strength of the relationship with regard to male
sample is very weak (Phi = .05378) and even weaker for the female sampled population
(Phi = .03922). The findings revealed that 0.2892 percent of the variation in marijuana
smoking was due to the males’ age compared with 0.01538 for females (i.e. Coefficient
of determination).
The findings showed that, 10.3 percent (N=206) of males who were less than
and/or equal to 15 years of age were smokers compared with 3.2 percent (N=79) of
females. On the other hand, 13.8 percent (N=212) of respondents 16 years and older
were smoked marijuana compared with only 4.8 percent (N=95) were females.
162
Some 89.7 percent (N=1,788) of male respondents less than or equal to 15 years
of age were non-smokers compared to 96.8 percent (N=2,355) female respondents.
Furthermore, 86.2 percent (N=1,320) of male respondents ages 16 years and older were
non-smokers compared to 95.2 percent (N=1,893) of females of the same age.
Table 5.1.6: RELATIONSHIP BETWEEN ACADEMIC PERFORMANCES AND MARIJUANA SMOKING, (N=7,808)
ACADEMIC PERFORMANCE
MARIJUANA SMOKING
Number and
Percent
Number and
Percent
Number and
Percent
Above Average Average Below Average
Non-Usage 643 (93.6%) 6027 (93.0%) 556 (86.6%)
Usage 44 (6.4%) 452 (7.0%) 86 (13.4%)
ρ<0.05
The findings indicated that there was a statistical relationship between academic
performance and marijuana smoking (2(2) = 36.094, p<0.001), very weak statistical
correlation (cc = 0.06783). Based on Table 8, approximately 94 percent (N=643) of
those who had an academic performance that was above average were non-smokers
compared with 87 percent (N=556) of those with an academic performance of below
average and 93% at the average level. Approximately 6 percent (N=44) of respondents
who had an academic performance above average were smokers in comparison to 13
percent (N=86) of them with an academic performance below average and 7 percent at
the average grade.
163
Table 5.1.7: RELATIONSHIP BETWEEN ACADEMIC PERFORMANCES AND MARIJUANA SMOKING, CONTROLLED FOR GENDER
ACADEMIC PERFORMANCES
MARIJUANA
SMOKING
Number and
Percent
Number and
Percent
Number and
Percent
Above Average Average Below Average
Non-UsageMale 272 (88.3%)
Female 371(97.9%)
Male 2439 (88.9%) Female 3588(96.1%)
Male 328 (82.2%) Female 228 (93.8%)
Usage Male 36 (11.7%)
Female 8(2.1%)
Male 305(11.1%) Female 147(3.9%)
Male 71(17.8%) Female 15(6.2%)
ρ value < 0.05
Based on the findings, irrespective of the gender of the sampled population, there was a
significant statistical relationship between academic performance and marijuana smoking
(2(2) = 14.80237, ρ value = 0.001 and 2(2) =6.59627, ρ value = 0.037 for males and
females respectively). The strength of the association between the variable for male is
very weak (cc = 0.06549) and even weaker for females (cc = 0.03888).
From Table 9, 11.7 percent (N=36) of respondents with academic performance
that was above average and less than or equal to 15 years of age smoked ganja compared
to 2.1 percent of female respondents of the same age. Some 17.8 percent (N=71) of
respondents who indicated that their academic performance was below average were
males compared to 6.2 percent of female respondents.
164
Continuing, there were approximately 6 times more male than female respondents
who had an academic performance in excess of average compared to approximately 3
times more male than respondents who obtained less than below average performance.
Furthermore, at an average academic performance level, there were approximately 3
times more male than female respondents.
165
TABLE 5.1.8: SUMMARY OF TABLES Dependent Variable
MARIJUANA SMOKING
Independent Variables
Religiosity Low Moderate High
Religiosity (controlled)male lowmale moderatemale highfemale lowfemale moderatefemale high
Non-Usage
294 (84.2%)***1213 (89.2)***5780 (93.8)***
152 (78.4%)***673 (84.7%)***
2231 (90.1%)***142 (91.6%)***540 (95.6%)***
3549 (96.3%)***
Usage
55 (15.8%)***147 (10.8%)***
380 (6.2)***
42 (21.6%)***122 (15.3%)***244 (9.9%)***13 (8.41%)***25 (4.4%)***
136 (3.7%)***
Academic Performance Above Average Average Below Average
Academic Performance (controlled)male above averagemale averagemale below averagefemale above averagefemale averagefemale below average
643 (93.6%)***6027 (93.0%)***556 (86.6%)***
272 (88.3%)***2439(88.9%)***328 (82.2%)***371 (97.9%)***
3588 (96.1%)***228 (93.8%)***
44 (6.4%)***452 (7.0%)***86 (13.4%)***
36 (11.7%)***305 (11.1%)***71 (17.8%)***
8 (2.1%)***147 (3.9%)***15 (6.2%)***
166
Age 15 and below 16 and above
Age (controlled)male 15 and belowmale 16 and abovefemale 15 and belowfemale 16 and above
4143(93.6%)***3213 (91.3%)***
1788 (89.7%)***1320 (86.2%)***2355 (96.8%)***1893 (95.2%)***
285 (6.4%)***307(8.7%)***
206 (10.3%)***212 (13.8%)***
79 (3.2%)***25 (4.8%)***
Note: *** represents a Ρ value < 0.05
167
CHAPTER 6
Hypothesis 3: There is a statistical difference between the pre-Test and the post-Test scores.
Analysis of Findings
SOCIO-DEMOGRAPHIC INFORMATION
43%
57%
male female
Figure 6.1.1: Gender Distribution
Of the sampled population of 68 students, 57 percent (n = 39) were females compared to
43 percent (n = 29) males; (See Figure 6.1.1, above) with an averaged age of 14 years 10
months (14.87 yrs.) ± 0.420 years, and a minimum age of 14 years and a range of 2 years
(See Table 4.1, below). The sample was further categorized into two groupings. Group
One (i.e. the Experimental) had 52.9 percent (n = 36) students compared to Group Two
with 47.1 percent (n = 32). In respect the class distribution of the sample, 52.9 percent
168
(n = 36) were in grade 9 Class One compared to 47.1 percent (n =32) who were in grade
9 Class Two.
primary all age preparatory
Figure 6.1.2: Typology of previous School
Based on Figure 6.1.2 (above), of the 68 students interviewed, 38.2 percent (n= 26) were
from primary schools across Jamaica compared to 30.9 percent (n = 21) of all-all schools
and 30.9 percent (n = 21) from preparatory schools.
Table 6.1.1: Age Profile of Respondent
Details Frequency (n = ) Percentage
(in years)
14 11 16.2 15 55 80.9 16 2 2.9Mean age 14.87 yearsStandard deviation 0.42 yrs.
Based on Table 6.1.1 (above), the majority of the sampled population (80.9 %) was 15
year-old, compared to 2.9 percent and 16.2 percent of ages 16 and 14 years respectively.
From the preponderance of 15 year olds, in this sample, the findings of this study are
primarily based on this age cohort’s responses.
169
Table 6.1.2: Examination scores
Details Pre-Test I Post-Test II
% %
Mean 49.22 70.68
Median 47.50 67.50
Mode 56.00 67.00
Standard deviation 16.165 14.801
Skewness
Minimum
Maximum
0.004
21.00
82.00
-0.119
41.00
98.00
In respect to Examination Scores, on Test I, the average score was 49.22 percent ±
16.165 percent (i.e. standard deviation), with a median of 47.5 percent and a minimum
score of 21.0 percent and a maximum score of 82.00 percent (See Table 6.1.2), with the
most frequent score being 56.0 percent. The Examination Scores of Test II were higher
as the average score of 70.68 percent ± 14.801 (i.e. standard deviation), with a median
score of 67.5 percent and minimum and maximum score of 41.0 percent and 98.0 percent
respectively. The most frequently occurred score was 67.0 percent; with the Test II
skewness being negative 0.119 compared to Test I of 0.004 percentage-point. (See
Figures 6.1.3 & 6.1.4, below)
170
Figure 6.1.3: Skewness of Examination I (i.e. Test I)
The sampled population Mathematics test scores on Test I showed a marginally
positively skewness of 0.004. The standard deviation of 16.17 squared percentage points
indicate that generally the students’ scores are relatively dispersed compared to Test II.
Figure 6.1.4: Skewness of Examination II (i.e. Test II)
Based on Figure 6.1.4, the Test I’s scores are marginally skewed with a standard
deviation of 14.80 percentage points. Generally, the individual scores are relatively well
dispersed.
171
BEFORE INTERVENTION
Undecided32%
Disagree53%
Strongly disagree
15%
Undecided Disagree Strongly disagree
Figure 6.1.5: Perception of Ability
Of the sampled population (n = 68), in respect to student’s perception of their ability,
32.0 percent (n = 22) indicated that they were undecided about their ability in
Mathematics compared to 53 percent (n=36) who said their ability was poor and 15
percent (n = 10) who reported that their ability was very poor. (See, Figure 6.1.5).
Generally, students had a low perception of their ability to apply themselves in
successfully problem-solving mathematical questions as needed by their teachers.
0
5
10
15
20
25
30
35
40
45
50
strongly agree agree undecided
Figure 6.1.6: Self-perception
172
Figure 6.1.6 indicated that prior to the Mathematics intervention mechanism, generally,
students self-perception was extremely good (strongly agree, approximately 68 %) and
good (agree, 29 %) compared to approximately 3 percent (n = 2) who were undecided
none who had a low self-perception within the context of Mathematics.
0
10
20
30
40
50
60
strongly agree agree undecided
Figure 6.1.7: Perception of Task
From Figure 6.1.7, 77.9 percent (n = 53) of the respondents were ‘undecided’ in regard
to the ‘perception of task’. On the other hand, some 22.1 percent of the sampled
population were cognizant of their task assignment, of which approximately 3 percent
(n= 2) reported that knew exactly what are required of them in Mathematics.
173
0
5
10
15
20
25
30
35
40
45
50
agree undecided Disagree Stronglydisagree
Figure 6.1.8: Perception of Utility
Of the sampled population of 68 students, only 1.4 percent (n=1) reported that
Mathematics is relevant in their general life compared to 86.7 percent (n=59) who
believed that the subject is not relevant to general work and some 12 percent (n=8) who
were not sure (‘undecided’).
0
5
10
15
20
25
30
35
40
45
50
stronglyagree
agree undecided Disagree Stronglydisagree
Figure 6.1.9: Class environment influence on performance
Prior to the introduction of the intervention mechanism, approximately 94 percent (n=64)
of the respondents believed that an interactive class environment can influence their
performance in the subject compared to 4.4 percent (n=3) who reported that this approach
did not make a difference in the learning of Mathematics.
174
AFTER INTERVENTION
0
10
20
30
40
50
60
strongly agree agree undecided Disagree
Figure 6.1.10: Perception of Ability
On completion of the teaching intervention, of the sampled population (n = 68), 76.0
percent (n = 51) indicated that they were undecided about their ability in Mathematics
compared to 16.17 percent (n=11) who said their ability was good and 3 percent (n = 2)
who reported that their ability was very good, compared to 4.4 percent (n=3) who rated
themselves within a poor perspective. (See, Figure 6.1.10). Generally, most of the
students change the ratings of themselves from varying degrees of poor to undecided.
This perceptual transformation is a gradual change in a higher awareness of their ability
to problem-solve mathematical questions.
175
0
5
10
15
20
25
30
35
40
45
agree undecided Disagree Stronglydisagree
Figure 6.1.11: Self-perception
Based on Figure 6.1.11, predominantly (61.8%, n=42) the students disagreed with view
that attending Mathematics classes are a waste of time and ‘attending making them
nervous’ compared to 1.5 percent who reported that they felt it was a waste of time and
that they were nervous before attending Mathematics sessions.
0
5
10
15
20
25
30
35
40
strongly agree agree undecided
Figure 6.1.12: Self-perception
Approximately 59 percent (n=40) of the students reported that they were very confident
in themselves with 38.7 percent (n=27) indicated that they were just confident compared
to 1.5 percent (n=1) who reported that they were undecided and none suggested low self-
perception after the intervention. (See, Figure 6.1.12)
176
0
5
10
15
20
25
30
35
40
45
50
undecided Disagree Strongly disagree
Figure 6.1.13: Perception of Task
Generally, (See, Figure 6.1.13), 72.1 percent (n = 49) of the respondents reported that
they were unsure of the mathematical task to be performed compared to 20.6 percent
(n=14) who indicated that they were ‘undecided’ in regard to the ‘perception of task’.
0
5
10
15
20
25
30
35
40
45
50
agree undecided Disagree Stronglydisagree
Figure 6.1.14: Perception of Utility
Predominantly the students did not see the usefulness of Mathematics to their general
environment (86.8 percent, n = 51). Of the 51 respondents who were not able to foresee
the uses of Mathematics outside of the actual subject, 16.7 percent (n=11) reported that
Mathematics is absolutely irrelevant to their general world compared to 70.6 percent
(n=40) who believed that the subject is not relevant, with 10.7 percent (n =7) who were
unsure and some 2.9 percent (n=8) who reported a relevance of the subject matter to other
areas of their lives (See, Figure 6.1.14).
177
0
5
10
15
20
25
30
35
40
45
strongly agree agree undecided
Figure 6.1.15: Class environment influence on performance
On completion of the intervention exercise, 94.1 percent (n=64) of the respondents
reported that involvement in class and the general integrated class environment
influenced their performance in the discipline compared to 5.9 percent (n=4) who were
undecided, in comparison to none who reported that the general class environment
affected their performance in Mathematics. (See, Figure 6.1.15, above)
178
CROSS-TABULATIONS
Table 6.1.3(a): Class distribution by gender
GENDER Total
Male Female
CLASS 9(1)
16 (55.2%) 20 (51.3%) 36 (52.9%)
9(2)
13 (44.8%) 19(48.7%) 32 (47.1%)
Total
29 39 68
Of 68 students of this sample, 57.4 percent (n=39) were females compared to 42.6
percent (n=29) males. Of the 42.6 percent of the male respondents, 55.2 percent (n=16)
were in class one and 44.8 percent (n=13) in class two compared to 51.3 percent (n=20)
of females in class one and 48.7 percent (n=19) in class two (See, Table 6.1.3(a)).
179
Table 6.1.3(b): Class distribution by age cohorts
AGE Total
14 15 16 CLASS Experimental 8 27 1 36 72.7% 49.1% 50.0% 100.0% Controlled 3 28 1 32 27.34% 50.9% 50.0% 100.0%
Total 11 55 2 68
Approximately 53 percent (n=36) of the sampled population were in the experimental
group in comparison to some 47 percent (n=32) who were within the controlled group.
Approximately 81 percent (n=55) of the respondents were 15 years old, of which 50.9
percent (n=28) were in class two (i.e. the controlled group) compared to 49.1 percent who
were in class two (i.e. the experimental group). (See, Table 6.1.3(b)).
180
Table 6.1.3(c): Pre-test Score by typology of group
GROUP TYPE Total
experimental
group control group RETEST_1 Below 40 % 8 13 21 22.2% 40.6% 30.9% 41 - 59 % 20 10 30 55.6% 31.3% 44.1% 60 - 70 % 4 6 10 11.1% 18.8% 14.7% 71 - 80 % 3 3 6 8.3% 9.4% 8.8% Above 80 % 1 0 1 2.8% .0% 1.5%
Total 36 32 68
Table 6.1.3(d): Post-test Score by typology of group
GROUP TYPE Totalexperimental
group control group RETEST_2 41 - 59 % 5 16 21 13.9% 50.0% 30.9% 60 - 70 % 8 7 15 22.2% 21.9% 22.1% 71 - 80 % 7 5 12 19.4% 15.6% 17.6% Above 80 % 16 4 20 44.4% 12.5% 29.4%
Total 36 32 68
The results reported in Tables 4.1.3 (c) and (d) revealed that prior to the intervention
(pre-test – See, Table 6.1.3 c), 30.9 percent (n=30) of the respondents got grades ranging
from 0 to less than 40 percent, of which 40.6 percent (n=13) were within the controlled
group compared to 22.2 percent (n=8) were in the experimental group. Approximately 2
percent (n=1) of the sampled population got scores in excess of 80 percent, and the
181
person was from the experimental group. On the other hand, after the student-centred
learning approach technique was used by the teacher (post-test scores), none of the
students got scores which were lower than 40 percent. (See, Table 6.1.3d). Based on
Table 6.1.3(d), 29.4 percent (n=20) of the students got grades higher than 80 %, which
represents a 1350 percent increase over Test 1. This was not the only improvement as
scores on Test II increased in all categories except scores between 41 and 59 percent (i.e.
this was a decline of 100 %). On a point of emphasis, on Test II over Test I, more
students within the experimental group was observed excess in scores of 41 to 59%. In
addition, after the intervention, 44.4 percent (n=16) of the students within the
experimental category (n=36) scores marks higher than 80% compared to only 2.8
percent before the implementation of the intervention strategy by the teacher.
182
PAIRED-SAMPLE t TEST:
Table 6.1.4: Comparison of Examination I and Examination II
Details N Correlation Paired Difference
0.194Mean Std. de S.E t
Test I 68 49.22
-21.46 19.681 2.387 -8.990Test II 68 70.68
Significant (2-tailed) = 0.000
From Table 6.1.3, the paired-sample t test analysis indicates that for the 68 respondents,
the mean score on Test II (M = 70.68 %) was significant greater at the ρ value of 0.01
level (note: ρ value = 0.000) than average score on the first test (M= 49.22%). These
results also indicate that a positive correlation exist between the two test scores (r =
0.194) representing that those who score high on one of the test tend to score high on the
next test.
183
INDEPENDENT-SAMPLE t TEST
Table 6.1.5: Comparison across the Group by Tests
Details N Mean St. Deviation Levine’s Test t-test forEquality of
mean
Test I:Exper groupControl group
3632
50.3148.00
15.1317.42
F Sig2.55 0.115
Sig (2-tailed)0.5610.564
Test 2:Exper groupControl group
3632
76.8163.78
13.4813.25
0.013 0.909 0.0000.000
The independent-sample t test analysis (See, Table 6.1.4) indicates that 36 individuals in
the experimental group scored an average of 50.31 percent in the class, the 32 persons
within the controlled group had a mean score of 48.0 percent, and the mean difference
did not differ significantly at the ρ value of 0.05 (note: ρ value = 0.561). The Levene’s
test for Equality of Variance indicates for the experimental and the controlled groups do
not differ significantly from each other (note: p=0.115. On the other hand, in respect to
typology of groups and second test scores, the mean score for the experimental group was
76.8 percent (n=36) compared to 63.78 percent (n=32) for the controlled group, and that
means did differ significantly at the ρ value of 0.05 level (note: p=0.000). The Levene’s
test for Equality of Variance indicates for the experimental and the controlled group did
not statistical differ (note: ρ value = 0.909). Based on Table 6.1.4, the students who
were in the experimental group having been introduced to the student-centred learning
approach increased their grade score in Mathematics by approximately 53.0 percent
compared to the controlled group whose performance improved by 32.9 percent.
184
FACTORS AND THEIR INFLUENCE ON PERFORMANCE
Table 6.1.6: Analysis of Factors influence on Test II Scores
examssc2 Sum of Squares df Mean Square F Sig.
Between Groups 318.025 1 318.025 1.462 .231Within Groups 14358.857 66 217.558Total 14676.882 67
Of the sampled population (n=68), for the bivariate analysis of factors on Test II scores,
the mean scores between the groups was statistical not significant, ρ value more than 0.05
(note: Ρ value = 0.23136). Based on Table 6.1.6, the factors identified in this study are
not statistically explaining variation in performance of students on Test II.
Table 6.1.7: Cross-tabulation of Test II scores and Factors
Refac_2 Total
strongly agree agreeretest_2 41 - 59 % 19 (30.2%) 2 (40.0%) 21 (30.9%)
60 - 70 % 12 (19.0%) 3 (60.0%) 15 (22.1%)
71 - 80 % 12 (19.0%) 0 (0.0%) 12 (17.6%)
Above 80 % 20 (31.7%) 0 (0.0%) 20 (29.4%)
Total 63 5 68
χ2 (3) = 6.207, ρ value = 0.102
Table 4.1.7, further analyses the Test II scores from the perspective that identified factors
influences students’ performance and statistically this was not significant (χ2 (3) = 6.207,
Ρ value = 0.102). Despite the fact that entire sampled population (100%, n=68) either
36 The following are reasons why the parameter estimate is not significant – (1) inadequate sample size; (2) type II error, (3) specification error, and (4) restricted variance in the independent variable(s).
185
strongly agreed or agreed to the questions on factors, these were not statistically found to
contributory factor that influences the change in academic performance. It should be
noted that this be a Type II error. In that, the ideal sample size for cross tabulation is in
excess of 200 cases with a stipulated minimum of more than 5 responses to a cell, this
prerequisite was not the case as the sample size for this study was 68 students. Therefore,
the fact that there is not statistical relationship between the examined variables may be as
a result of a Type II error (i.e. meaning, statistically indicating that no relationship exist
between the factors but in reality a relationship does exists, and the primary reason is due
to the relatively small sample size).
Table 6.1.8: Bivariate relationship between Student’s Factors and Test II scores
Test II Scores Other Total
No Yesretest_2 41 - 59 % 15 6 21 29.4% 35.3% 30.9% 60 - 70 % 9 6 15 17.6% 35.3% 22.1% 71 - 80 % 10 2 12 19.6% 11.8% 17.6% Above 80 % 17 3 20 33.3% 17.6% 29.4%
Total 51 17 68
χ2 (3) = 3.454, ρ value = 0.327
Students did note that a number of factors contribute to their low academic performance
in Mathematics, to which the researcher sought to unearth any merit to this perception.
Based on Table 6.1.8, there is not statistical association between the identified factors
noted by students and academic performance. (χ2 (3) = 3.454, ρ value = 0.327) Hence,
collectively, issues such as lighting, resources, and noise and communication barriers
186
were not statistically responsible for improvements in students’ test scores on the second
Mathematics examination. Even when the identified factors were disaggregated, none of
them was found to contribute to the increased Test II scores (i.e. light: χ2 (3) = 1.298, ρ
value = 0.730; communication barriers: χ2 (3) = 2.330, ρvalue = 0.5.07; resources χ2 (3) =
2.126, ρ value = 0.547 and noise: χ2 (3) = 1.169, ρ value = .760). It should be noted that
this is a Type II error (See Appendix 2). In that, the ideal sample size for cross tabulation
is in excess of 200 cases with a stipulated minimum of more than 5 responses to a cell,
this prerequisite was not the case as the sample size for this study was 68 students.
Therefore, the fact that there is not statistical relationship between the examined variables
may be as a result of a Type II error (i.e. meaning, statistically indicating that no
relationship exist between the factors but in reality a relationship does exists, and the
primary reason is due to the relatively small sample size).
187
CHAPTER 7
Hypothesis 4:
General hypothesis –
Ho: There is no statistical relationship between expenditure on social programmes (public
expenditure on education and health) and levels of development in a country; and
H1: There is a statistical association between expenditure on social programmes (i.e.
public expenditure on education and health) and levels of development in a country
ANALYSES AND INTERPRETATION OF DATA
Univariate Analyses
Table 7.1.1: Descriptive Statistics - Total Expenditure on Public Health (as percentage of GNP HRD, 1994)
TOTAL EXPENDITURE on PUBLIC HEALTH as percentage of GNP (HRD, 1994)
Mean 4.6140
Standard deviation 2.1489
Skewness 0.9860
Minimum 0.8000
Maximum 13.3000
From table 7.1.1, the data is trending towards normalcy, as the skewness is 0.9860 and so
the distribution is relatively a good statistical measure of the sampled population (see
188
figure 1.2 below). A mean of 4.614 shows that approximately 4.614 per cent of the Gross
National Production (GNP) is spent on public health ± 2.1489, with a maximum of 13.3%
0.0 2.0 4.0 6.0 8.0 10.0 12.0 14.0
1990: TOTAL EXPENDITURE ON HEALTH AS PERCENTAGE OF GDP (HDR 1994)
0
5
10
15
20
25
Fre
qu
en
cy
Mean = 4.614Std. Dev. = 2.1489N = 145
1990: TOTAL EXPENDITURE ON HEALTH AS PERCENTAGE OF GDP (HDR 1994)
Figure 7.1.1: Frequency distribution of total expenditure on health as % of GDP
189
Table 7.1.2: Descriptive statistics of Expenditure on Public Education (as percentage of GNP, HRD, 1994)
PUBLIC EXPENDITURE on PUBLIC EDUCATION as percentage of GNP (HRD, 1994)
Mean 4.5340
Standard deviation 1.9058
Skewness 0.1340
Minimum 0.0000
Maximum 10.600
It can be concluded from the data collected and presented in the table above that the data
is relatively normally distributed (see Figure 4.2 – skewness is 0.134) and therefore is a
good measure of the sample population. The mean amount of public expenditure on
public education as a percentage of GNP is 4.534 ± 1.91. This indicates that on an
average that approximately of 4.534 per cent of the Gross National Production (GNP) is
spent on public education.
Figure 4.2:
190
0.0 2.0 4.0 6.0 8.0 10.0 12.0
PUBLIC EXPENDITURE ON EDUCATION AS PERCENTAGE OF GNP (HDR 1994)
0
5
10
15
20
Fre
qu
en
cy
Mean = 4.534Std. Dev. = 1.9058N = 115
PUBLIC EXPENDITURE ON EDUCATION AS PERCENTAGE OF GNP (HDR 1994)
Figure 7.1.2: Frequency distribution of total expenditure on education as % of GNP
191
Table 7.1.3: Descriptive statistics of Human Development (proxy for development)HUMAN DEVELOPMENT INDEX
Mean 2.0700
Standard deviation 0.7820
Skewness -0.1180
Minimum 1.000
Maximum 3.000
Based on Table 7.1.3 above, the average human development index reads 2.07 ± 0.78,
with a negligible skewness of – 0.118. The table shows that the maximum value for
human development is 3 with a minimum of 1.
192
0.5 1 1.5 2 2.5 3 3.5
1993: HUMAN DEVELOPMENT INDEX IN THREE CATEGORIES: 1 = LOW HUMAN DEVELOPMENT, 2 = MEDIUM HUMAN
DEVELOPMENT, 3 = HIGH HUM
0
20
40
60
80
100
Fre
qu
ency
Mean = 2.07Std. Dev. = 0.782N = 165
1993: HUMAN DEVELOPMENT INDEX IN THREE CATEGORIES: 1 = LOW HUMAN DEVELOPMENT, 2 = MEDIUM
HUMAN DEVELOPMENT, 3 = HIGH HUM
Figure 7.1.3: Frequency distribution of the Human Development Index
193
In seeking with the attempt of making this text simple and extensive, I will not only
provide an analysis of the generated output from a Pearson statistical test but will
illustrate how this should be executed in SPSS. Before we are able to begin the
process, let us remind ourselves of the hypothesis:
194
H1: There is a statistical association between expenditure on social programmes (i.e. public
expenditure on education and health) and levels of development in a country (dependent variable –
HDI, which measures levels of development; and independent variables – public expenditure on
education, public expenditure on health care).
Figure 7.1.4: Running SPSS for social expenditure on social programme
step 1: select analyze
195
Figure 7.1.5: Running bivariate correlation for social expenditure on social programme
Step 2: Select correlate, then bivariate
196
Figure 7.1.6: Running bivariate correlation for social expenditure on social programme
This result from step 2
197
Step 3: Select the dependent and the independent variables
198
You would have accomplished a lot from just generating the tables, but the most important aspect is not in the production of the tables but it the analysis of the hypothesis. Hence, I will analyze the results, below.
Step 4: Select paste then ‘run’ or ok, which then give, Output
199
PEARSON’S MOMENT CORRELATION: BIVARIATE ANALYSIS37
Table 7.1.4: Bivariate relationships between dependent and independent variables
PUBLIC EXPENDITURE
ON EDUCATION
AS PERCENTAGE OF GNP (HDR
1994)
HUMAN DEVELOPMENT
INDEX: 0 = LOWEST HUMAN
DEVELOPMENT, 1 = HIGHEST
HUMAN DEVELOPMENT
(HDR, 1997)
1990: TOTAL EXPENDITURE ON HEALTH AS PERCENTAGE OF GDP (HDR
1994)
PUBLIC EXPENDITURE ON EDUCATION AS PERCENTAGE OF GNP (HDR 1994)
Pearson Correlation
1 .413(**) .435(**)
Sig. (2-tailed)
. .000 .000
N 115 114 106
HUMAN DEVELOPMENT INDEX: 0 = LOWEST HUMAN DEVELOPMENT, 1 = HIGHEST HUMAN DEVELOPMENT (HDR, 1997)
Pearson Correlation
.413(**)1 .395(**)
Sig. (2-tailed)
.000 . .000
N 114 165 142
1990: TOTAL EXPENDITURE ON HEALTH AS PERCENTAGE OF GDP (HDR 1994)
Pearson Correlation
.435(**).395(**)
1
Sig. (2-tailed)
.000 .000 .
N 106 142 145
** Correlation is significant at the 0.01 level (2-tailed).
37 See Appendix IV
200
Bivariate relationship between public expenditure on education and human development
From Table 7.1.4, the results indicated that there was a statistical relationship between
public expenditure on education as a percentage of GNP and levels of human
development based on the population sampled. The strength of the relationship is
moderate (cc = 0.413 or 41.3 %) and this indicated that there is a positive relationship
public expenditure on education as a percentage of GNP and human development.
The coefficient of determination indicates that public expenditure on education as
a percentage of GNP explains approximately 17.06 percent of the variation in levels of
human development of the population sampled. A significant portion of the countries
surveyed (82.94%) is not explained in terms of its expenditure on education.
Bivariate relationship between total expenditure on health and human development
From Table 1.4, the results indicate that there is a statistical relationship between total
expenditure on health as a percentage of GDP and levels of human development. The
strength of the relationship is moderate which shows that there is a positive relationship
total expenditure on health as a percentage of GDP and human development. The
coefficient of determination indicates that total expenditure on health as a percentage of
GNP explains approximately 15.68 per cent of the proportion of variation in levels of
human development of the population sampled. The unexplained variation of 84.32%
which indicates that although total expenditure on health explains a particular percent of
the variation in development, a significantly larger percent of that variation is not
explained by total expenditure on health.
201
TABLE 7.1.5: SUMMARY OF HYPOTHESES ANALYSIS
VARIABLES COUNT (Ρ value )
Rejected Null Hypotheses (i.e. rejected Ho):
TOTAL EXPENDITURE ON HEALTH AND HUMAN DEVELOPMENT 114 (0.001)
PUBLIC EXPENDITURE ON HEALTH AND HUMAN DEVELOPMENT 142 (0.001)
202
CHAPTER 8
Hypothesis 5:
GENERAL HYPOTHESIS:
The health care seeking behaviour of Jamaicans is a function of educational level,
poverty, union status, illnesses, duration of illnesses, gender, per capita consumption,
ownership of health insurance policy, and injuries. [ Health Care Seeking Behaviour =
f( educational levels, poverty, union status, illnesses, duration of illnesses, gender, per
capita consumption, ownership of health insurance policy, injuries)]
DATA INTERPRETATIONS
SOCIO-DEMOGRAPHIC INFORMATION
Table8.1.1: AGE PROFILE OF RESPONDENTS (N = 16,619) Particulars Years
Mean 39.740
Standard deviation 19.052
Skewness 0.717
From table 1 above, the skewness of 0.717 shows that there is a clear indication that the
data set is not normal, and so the researcher logged this variable in order to reduce the
203
skewness so that the value will be a relative good statistical measure for the sampled
population (n=16,619 respondents). The mean age of the sampled population is 39 years
and 9 months (39.740 years). Of the population sampled, the minimum age was 15 years
and the maximum age was 99 years. The standard deviation (of 19.052) shows a wide
spread from the mean of the scatter values of the sampled distribution.
Table 8.1.2: LOGGED AGE PROFILE OF RESPONDENTS (N = 16,619) Particulars Years
Mean 3.5983
Standard deviation 0.47047
Skewness 0.014
Kurtosis -1.014
From table 8.1.2 above, after the variable was logged (age), the skewness was 0.014
which shows minimal skewness that is a better relative statistical measure for the
sampled population (n=16,619 respondents). The sampled population has a mean age of 3
years and 7 months (3.5983 years) with a standard deviation of 0.47047 that shows a
narrow spread from the mean of the scatter values of the sampled distribution.
Table 8.1.3: HOUSEHOLD SIZE (ALL INDIVIDUALS) OF RESPONDENTSParticular Individuals
Mean 4.741
Median 4.000
Standard deviation 2.914
Skewness 1.503
204
The findings from the sampled population of the Survey of Living Condition (SLC 2002)
in table 1 above shows a skewness of 1.503 that is an unambiguous indication that the
data set is not close normal and so is not a relative good statistical measure of the
measure of central tendency of this population sampled (n=16,619 respondents).
Therefore, the researchers use the median, as this is a better measure of central tendency.
The median number of individuals within the sampled population is four persons. Of the
population sampled, the minimum number of individuals with a household was one
person and the maximum was 23 people. The standard deviation (of 2.914) shows a
relatively close spread from the median of the scatter values of the sampled distribution.
Of the sampled population (n=16,619 people beyond and including 15 years),
there were 8,078 males (i.e. 48.6 %) and 8,541 females (i.e. 51.4%). Furthermore, 92.1
percent (n=13,339) of the sampled respondents had secondary education and lower [see
Table 8.1.] compared with 7.9 percent (n=1142) at the tertiary level. The valid response
rate in regards to type of education was 87.1 percent (that is, of the sampled population of
sixteen thousand, six hundred and nineteen people). In addition, 14,009 cases were
included in the analysis (or 84.3 percent) with 2,610 missing cases (or 15.7 percent).
Table 8.1.4: UNION STATUS OF THE SAMPLED POPULATION (N=16,619)Particular Frequency Percent
Married 3,907 25.4
Common law 2,608 16.4
Visiting 2,029 12.7
Single 5,638 35.4
None 1,757 11.0
Total 15,939 100.0
205
Based on the findings of this survey, of the sampled population (n =16,619), the valid
response rate to union status was 95 percent. The survey showed that 35.4 percent (n =
5,638) of the sample was single, 25.4 percent (n = 3,907) was married, 16.4 percent (n =
2,608) was in common law union and 11.0 percent (n = 1,757) of the same sample was in
no union. Union status was further classified into two (2) main groups; firstly, living
together and secondly, not living together. Collectively, 51.9 percent of the respondents
(n = 8,272) were not living together and 48.1 percent (n = 7,667) were living together.
Comparatively, the response rate was 95.9 percent (n = 15,939) to none response rate of
4.1 percent (n = 680).
Table 8.1.5: OTHER UNIVARIATE VARIABLES OF THE EXPLANATORY MODEL
Particular Frequency Percent
Gender Male 8078 48.6 Female 8541 51.4
Dummy educational LevelPrimary 7294 50.4Secondary 6045 41.7Tertiary 1142 7.9
Health InsuranceYes 1919 11.8No 14292 88.2
Dummy union StatusWith a partner 8544 53.6Without a partner 7395 46.4
PovertyPoor 5844 35.2Middle 6762 40.7Rich 4013 24.1
206
From Table 8.1.5, of the sampled population (n=16,619), 51.4 percent (N=8541)
were females compared with 48.6 percent (N=8078) males. The findings revealed that
were 35.2 percent (5844) poor people compared with 40.7 percent (N=6762) within the
middle class with 24.1 percent (N=4013) of the sample in the upper (rich) categorization.
With regard to the union status of the sampled group, 53.6 percent (N=8544) had a
partner compared with 46.4 percent (7395) who did not have a partner. Furthermore, the
educational level of the respondents was 50.4 percent (N=7294) in primary category with
41.7 percent (N=6045) in the secondary grouping compared with 7.9 percent (N=1142) in
the tertiary categorization. With respect to the issue of availability of health insurance,
the findings revealed that 88.2 percent (14,292) of the sampled population did not possess
this medium compared with 11.8 percent (1919) that had access.
Table 8.1.6: VARIABLES IN THE LOGISTIC EQUATIONParticular β S.E Wald df Significant Exp (β)Illnesses 2.336 .075 969.894 1 .000 10.338Injuries .863 .181 22.655 1 .000 2.370Poverty 45.938 2 .000Poverty 1 .127 .056 5.128 1 .024 1.135Poverty 2 .332 .050 44.601 1 .000 1.394Per capita consumption
.094 .030 10.117 1 .001 1.099
Union status -.169 .040 18.024 1 .000 0.845Gender .793 .039 418.533 1 .000 2.2210Health insurance .664 .064 106.383 1 .000 1.942Age .022 .001 359.375 1 .000 1.022Levels of education
.274 .085 10.332 1 .001 1.315
Constant - 3.024 .319 89.691 1 .000 0.049
Note: If the ρ value ≤ 0.05, then this indicates that the corresponding variable is
significantly associated with changes in the baseline odds of not seeking health care.
207
Based on table 8.1.6, illnesses contributes the most (i.e. Exp (β) =10.338) to
health seeking behaviour. The relationship between illnesses and health seeking
behaviour is significant (Ρ value = 0.000 ≤0.05). Furthermore, positive β values of 2.336
as it relates to illnesses indicate that as people move from no illnesses to illnesses, they
will seek more health care. Given that, the logit is positive for illnesses, so we know that
being ill increases the odds of seeking health care.
The value in table 4 in regards to injuries is not surprising as is inferred from the
literature. This variable second ranked (injuries) in contributing to health seeking
behaviour (i.e. Exp (β) = 2.370) for individuals, ages 15 to 99 years. Furthermore, a
positive β value of 0.863 indicates that with the increasing number of injuries, the
sampled population sought more health care (or health seeking behaviour increases).
With the Ρ value = 0.001 ≤ 0.05, the logit is positive for injuries, and this suggests that
being injured increases the odds of seeking health care.
As also indicated in table 4, there is a significant relationship between gender and
health seeking behaviour (ρ value = 0.000 ≤0.05). Based on the Exp (β) of 2.210, gender
is the third largest contributor to the health seeking behaviour. In addition, a positive β
value of 0.793 indicates that females sought more health care in comparison to males.
Further, a positive logit in relation to gender suggests that being female increases the
odds of seeking health care.
The findings in table 8.1.6 concur with the literature as it spoke to a positive
relationship between possessing health insurance and individual seeking health (ρvalue =
0.000 ≤0.05). Herein, health policy contributes the fourth most to the model of health
seeking behaviour (Exp (β) of 1.942). The positive β (of 0.664) suggests that an
208
individual who holds a health policy is more likely to seek health care in contrast to no-
health policyholders. In addition, this positive logit of the sampled population infers that
having a health insurance increases the odds of seeking health care.
The literature review spoke to a direct relationship between moving from lower
education to higher education and health seeking behaviour (β of 0.274, ρ value = 0.000
≤0.05). The positive β reinforced the literature that health seekers are more of a higher
educational type. Further, a positive logit in relation to levels of education suggests that
being within a higher education type increases the odds of seeking health care.
In respect to ages of the respondents (15 years ≤ ages ≥99 years), there is a
statistical significant relationship between the older one gets and an increase in his/her
health seeking behaviour (ρ value = 0.000 ≤0.05). This means that for each additional
year that is added to ones life, he/she seeks additional health care. Furthermore, positive
logit (based on table 4) suggests that as age increase by each additional year, the odds of
seeking health care increases.
The information presented in table 4 with regard union status indicates that people
who had partner are more likely to seek health care compared with those who do not β (of
-0.169) and a ρ value of 0.000 ≤0.05. The reality was that union status contributes the
least to the health seeking behaviour (or the model). With a negative logit (from table 4)
in regards to union status, this suggests that as union status decrease from living to not
living together, the odds of seeking health care decreases.
The per capita consumption of the sampled population clearly indicates that a
direct significant relationship exists between this variable and dependent variable (health
seeking behaviour, ρ value of 0.001 ≤0.05). The Exp (β) of 1.099 values determines that
209
per capita consumption contributes the third least to the model. Furthermore, the positive
β indicates that as per capita consumption increases by one additional dollar, health-
seeking behaviour increases. Given that, the logit is positive we know that increases in
per capita consumption increases the odds of seeking health care.
Table 8.1.7: CLASSIFICATION TABLE
Predicted
Health seeking behaviour
PercentageCorrect
Observed No Yes No 6,452 1.191 84.4 Yes 3,008 3,358 52.7Overall percentage 70.0
The literature review perspective was that there were relationships between the
dependent and the independent variables, the findings of this survey unanimously support
those positions. This means that there were statistical significant relationships between
each hypothesis (i.e. ρvalue ≤ 0.05). The variables tested in the model all predict the
health seeking behaviour of Jamaicans (of ages 15 to 99 years) but to varied degree (Exp
(β). From the model predictor; illnesses, injuries and gender offered the strongest
influence. This, therefore, means that people generally tend to seek health care when they
are ill or injured and of a particular gender (female). Based on table 5 above, the model
correctly predicts 52.7 percent of people in the sample will seek health care. However,
the model correctly predicts that 84.4 percent of the will not seek health care. In respect
to the overall predictor of the model, 70.0 percent is correctly predicted from the variable
chosen of the sample size. The Nagelkerke R square of .284 indicates that, 28.4 percent
of the variation in health care seeking behaviour of Jamaicans of ages 15 to 99 years is
explained by the nine variables in the model.
210
CHAPTER 9
Hypothesis 6:
GENERAL HYPOTHESIS
There is a negative correlation between access to tertiary level education and poverty
controlled for sex, age, area of residence, household size, and educational level of parents
(see Appendix III)
211
ANALYSES AND INTERPRETATION OF DATA
Table 9.1.1: UNIVARIATE ANALYSESVariables Frequency (Percent)
Educational LevelNo formal schooling 118 (0.8)Primary education 6956 (48.1)Secondary education 231 (43.1)Tertiary education 1142 (7.9)AgeMean 40.5 yrs Standard deviation 18.839
Skewness 0.713Jamaica’s Pop. QuintilePoor 5629 (34.97)Lower Middle Class 3146 (19.5)Upper Middle Class 3400 (21.1)Rich 3957 (24.5)Gender (Sex)Male 7822 (48.5)Female 8310 (51.5)
Geographic Locality of JamaicansKingston Metropolitan Area (KMA) 3397 (21.1)
Other Towns 3046 (18.9)Rural Areas 9689 (61.0)Union StatusMarried 3906(25.2)Common law 2607 (16.8)Visiting 2017 (13.0) Single 5368(34.6)None 1605 (10.4)Household SizeMean 4.7035Standard deviation 2.917Skewness 1.531Access to Tertiary EducationNo Access 16422 (89.4)Access 1943 (10.6)Poverty StatusNon-poor 10503(65.1)
212
Poor 5629 (34.9)
1 The index on access to tertiary level education begins with a of 0.00 to a high of 1.0
Of the sampled population of 16,123 respondents, there are 48.5 percent (n = 7,822)
males and 51.5 percent (n = 8310) females. This sample is a derivative of the general
sample of 25,007. From table 4(i), above, the incidence of poverty is 34.9 percent (n =
5,629). The findings reveal that 25.2 percent (n = 3906) of the sampled population are
married compared to 16.8 percent (n = 2,607) in cohabitant (i.e. common law)
relationship, with 13.0 percent (n = 2,017) in visiting unions, compared to 34.6 percent (n
= 53) in single relationships, with 10.4 percent (n= 1605) not indicating a union choice.
The average number of individuals per household is approximately five (4.7035 ±
2.917) with a standard deviation of approximately three persons. As results in Table 4 (i)
indicate, the household size variable has a skewness of 1.5 persons, indicating dispersion
away from normality. It is this finding that made the researcher logged the variable in
order to remove some degree of the skewness.
A preponderance of the sampled population is from the rural zones (i.e. 61.0 percent,
n = 9,689) compared to 21.1 percent (n = 3,397) who reside in Kingston Metropolitan
Areas, and 18.9 percent from Other Towns. The minimum age for the sampled group is
16 years with an averaged age of 40 years and a standard deviation of 19 years, (40 years
6 months = -18.839). The age variable has a positive skewness of 0.733 to which the
researcher logged (natural log) in order to reduce some degree of the variable’s skewness.
Despite a preponderance of sample being within the poor categorization (≈35
percent), only 7.9 percent (n=1142) of the sampled population (n=16132) has or is
pursuing a tertiary level education. In Table 4 (i), the findings reveal that people who
have had no formal schooling are less than 1 percent (0.8 percent, n = 118) compared to
213
approximately 48.1 percent (n = 6,956) of people who are pursuing or have not
completed primary level education whereas 43.1 percent (n = 6231) are at the secondary
level with the formal education system.
214
Table 9.1.2: FREQUENCY DISTRIBUTION OF EDUCATIONAL LEVEL BY
QUINTILE
EducationalLevel
Jamaica’s Population Quintile DistributionPoor Lower Middle Upper Middle Rich
Frequency (Percent)
No formal 73 (1.4) 12(0.4) 16 (0.5) 17 (0.5)
Primary 2,886 (55.9) 1,442(51.3) 1,393 (46.4) 1,235 (35.5)
Secondary 2,069 (40.1) 1,248 (44.4) 1,386 (46.2) 1,528 (44.0)
Tertiary 135 (2.6) 108 (3.8) 205 (6.8) 694 (20.0)Ρ value = 0.001, χ2 (9) = 1127.55, Lambda (i.e. λ) = .051
As indicated in Table 9.1.2, there was a statistical relationship between persons within the
population quintile and educational level (ρ value = .001 < 0/05, χ2 (9) = 1,127.55). A
lambda value of 0.051 indicates that there is a direct relationship between higher levels of
educational attainment and affluence. Table 9.1.1 showed that 2.6 percent of the poor has
access to tertiary level education compared to 20.0 percent of the rich, and 10.6 percent
of the middle class. Approximately 64 percent (64.28 %) less rich person have less than
primary school education compared to the poor (see Table 9.1.1, above). In the primary
level of education, the poor has more people in this categorization than the other
classification (i.e. lower middle/upper middle class and rich). With respect to secondary
level educational attainment, the poor have the least number of attendances in the social
class stratification (i.e. quintile distribution).
215
Table 9.1.3: FREQUENCY DISTRIBUTION OF JAMAICA’S POPULATION BY QUINTILE AND GENDER
Pop. Quintile
Gender of Respondents
MaleFrequency (%)
FemaleFrequency (%)
Poor 2606 (33.3) 3023 (36.4)
Lower Middle Class 1514 (19.4) 1632 (19.6)
Upper Middle Class 1643 (21.0) 1757 (21.1)
Rich 2059 (26.3) 1898 (22.8)
ρ value = 0.001, χ2 (3) = 30.957
When gender is cross tabulated with population quintile, 36.4 percent (n = 3023) of the
sampled population who are females are in the poor categorization compared to 33.3
percent males. In the affluence classification, 26.3 percent (n=2059) are males compared
to 22.8 (n=1898) being females. From the data (Table 9.1.3), irrespective of a person’s
gender, within the middle class groupings, population quintile distribution is the same.
This finding reveals that approximately 4 percent more males are richer than females
(22.8 %), compared to 3.1 percent more poor females than their male counterparts. It can
be safely deduced from the data that poverty is more a female issue (36.4 %) than a male
phenomenon (33.3%).
216
Table 9.1.4: FREQUENCY DISTRIBUTION OF EDUCATIONAL LEVEL BY QUINTILE
Union StatusJamaica’s Population Quintile Distribution
Poor Lower Middle Upper Middle RichFrequency (Percent)
Married 1213(22.5) 710 (23.4) 827 (25.3) 1156 (30.4)Common law 972(18.0) 550(18.1) 637 (19.57) 448 (11.8)Visiting 672 (12.4) 358(11.8) 406 (12.4) 581 (15.3)
Single 1905 (35.3) 1099 (36.2) 1102(33.7) 1262 (33.2)None 639(11.8) 319 (10.5) 2969(9.1) 351(9.2)Ρ value = 0.001, χ2 (12) = 187.77
Collectively, 30.4 percent (n=1156) of the sampled population who are affluent (i.e. rich)
indicate that they are married compared to 22.5 percent (n=1213) of those who are poor,
23.4 percent (n=710) of those in the lower middle class in comparison to 25.3 percent
(n=827) in the upper middle class. Approximately 12 percent (11.8 %) of the rich report
that they are in cohabitated relationship compared to 18 percent (n=972) in the poor
categorization, and 19.6 percent (n=637) in the upper middle class in contrast to 18.1
percent (n=550) of those in lower middle class. Within the categorization of the single
union status, the differences in each quintile are marginal (Table 9.1.4).
217
Table 9.1.5: FREQUENCY DISTRIBUTION OF POP. QUINTILE BY HOUSEHOLD SIZE
Household size
Jamaica’s Population Quintile Distribution
Frequency (%) Frequency (%) Frequency (%) Frequency (%)Poor Lower Middle Upper Middle Rich
1 229 (4.11) 149 (4.7) 304 (8.9) 838(21.2)
2 427(7.6) 354(11.3) 507(14.9) 977(24.7)
3 567(10.1) 466(14.8) 614(18.1) 822(20.8)
4 702(12.5) 520(16.5) 631(18.6) 615(15.5)
5 863(15.3) 503(16.0) 499(14.7) 359(9.1)
6 764(13.6) 439(14.0) 311(9.1) 193(4.9)
7 650(11.5) 305(9.7) 260(7.6) 59(1.5)
8 516(9.27) 151(4.8) 133(3.9) 45(1.5)
9 282(5.0) 91(2.9) 36(1.1) 18(0.5)
10 171(3.0) 41(1.3) 44(1.3) 8(0.2)
11 106(1.9) 53(1.7) 26(0.8) 8(0.2)
12 114(2.0) 14(0.4) 9(0.3) 0(0)
13 84(1.5) 9(0.3) 0(0.0) 8(0.2)
14 53(0.9) 7(0.2) 16(0.5) 0(0.0)
15 12(0.2) 17(0.5) 0(0.0) 7(0.2)
16 26(0.50) 8(0.3) 0(0.0) 0(0.0)
17 17(50.0) 0(0.0) 10(0.3) 0(0.0)
18 7(0.1) 8(0.3) 0(00.0) 0(0.0)
19 7(0.1) 11(0.3) 11(0.3) 0(0.0)
21 26(0.5) 0(0.0) 0(0.0) 0(0.0)
23 13(0.2) 0(0.0) 0(0.0) 0(0.0)
Ρ value = 0.001, χ2 (60) = 3397.06
The findings in Table 9.1.5 reveal there is a statistical association between population
quintile and household size. Even more importantly, 21.2 percent (n=838) of the affluent
218
has a one member household compared to 8.9 percent (n=304) in the upper middle class
and 4.7 percent (n=149) of the poor. Comparatively, the rich do not have a 16-member
family household or more in comparison to poor, which have household ranging for one-
member to 23 members. Collectively the affluent family type has the majority of their
household size being between 1 to 4 members compared to the majority of the poor that
have household sizes from 4 to 7 members.
Table 9.1.6: BIVARIATE ANALYSIS OF ACCESS TO TERTIARY EDU. & POVERTY STATUS
Access to tertiary education
Poverty Status
Non-poorFrequency (%)
PoorFrequency (%)
No Access 8146 (83.3) 5116 (95.3)
Access 1631 (16.7) 254 (4.76)
ρvalue = 0.001, χ2 (1) = 454.432
The substantive issue of this study is ‘there a relationship between poverty status and
access to tertiary level education’ as indicated in Table 8.1.6, there is a statistical
association between poverty status and access to tertiary level education. Similarly, 95.3
percent (n=5116) of the poor indicate that they had no access to tertiary level education
compared to 8.3 percent (n=8146) of those who are non-poor (i.e. from lower middle
class to rich). Some 5 percent (4.76) of the poor reported that they had access to tertiary
level education in contrast to 16.7 percent for the non-poor. This finding indicates that a
preponderance ( 71.5%) of non-poor had access to tertiary education than the poor.
219
Table 9.1.7: BIVARIATE ANALYSIS OF ACCESS TO TERTIARY EDU. & GEOGRAPHIC LOCALITY OF RESIDENTSAccess to tertiary education
Geographic Locality of residents
KMA Other Towns Rural Areas
Frequency (%) Frequency (%) Frequency (%)No Access 2348 (76.1) 2446 (85.0) 8468 (92.2)
Access 738 (23.9) 430 (15.0) 717 (7.8)
Ρ value = 0.001, χ2 (2) = 570.550
The findings in Table 9.1.7 reveals that 92.2 percent (n=8468) of the residence of rural
areas do not have access to tertiary level education compared to 76.1 percent (n=2348) of
those who dwell in Kingston Metropolitan Areas and 85.0 percent (n=2446) of those who
live in Other Towns. However, 7.8 percent (n=717) of the sampled population who
reside in the rural areas have access to tertiary level education followed by 15 percent
(n=430) of those who reside in Other Towns have access to post-secondary education
compared to 23.9 percent (n=738) of those in Kingston Metropolitan area.
220
Table 9.1.8: BIVARIATE ANALYSIS OF GEOGRAPHIC LOCALITY OF RESIDENTS & POVERTY STATUS
Geographic Locale
Poverty Status
Non-poorFrequency (%)
PoorFrequency (%)
Kingston Metropolitan
Area(KMA)
2808 (26.7) 589 (17.3)
Other Towns 2139 (20.4) 907 (16.1)
Rural Areas 5556 (52.9) 4133 (73.4)
Ρ value = 0.001, χ2 (1) = 752.934
According to 73.4 percent (n=1433) of the poor, they live in rural areas in comparison to
52.9 percent (n=5556) of the non-poor. From Table 9.1.8), 17.3 percent of the poor live
in Kingston Metropolitan Area compared to 26.7 percent (n=2808) of the non-poor. On
the other hand, 20.4 percent (n=2139) of the middle, upper and rich classes live in Other
Towns as against the poor. The findings clearly show that poverty is substantially a
Rural Area phenomenon as against Other Towns or in urban zones. Statistically, there is
a significant association between poverty status and access to tertiary level education
(ρvalue = 0.001 < 0.05, χ2 (1) = 752.934).
221
Table 9.1.9: BIVARIATE RELATIONSHIP BETWEEN ACCESS TO TERTIARY LEVEL EDUCATION BY GENDER
Access to tertiary level ed.
Gender of Respondents
MaleFrequency (%)
FemaleFrequency (%)
No Access 6684 (90.2) 6578 (85.1)
Access 729 (9.8) 1156(14.9)
ρvalue = 0.001, χ2 (1) = 90.812
The findings in Table 9.1.9 reveal that there is a statistical association between
gender determining access to post-secondary level education (χ2 (1) = 90.812, ρ value =
0.001<0.05). The sampled population constitutes 90.2 percent (n=6684) males not
having access to tertiary level education in comparison to 85.1 percent (n=6578) of
females. Using the data in Table 4.7 (ii), approximately 34 percent more females are
accessing post-secondary level education than their male counterparts (i.e. 14.9 percent
female to 9.8 percent males).
222
Table 9.1.10: BIVARIATE RELATIONSHIP BETWEEN ACCESS TO TERTIARY LEVEL EDUCATION BY GENDER CONTROLLED FOR POVERTY STATUSPoverty Status Sex of individual Total male female 0 = Non-poor Access to tertiary
education0 = No access Count
4269 3877 8146
% within Sex of individual
86.7% 79.9% 83.3%
1 = Access Count 657 974 1631 % within Sex of
individual13.3% 20.1% 16.7%
Total Count 4926 4851 9777 1 = Poor Access to tertiary
education0 = No access Count
2415 2701 5116
% within Sex of individual
97.1% 93.7% 95.3%
1 = Access Count 72 182 254 % within Sex of
individual2.9% 6.3% 4.7%
Total Count 2487 2883 5370
Non-poor: Ρ value = 0.001, χ2 (1) = 79.905; Poor Ρ value = 0.001, χ2 (1) = 34.612
As indicated by Table 9.1.10, gender is a complete explanation for access to post-
secondary level education as even when controlled for poverty status, there is still a
statistical association (Non-poor: ρ value = 0.001, χ2 (1) = 79.905; Poor Ρ value = 0.001,
χ2 (1) = 34.612). According to the data (Table 4.7(iii)) above, 86.7 percent (n=4269) of
the males are not able to access post-secondary level education who are with the non-
poor categorization compared to 79.9 percent (n=3877) females. In respect to the poor,
97.1 percent (n=2415) are not able to access tertiary level education compared to 93.7
percent. On the contrary, 6.3 percent (n=182) of the females are able to access post-
secondary level education despite the social setting of being poor compared to 2.9 percent
(n=72) of the males.
223
Table 9.1.11: Regression Model Summary Model
1Model
2Model
3Model
4Model
5Model
6Model
7Model
8Model
9Model
10
Independent:Dependent variable: Access to Tertiary Level Education
Constant .121 .097 .084 .294 .317 .341 .430 .385 .394 .394
Poverty Status
-.094* -.079* -.077* -.077* -.079* -.076* -.065* -.065* -.065* -.065*
DummyKMA
.093* .095* .093* .091* .060* .060* .060* .060* .061*
DummyMarried
.045* .066* .066* .066* .072* .077* .083* .083*
LoggedAge
-.059* -.060* -.059* -.069* -.056* -.058* -.058*
DummyGender
-.038* -.037* -.041* -.043* -.046* -.046*
DummyRural
-.042* -.041* -.041* -.041* -.041*
LoggedHousehold size
-.033* -.040* -.040* -.040*
Dummy child of spouse
.039* .035* .035*
Dummy partner
-.017* -.016*
Dummy helper
-.112*
n 14912 14912 14912 14912 14912 14912 14912 14912 14912 14912
Ρ value .001 .001 .001 .001 .001 .001 .001 .001 .001 .001R .179 .232 .246 .266 .277 .284 .290 .295 .296 .296
R2 .032 .054 .060 .071 .076 .080 .084 .087 .087 .088
Error term .24577 .24298 .24217 .24083 .24010 .23960 .23915 .23878 .23871 .23867
F statistic 494.98 425.771 319.1 283.844 246.866 217.232 195.002 177.114 158.592 143.319
ANOVA (sig)
0.001 0.001 0.001 0.001 0.001 0.001 0.001 0.001 0.001 0.001
Model 1 [ Y= β0 + β1x1 + ei ] - where Y represents Index on Access to Tertiary Education, β0 denotes a constant, ei means error term and β1 indicates the coefficient of poverty x1 represents the variable poverty
Model 10 [Y= β0 + β1x1 + …+ βnxn ei]
* significant at the two-tailed level of 0.001
The findings in Table 9.1.11 above reveal that final model (i.e. Model 10) constitutes all
the determinants of access to tertiary level education. Model 10 has a Pearson’s
224
Correlation coefficient of 0.296 indicating that the relationship is a weak one. The
coefficient of determination, r2, (in Table 9.1.8 from Model 10) is 0.088 representing that
a 1 percent change in the determinants of (poverty status, area of residence, union status,
age, gender, household size, relationship with head of household) in predictor changes
the predictand by 8.8 percent to the sample observation is not a good fit. This means that
less that 8.8 percent of the total variation in the Yi is explained by the regression.
As shown in Table 9.1.11, Model 10, Testing Ho: β=0, with an α = 0.05, the
researcher can conclude that the linear model provides a good fit to the data from a F
value of [8.164, 0.057] = 143.319 with a ρ < 0.05.
The overall assessment of this causal model climax in Model 10, and so should be
disaggregated in order for a comprehensive understand of the phenomenon of poverty
and its influence on access to tertiary level education along with other determinants.
With all things being constant, access to tertiary level education has a value of 0.394 (i.e.
moderate access). From the findings in Table 4.8, poverty status is a negative value of
0.065 indicating that poverty is indirectly related to access to tertiary level education with
all other things held constant. On the other hand, there is a direct relationship between
person living in the Kingston Metropolitan Area and access to tertiary level education
compared to inverse relationship that exists between the rural residents and access to this
degree of education.
The results in Table 9.1.11 (Model 10) show that inverse association between
household size and access to post-secondary level education. This denotes that the larger
the household size becomes, the less likely that the individuals of that family will access
tertiary level education. Hence, household will smaller size means that the people therein
225
are more likely to attend post-secondary education. The data show for the age variable a
valuation of -0.058 that this indicates that younger people are more likely to access post-
secondary education than older persons. It is found that married people are more likely to
access post-secondary education in comparison to people in union status which is single,
none, visiting or common-law.
In relation to the issue of gender and access to post-secondary level education, a
value of negative 0.046 implies that men are less likely to access tertiary level education
than their female counterparts. The valuation indicates that women are 0.046 more likely
to attend post-secondary education than men. The results in Table 9.1.8 above show
helpers are less likely to access post-secondary education in comparison to the child of
the spouse. Compared to the child of the spouse concerning access to education, the
partner is more likely to acquire a post-secondary level education than the partner. The
latter elements are in regard to the question, ‘What is your relationship with the head of
the household’?
The focus of this text is the provision of materials that make a difference in the analysis
of SPSS output, and with this being the aim, one of my responsibility is in assisting with
the execution the various SPSS commands, which will generate the necessary output.
Hence, I will use an example of some metric variable which are not skewed to produce a
regression output. (See Appendix VII)
226
CHAPTER 10
Hypothesis 7:
There is an association between the introduction of the Inventory Readiness Test and the Performance of Students in Grade 1
ANALYSIS OF FINDINGS
Table 10.1.1: Univariate Analysis of Parental Information
Description Frequency (Percent)
Typology of School:SLB 18 (51.4)
KC 17 (48.6)
Gender:Male 7 (20)Female 28 ((80)
No. of children living at home0 17 (50)1 14 (40)2 2 (5.7)3 1(7.9)
No. of hours spent with childMean 9.77 hrsMedian 2.00 hrsMode 1.00 hrsStandard deviation 27.0 hrs
Of the sampled population (35 respondents), 51.4 percent (n=18) sent their children to
SLB compared to 48.6 percent (n=17) who sent them to KC. Approximately eight
percent (n=28) were females and 20 percent (n=7) males. Of the total respondents
227
interviewed, 50 percent (n=17) reported that they had no children under 6 years old living
at home, 40 percent (n=14) had 1 child, 5.7 percent (n=2) two children compared to 7.9
percent (n=1) had 3 children. When asked “how many hours spent with child?” the
average hours was approximately 10 ± 27 hours with the most frequent being 1 hour.
Table 10.1.2: Descriptive on Parental Involvement
Details Frequency (Percent)
Educational InvolvementMean 3.77Median 3.80Mode 3.6Standard deviation 0.89Skewness -0.395
Psychosocial InvolvementMean 3.4Median 3.4Mode 3.0Standard deviation 0.67Skewness -0.105
From the respondents’ information, they reported that educational involvement was 3.77
(i.e. agree) ± 0.89 with a skewness of -0.395 (i.e. this is negligible negative skewness);
psychosocial involvement was 3.4 (i.e. undecided) ± -0.105.
228
Table 10.1.3: Univariate Analysis of Teacher’s Information
Details Frequency (Percent)
Gender:Male 0 (0.0)Female 2 (100)
Age 31 to 40 years 1 (50.0) 41 to 50 years 1 (50.0)
Educational levelSecondary school diploma 1 (50.0)Teacher’s college diploma 1 (50.0)
Duration at this school11 years 1 (50.0) 12 years 1 (50.0)
Self-reported Learning EnvironmentUndecided 1 (50.0)Agree 1 (50.0)
Of the sampled population (2 teachers), 100 percent (n=2) were females compared to 0
percent males, with 50 percent (n=1) being 31 to 40 years and 50 percent (n=1) 41 to 50
years. The highest level of education was teacher’s college diploma (50%, n=1) followed
by secondary school diploma (50%, n=1). The minimum number of years spent at each
school is 11 years.
When the teachers were asked about the learning environment, 50 percent (n=1) was
undecided with 50 percent (n=1) agreeing.
229
Table 10.1.4: Univariate Analysis of ECERS-R Profile
Details Rating (Averaged score)
General (n=35) SLB (n=18) KC (n=18)Space and Furnishings 2.5 2.5 2.38
Personal Care Routines 2.0 1.8 2.17
Language-Reasoning 5.0 5.0 5.25
Activities 4 3.4 4.0
Interaction 5 6.6 5.0
Program Structure 6.0 6.0 6.00
Parents and Staff 5.0 5.17 5.33
From the average score of ECERS-R profile, overall, the space and furnishings in each
school was low but this was even lower in KC compared to SLB. With respect to
personal care routines offered, generally, it was poor with SLB depicting a lower
averaged score than KC. Language reasoning, on the other hand, was high (average of 5
out of 7) with KC showed a marginal higher rating than SLB. Overall, programme
structure was received the highest score (6 out of 7) and this was consistent across the
two school types. The averaged score received on activities was moderate (4) for KC but
weak (3.4) for SLB. On the other hand, interaction in SLB was higher (6.6) compared to
KC (5). Parent and staff rating were good in both institutions with KC marginally
receiving a better score than SLB.
230
Table 10.1.5: Bivariate Analysis of Self-reported Learning Environment and Mastery on Inventory Test
Final Report (before grade 1)
Learning Environment
Final Report (before grade 1)
Pearson Correlation 1 .344
Sig. (2-tailed) . .043 N 35 35Learning Environment Pearson Correlation .344 1 Sig. (2-tailed) .043 . N 35 35
* Correlation is significant at the 0.05 level (2-tailed).
From Table 10.1.5, there is a statistical significant relationship between Inventory Test
scores of Grade 1 students and their learning environment (ρ value = 0.043 <0.05). The
relationship is a weak positive one (Pearson Correlation Coefficient = 0.344 or 34.4 %).
This denotes that students’ learning environment explains 34.4 percent of readiness for
Grade 1. Statistically, although, this a weak relationship, for any single variable (i.e.
learning environment) to explain 34.4 percent of a relationship, the independent variable
(learning environment) has a very strong influence on readiness of students.
231
Table 10.1.6: Relationship between Educational Involvement, Psychosocial & Environment Involvement and Inventory Test
Final Report (before grade
1)
Educational Involvement
Psychosocial & Environmental Involvement
Final Report (before grade 1)
Pearson Correlation
1 .001 .241
Sig. (2-tailed) . .995 .162 N 35 35 35Educational Involvement
Pearson Correlation
.001 1 .735
Sig. (2-tailed) .995 . .000 N 35 35 35Psychosocial & Environmental Involvement
Pearson Correlation
.241 .735 1
Sig. (2-tailed) .162 .000 . N 35 35 35
** Correlation is significant at the 0.01 level (2-tailed).
Of the sampled population (n=35) parents of grade 1 students, no statistical relationship
existed between educational (ρ value = 0.995>0.05) psychosocial and environmental
involvement (ρ value 0.162>0.05) of parents and students readiness for grade 1. This
finding may be due to a Type I error, as the sample size is too small. In that when the
sample size was weighted by 6, 10 and so on, a with a new sample size of (i.e. weight 6 =
200, weight 10 = 350), a statistical relationship existed between the independent variable
(i.e. educational involvement, psychosocial and environmental involvement) and the
dependent variable (i.e. Readiness for grade 1 using the Inventory Readiness Test scores).
232
Table 10.1.7: BIVARIATE ANALYSIS OF THE INDEPENDENT VARIABLES AND READINESS FOR GRADE 1Final Report
(before grade 1)
Personal Care
Routines
Language-Reasoning
Activities Interaction Parents and Staff
PROGRAM Space and Furniture
N 35 35Personal Care Routines
Pearson Correlation
.344 1
Sig. (2-tailed) .043 N 35 35 35Language-Reasoning
Pearson Correlation
.344 1.000 1
Sig. (2-tailed) .043 N 35 35 35 35Activities Pearson
Correlation.344 1.000 1.000 1
Sig. (2-tailed) .043 N 35 35 35 35 35Interaction Pearson
Correlation-.344 -1.000 -1.000 -1.000 1
Sig. (2-tailed) .043 N 35 35 35 35 35 35 35Parents and Staff Pearson
Correlation.344 1.000 1.000 1.000 -1.000 1 .
Sig. (2-tailed) .043 .000 N 35 35 35 35 35 35 35 35PROGRAM Pearson
Correlation.
Sig. (2-tailed) . N 35 35 35 35 35 35 35 35Space and Furniture
Pearson Correlation
-.344 -1.000 -1.000 -1.000 1.000 -1.000 1
Sig. (2-tailed) .043N 35 35 35 35 35 35 35 35
* Correlation is significant at the 0.05 level (2-tailed).** Correlation is significant at the 0.01 level (2-tailed).
233
From Table 10.1.7, independently each of the following ECERS-R variables (i.e. Parents
and Staff, Space and Furnishing, Personal Care Routines, Language-Reasoning,
Activities and Interaction) has a statistical (ρ value 0.043 < 0.05) significantly
relationship with Readiness of grade 1 pupils. Generally, singly, the weight of each
relationship was very strong (i.e. despite Pearson’s Correlation Coefficient value of
0.344). Of the seven ECERS-R profile, programme (i.e. Program) structure is the only
one that was not statistically significant, with space and furnishing, and interaction
reporting a negative relationship (Pearson’s r = -0.344) and the other with a positive
association (Pearson’s Correlation Coefficient = 0.344). A positive association, for
example between Parents and staff, and Readiness of Grade 1 pupils, denotes that the
greater the parents and staff score the higher the readiness of the child who enters grade
1. On the other hand, a negative score, for example a relationship between interaction
and Readiness Test score, a low interaction will produce a high readiness on the
Inventory Test. This may be explained by what constitutes interaction, as a low grade
was reported for ‘supervision of gross motor activities’ compared to discipline, staff-child
interaction, interactions among children and general supervision of children that do not
directly influence readiness of a student on an examination.
234
Table 10.1.8: School type by Inventory Readiness Score (in %)
School Type TotalSLB KC
Final Report (before grade 1)
Non-mastery 88.9 58.8 74.3
Mastery 11.1 41.2 25.7 Total 18 17 35
Χ2 (1) = 4.137, ρ value = 0.049
There is a statistical relationship between type of school attended before grade 1 and
score on inventory test (i.e. Χ2 (1) = 4.137, Ρ value = 0.049). Of the 35 students in Grade
1, 88.9 percent of them got non-mastery from SLB compared to 58.8 percent of those
who attended KC. Of those who mastery the inventory test (n=9, 25.7%), 41.2 percent
attended KC compared to 11.1 percent who attended SLB. Embedded in this finding is
the super performance of students who went to KC basic.
235
CHAPTER 11Hypothesis 8:
The people who perceived themselves to be in the upper class and middle class are more so than those in the lower (or working) class do strongly believe that acts of incivility are only caused by persons in garrison communities
Table 11.1.1: INCIVILITY AND SUBJECTIVE SOCIAL STATUS
Case Processing Summary
1728 99.8% 3 .2% 1731 100.0%Incivility * Social StatusN Percent N Percent N Percent
Valid Missing Total
Cases
Column Totals and Totals
Incivility * Social Status Crosstabulation
296 8 96 400
37.0% 1.0% 100.0% 23.1%
17.1% .5% 5.6% 23.1%
472 120 0 592
59.0% 14.4% .0% 34.3%
27.3% 6.9% .0% 34.3%
32 688 0 720
4.0% 82.7% .0% 41.7%
1.9% 39.8% .0% 41.7%
0 8 0 8
.0% 1.0% .0% .5%
.0% .5% .0% .5%
0 8 0 8
.0% 1.0% .0% .5%
.0% .5% .0% .5%
800 832 96 1728
100.0% 100.0% 100.0% 100.0%
46.3% 48.1% 5.6% 100.0%
Count
% within Social Status
% of Total
Count
% within Social Status
% of Total
Count
% within Social Status
% of Total
Count
% within Social Status
% of Total
Count
% within Social Status
% of Total
Count
% within Social Status
% of Total
1=Strongly agree
2=Agree
3=Disagree
4=Strongly disagree
8
Incivility
Total
1=Lower(Working)
Class2=Middle
Class3=UpperMiddle
Social Status
Total
236
Chi-Square Tests
1425.277a 8 .000
1629.762 8 .000
220.288 1 .000
1728
Pearson Chi-Square
Likelihood Ratio
Linear-by-LinearAssociation
N of Valid Cases
Value dfAsymp. Sig.
(2-sided)
6 cells (40.0%) have expected count less than 5. Theminimum expected count is .44.
a.
Symmetric Measures
.672 .000
1728
Contingency CoefficientNominal by Nominal
N of Valid Cases
Value Approx. Sig.
Not assuming the null hypothesis.a.
Using the asymptotic standard error assuming the null hypothesis.b.
INTERPRETATION OF INCIVILITY AND SUBJECTIVE SOCIAL STATUS (using the information from Tables 1.1, above)
Based on Tables 11.1.1, the results reveal that there is a statistical relationship
between‘incivility’ and ‘subjective social class’ (χ2 (8) = 1425.28, Ρ value = 0.001 <
0.05). The findings show that there is a direct association ‘incivility’ and ‘subjective
social class’ (i.e. this is based on the positive value of 0.672). The strength of the
relationship is moderately strong (cc = 0.672). Approximately 45 % (i.e. cc2 * 100 –
0.672 * 0.672 * 100) of the proportion of variation in ‘incivility’ is explained by an
incremental change from one subjective social class to the next (for example, a
movement from lower class to middle class or from middle class to upper class).
237
Of the respondents who had indicated ‘strongly agree’ (n=400, 23.1%), 37.0%
percent of them (n=296) were from the ‘lower class’ while 1.0 % (n=8) were from
‘middle class’ compared to 100 % (n=96) who classified themselves as being in the
‘upper class’. Of those responded ‘Agree’ (n=592, 34.3%), 59.0% (n=472) of them were
within the ‘lower class’, 14.4% (n=120) in the ‘middle class’ and 0.0% (n=0) from the
‘upper class’. While of those who ‘disagree[d]’ with ‘incivility’ (41.7%, n=720), 4.0 %
(n=32) were ranked in the ‘lower class’, 82.7% (n=688) from the ‘middle class’ and 0%
(n=0) within the ‘upper class’. Ergo, we accept the H1 (alternative hypothesis) and by so
doing reject the Ho (i.e. the null hypothesis).
Let us assume that within the ‘Symmetric measure’ the ‘approximate significant’ (i.e. the Ρ value) was greater than 0.05 (for example 0.256), the analysis would read:
The results in Tables 1.1 above, indicate that there is no statistical relationship between
the ‘incivility’ and ‘subjective social class’ (χ 2(8) = 0.256, p>0.05) of the population
sampled. This implies that perception on ‘incivility’ is not associated (or related) in no
statistical way with ones classification of him/herself within the social strata of society.
Thus, we reject the H1 (alternative hypothesis) or fail to reject the Ho (i.e. the null
hypothesis).
(Note briefly – this none relationship must be explained and/or justified using empirical
data or the result may argue that this is due to a Type II Error – See Appendix II). Type II
Errors occur, when the statistical correlation reveals no relationship but in reality an
association does exist. This may be as a (i) the sample size is ‘too’ small; (ii) ‘too’ many
of the cells in the cross tabulations have less than ‘5’ respondents; (iii) errors exist in the
data collection process and (iv) issues relating to validity and/or reliability.
238
CHAPTER 12
Table 12.1.1: Do you believe that corruption is a serious problem in Jamaica?
Frequency PercentValid
PercentCumulative
PercentValid Not a serious
problem35 3.1 3.2 3.2
Somewhat serious
185 16.2 16.7 19.9
Very serious 886 77.7 80.1 100.0 Total 1106 97.0 100.0Missing -99.00 24 2.1 -98.00 2 .2 -88.00 8 .7 Total 34 3.0Total 1140 100.0
As shown in Table? majority of the respondents indicated that corruption is a very serious
problem in Jamaica (80.1%, n=886), with approximately 17% (n=185) ‘somewhat serious’
compared to 3.2% (n=35) who remarked it was ‘not a serious problem.
Table 12.1.2: Have you or someone in your family known of an act of corruption in the last 12 months?
Frequency PercentValid
PercentCumulative
PercentValid Yes 406 35.6 40.1 40.1 No 606 53.2 59.9 100.0 Total 1012 88.8 100.0Missing
-99.0026 2.3
-98.00 96 8.4 -88.00 6 .5 Total 128 11.2
239
Total 1140 100.0
Of the sampled population (n=1140), 88.8% (n=1012) responded to this question. The results
indicated that approximately 60% (n=606) of the respondents believed ‘No’ compared to 40%
(n=406) who remarked ‘Yes’.
Table 12.1.3: Gender of Respondent
Frequency PercentValid
PercentCumulative
PercentValid Male 511 44.8 46.8 46.8 Female 581 51.0 53.2 100.0 Total 1092 95.8 100.0Missing -99.00 43 3.8 -88.00 5 .4 Total 48 4.2Total 1140 100.0
Of the sampled population (n=1140), approximately 45 percent (n=511) were males compared to 51 percent (n=581) who were females. The non-response rate was approximately 4 percent.
240
Table 12.1.4: In what Parish do you live?
Frequency PercentValid
PercentCumulative
PercentValid Clarendon 105 9.2 9.3 9.3 Hanover 59 5.2 5.2 14.6 Kingston 112 9.8 9.9 24.5 Manchester 122 10.7 10.8 35.3 Portland 95 8.3 8.4 43.8 Saint
Andrew18 1.6 1.6 45.4
Saint Ann 70 6.1 6.2 51.6 Saint
Catherine143 12.5 12.7 64.3
Saint Elizabeth
77 6.8 6.8 71.1
Saint James 106 9.3 9.4 80.6 Saint Mary 30 2.6 2.7 83.2 Saint
Thomas74 6.5 6.6 89.8
Trelawny 52 4.6 4.6 94.4 Westmorela
nd63 5.5 5.6 100.0
Total 1126 98.8 100.0Missing -99.00 14 1.2Total 1140 100.0
241
Table 12.1.5: Suppose that you, or someone close to you, have been a victim of a crime. What would you do...?
Frequency PercentValid
PercentCumulative
PercentValid Report it to an
influential neighbour or don
89 7.8 8.3 8.3
Settle the matter yourself
72 6.3 6.7 14.9
Report it to a private security company
48 4.2 4.5 19.4
Report the crime to the police
802 70.4 74.5 93.9
Do nothing 35 3.1 3.2 97.1 Other 31 2.7 2.9 100.0 Total 1077 94.5 100.0Missing -99.00 46 4.0 -98.00 17 1.5 Total 63 5.5Total 1140 100.0
Generally, 74.5% (n=802) of the sampled population (n=1140) reported that they would inform the police in the event that someone that they know has been victimized by another. On the other hand, approximately 8% (n=89) indicated that they would use an influential community member or a ‘Don’, with some 7% (n=72) stating they would ‘settle matter themselves’.
242
Table 12.1.6: What is your highest level of education?
Frequency PercentValid
PercentCumulative
PercentValid No formal
education17 1.5 1.5 1.5
Primary/Prep school
51 4.5 4.6 6.1
All-Age school or some Secondary education
172 15.1 15.4 21.5
Completed secondary school
319 28.0 28.6 50.2
Vocational/Skills training
188 16.5 16.9 67.1
University graduate (Undergraduate)
250 21.9 22.4 89.5
Some professional training beyond university
69 6.1 6.2 95.7
Graduate degree (MSc, MA, PhD etc)
48 4.2 4.3 100.0
Total 1114 97.7 100.0Missing -99.00 20 1.8 -98.00 2 .2 -88.00 4 .4 Total 26 2.3Total 1140 100.0
Most of the sampled population had attained at completed secondary (i.e. high) school education (28%, n=319); with 21.9% (n=250) an undergraduate level, 16.5% (n=188) a vocational level education, 15.1% (n=172) and 6.1% professional. The non-response rate was approximately 2% (n=26)
243
Table 12.1.7: In terms of work, which of these best describes your present situation?
Frequency PercentValid
PercentCumulative
PercentValid Employed, Full-
Time job497 43.6 43.9 43.9
Employed, Part-Time job
69 6.1 6.1 50.0
Seasonally employed
49 4.3 4.3 54.3
Temporarily employed
50 4.4 4.4 58.7
Self-employed 186 16.3 16.4 75.2 Unemployed,
out of work91 8.0 8.0 83.2
Retired 32 2.8 2.8 86.0 Housewife 17 1.5 1.5 87.5 Student 116 10.2 10.2 97.8 Sick/Disabled 25 2.2 2.2 100.0 Total 1132 99.3 100.0Missing -99.00 6 .5 -98.00 2 .2 Total 8 .7Total 1140 100.0
Of the surveyed population (n=1140), the response rate, for this question, was 99.3% (n=1132). Approximately 44% (n=497) of the sampled population were full-time employees, 16.4% (n=186) self-employed, 10.2 % (n=116) were students, 6.1% (n=69) part-time employees, 4.3 % (n=49) seasonally employed, 4.4% (n=50) temporarily employed, 2.8% (n=32) retirees, 2.2 % (n=25) physically challenged and 1.5 % (n=17) were housewives.
244
Table 12.1.8: Which best represents your present position in Jamaica society?
Frequency PercentValid
PercentCumulative
PercentValid Working
(lower) class562 49.3 50.9 50.9
Middle class 421 36.9 38.1 89.0 Upper-middle
class70 6.1 6.3 95.3
upper class 52 4.6 4.7 100.0 Total 1105 96.9 100.0Missing -99.00 27 2.4 -98.00 1 .1 -88.00 7 .6 Total 35 3.1Total 1140 100.0
Of the population surveyed (n=1140), the response rate was 96.9% (n=1105). Some 50.9 percent (n=562) perceived themselves to be within the working-class categorization, 38.1 percent (n=421) middle-class, 6.3 percent (n=70) within the upper-middle class compared to 4.7 percent (n=52) who said upper class.
Table 12.1.9: Age on your last birthday? N Valid 1058 Missing 82Mean 35.6805Std. Deviation 13.25951Skewness .710Std. Error of Skewness .075
The average age of the sampled population (n=1140) is 35 years and 8 months ± 13 years and 3 months. The non-response rate was 7 percent.
245
Table 12.1.10: Age Categorization of respondents
Frequency PercentValid
PercentCumulative
PercentValid 1= Young (less
than 26 yrs)289 25.4 27.3 27.3
2= middle-aged (between 25 and 60 yrs)
717 62.9 67.8 95.1
3= seniors (older than or equal to 60 yrs)
52 4.6 4.9 100.0
Total 1058 92.8 100.0Missing System 82 7.2Total 1140 100.0
The sampled population (n=1140) was predominately of people within the middle-aged categorization (67.8%, n=717) with 27.3 % (n=289) being young people compared to 4.9% (n=52) seniors.
246
Table 12.1.11: Suppose that you, or someone close to you, have been a victim of a crime. What would you do... * Gender of Respondent Cross tabulation
Gender of
Respondent Total Male Female Suppose that you, or someone close to you, have been a victim of a crime. What would you do
Report it to an influential neighbour or don
Count
43 43 86
% within Gender of Respondent
8.9% 7.9% 8.3%
Settle the matter yourself
Count39 33 72
% within Gender of Respondent
8.0% 6.0% 7.0%
Report it to a private security company
Count21 22 43
% within Gender of Respondent
4.3% 4.0% 4.2%
Report the crime to the police
Count356 413 769
% within Gender of Respondent
73.4% 75.6% 74.6%
Do nothing Count 15 17 32 % within Gender of
Respondent3.1% 3.1% 3.1%
Other Count 11 18 29 % within Gender of
Respondent2.3% 3.3% 2.8%
Total Count 485 546 1031 % within Gender of
Respondent100.0% 100.0% 100.0%
Chi-Square Tests
Value dfAsymp. Sig.
(2-sided)Pearson Chi-Square 2.964(a) 5 .706Likelihood Ratio 2.973 5 .704Linear-by-Linear Association
2.043 1 .153
N of Valid Cases1031
a 0 cells (.0%) have expected count less than 5. The minimum expected count is 13.64.
There is not statistical relationship that was found between the two variables.
247
Table 12.1.12: If involved in a dispute with neighbour and repeated discussions have not made a difference, would you...? * Gender of Respondent Cross tabulation
Gender of
Respondent Total Male Female If involved in a dispute with neighbour and repeated discussions have not made a difference, would you...?
Report it to an influential neighbour or don
Count
58 66 124
% within Gender of Respondent
12.1% 12.1% 12.1%
Settle the matter yourself
Count68 36 104
% within Gender of Respondent
14.2% 6.6% 10.2%
Report it to a private security company
Count12 13 25
% within Gender of Respondent
2.5% 2.4% 2.4%
Report the crime to the police
Count303 382 685
% within Gender of Respondent
63.4% 70.0% 66.9%
Do nothing Count 15 24 39 % within Gender of
Respondent3.1% 4.4% 3.8%
Other Count 22 25 47 % within Gender of
Respondent4.6% 4.6% 4.6%
Total Count 478 546 1024 % within Gender of
Respondent100.0% 100.0% 100.0%
248
Chi-Square Tests
Value dfAsymp. Sig.
(2-sided)Pearson Chi-Square 17.342(a) 5 .004Likelihood Ratio 17.464 5 .004Linear-by-Linear Association
4.666 1 .031
N of Valid Cases1024
a 0 cells (.0%) have expected count less than 5. The minimum expected count is 11.67.
When the respondents’ answers for “If involved in a dispute with neighbour and repeated
discussions have not made a difference, would you...?” was cross tabulated with ‘gender’, a
significant statistical association was found (χ2 (5) = 17.342, Ρ value =.004< 0.05). Some 12%
(n=124) of the respondents indicated that they would address the matter(s) through an influential
individual within the community or a don. Furthermore analysis revealed that both males and
females (12%) would use the same source – influential community member or ‘don’.
With regard to addressing the matter personally, approximately twice the number of males
(14.2%, n=68) would do this compared to females (6.6%, n=36). On the other hand, marginally
more females (70%, n=382) than males (63.4%, n=303) would inform the police, and a similar
situation existed in respect to ‘doing nothings and using ‘other’ approaches – females (4.4%,
n=24) and 3.1% (n=15) for males and females (4.6%, n=22) and 4.6% (n=25) for males
respectively.
249
Table 12.1.13: Do you believe that corruption is a serious problem in Jamaica? * Gender of Respondent Cross tabulation
Gender of
Respondent Total Male Female Do you believe that corruption is a serious problem in Jamaica?
Not a serious problem Count17 16 33
% within Do you believe that corruption is a serious problem in Jamaica?
51.5% 48.5% 100.0%
Somewhat serious Count 91 82 173 % within Do you
believe that corruption is a serious problem in Jamaica?
52.6% 47.4% 100.0%
Very serious Count 388 468 856 % within Do you
believe that corruption is a serious problem in Jamaica?
45.3% 54.7% 100.0%
Total Count 496 566 1062 % within Do you
believe that corruption is a serious problem in Jamaica?
46.7% 53.3% 100.0%
Chi-Square Tests
Value dfAsymp. Sig.
(2-sided)Pearson Chi-Square 3.376(a) 2 .185Likelihood Ratio 3.369 2 .186Linear-by-Linear Association 2.859 1 .091
N of Valid Cases1062
a 0 cells (.0%) have expected count less than 5. The minimum expected count is 15.41.
From Table, no statistical relationship exists between ‘Do you believe that corruption is a serious
problem in Jamaica’ and the Gender of the Respondents.
250
Table 12.1.14: Have you or someone in your family known of an act of corruption in the last 12 months? * Gender of Respondent Cross tabulation
Gender of
Respondent Total Male Female Have you or someone in your family known of an act of corruption in the last 12 months?
Yes Count 192 198 390
% within Have you or someone in your family known of an act of corruption in the last 12 months?
49.2% 50.8% 100.0%
No Count 257 321 578 % within Have
you or someone in your family known of an act of corruption in the last 12 months?
44.5% 55.5% 100.0%
% within Have you or someone in your family known of an act of corruption in the last 12 months?
46.4% 53.6% 100.0%
251
Chi-Square Tests
Value dfAsymp. Sig.
(2-sided)Exact Sig. (2-sided)
Exact Sig. (1-sided)
Pearson Chi-Square 2.128(b) 1 .145Continuity Correction(a)
1.941 1 .164
Likelihood Ratio 2.127 1 .145Fisher's Exact Test .149 .082Linear-by-Linear Association
2.126 1 .145
N of Valid Cases 968a Computed only for a 2x2 tableb 0 cells (.0%) have expected count less than 5. The minimum expected count is 180.90.
Based on the findings in Table, there is no statistical association between responses garnered
from “Have you or someone in your family known of an act of corruption in the last 12 months?”
tabulated by Gender of Respondent.
252
CHAPTER 13
Hypothesis 10: There is no statistical difference between the typology of workers in the construction industry and how they view 10-most top productivity outcomes
SOCIODEMOGRAPHIC CHARACTERISTICS
Categorization of respondents
45.9
33.8
13.56.8
05
101520253035404550
Fie
ld w
ork
forc
e
Fie
ldS
up
eri
nte
nd
en
t
Pro
ject
ma
na
ge
r
Exe
cutiv
e(C
EO
,P
resi
de
nt,
VP
)
Figure13.1.1: Categories that describe respondents’ position
Of the sampled population (n=80), the non-response rate was 7.5% (n=6).
Approximately 45.9% of the respondents (n=34) were from ‘Field workforce’, 33.8%
(n=25) ‘Field Superintendent’, 13.5% (n=10) ‘Project manager’ compared to 6.8% (n=5)
‘Executive’.
253
COMPANY’S ANNUAL WORK VOLUME
10.5
21.1
26.3
42.1
0
5
10
15
20
25
30
35
40
45
Un
de
r 2
5m
illio
nd
olla
rs
26
- 5
0m
illio
nd
olla
rs
51
- 1
00
mill
ion
do
llars
Ove
r 1
00
mill
ion
do
llars
Figure13.1.2: Company’s annual work volume
Based on Figure 1.2, 42.1% of the respondents (n=16) remarked that their company’s
annual work volume in dollars was ‘Over 100 million’, 26.3% between ’51 and 100
million’, 21.1% ’26 to 50 millions’ compared to 10.5% ‘under 25 million.
254
LABOUR FORCE – ‘ON AN AVERAGE PER YEAR’
23.1
48.7
28.2
05
101520
25303540
4550
Un
de
r 5
0
50
- 2
49
Ove
r 2
50
Figure13.1.3: Company’s Labour Force – ‘On an average per year’
Of the sampled population (n=80), using Figure 1.3, approximately 49% of the
respondents (n=19) said that their companies employed ’50 to 249’ employers per annum
per average, with some 28% remarked ‘over 250’ employees compared to 23% who said
‘under 50’ employees.
255
MAIN AREA OF CONSTRUCTION WORK
32.5 32.5
20.0
2.5
12.5
0
5
10
15
20
25
30
35
Co
mm
erc
ial
Re
sid
en
tial
Hig
hw
ay
Pu
blic
Wo
rks
Oth
er
Figure13.1.4: Respondents’ main area of construction work
Based on Figure 1.4, 50% of the respondents (n=40) responded to this question. Of the
respondents (n=40), approximately33% said ‘Commercial and Residential, 20%
remarked ‘Highways’, 2.5% ‘Public Works’ and 12.5% said ‘Other’.
256
SELF-PERFORMED IN CONTRAST TO SUB-CONTRACTED
11.6
20.9
32.6
23.3
11.6
0
5
10
15
20
25
30
35
1 -
10
%
11
- 2
5 %
26
- 5
0 %
51
- 7
5 %
76
- 1
00
%
Figure13.1.5: Percentage of work ‘Self-performed’ in contrast to ‘Sub-contracted’
Of the sampled population (n=80), the non-response rate was 46.2% (n=37). Of the
respondents (n=43), 11.6 % indicated that between ‘1 and 10%’ of their work was ‘Self-
performed’ compared to ‘Sub-contracted’, with 20.9% said between ’11 to 25%’, 32.6%
revealed ’51 to 75%’, with 23.3% make mention that it was between ’26 and 50%,
compared to 11.6% who mentioned ’76 – 100%.
257
AGE COHORT OF RESPONDENTS
14.9
37.8
25.721.6
0
5
10
15
20
25
30
35
40
18
- 2
4 y
rs
25
- 3
4 y
rs
35
- 4
4 y
rs
Ove
r 4
5 y
rs
Figure13.1.6: Percentage of work ‘Self-performed’ in contrast to ‘Sub-contracted’
Figure 1.6 revealed that the modal age (37.8%, n=28) group was 25 – 34 years.
Approximately 22% of the respondents were older than 45 years with 14.9% between the
age cohort of ’18-24’ years and another 25.7% being ’35 to 44’ years.
258
YEARS OF EXPERIENCE IN CONSTRUCTION INDUSTRY
35.1
2324.3 17.6
0
5
10
15
20
25
30
35
40
Un
de
r 5
yrs
5 -
9 y
rs
10
-1
9 y
rs
Ove
r 2
0 y
rs
Figure13.1.7: Years of Experience in Construction Industry
259
PRIMARY AREA OF EMPLOYMENT
35.1
2324.3
0
5
10
15
20
25
30
35
40
Kin
gst
on
an
d S
t.A
nd
rew
No
rth
Co
ast
Mig
rato
ry(c
om
bin
e a
an
d b
)
Figure13.1.8: Geographical Area of Employment
260
DURATION IN PRESENT EMPLOYMENT
0
5
10
15
20
25
30
35
40
45
50
Less than 2yrs
2 - 5 yrs 6 - 9 yrs Over 10 yrs
Figure13.1.9: Duration of service with current employer
When asked “How long have you been with your present employer?” 90 % of the
respondents (n=72) answered this question. Most of the respondents (50%, n=36)
indicated less than 2 years, with 22.2% (n=16) mentioned 2-5 years, 8.3% (n=6) said 6-9
years compared to 19.4% (n=14) saying over 10 years 9(see Figure 1.9).
261
PRODUCTIVITY CHANGES IN THE PAST FIVE YEARS
6.23.1
10.8
47.7
32.3
0
5
1015
2025
30
3540
4550
Sig
nifi
can
tlyd
ecr
ea
sed
De
cre
ase
dsl
igh
tly
Ha
s n
ot
cha
ng
ed
Imp
rove
dsl
igh
tly
Imp
rove
dsu
bst
an
tially
Figure13.1.10: Productivity changes over the past five years
Of the sampled population (n=80), the response rate was 81.3% (n=65). Of the
respondents (n=65), approximately 48% indicated that their company had ‘Improved
slightly’, with 32% mentioned ‘Improved substantially’, and some 11% remarked ‘Has
not changed’ compared to 3.1% who said ‘Decreased slightly’, with 6.2% mentioned
‘Significantly decreased’.
262
SELF-RATED PERCEPTION of PRODUCTIVITY IN CONSTRUCTION SECTOR
1 2 3 4 5 Mean Mode Median
11
Work force skill and experience? 4.45 5.00 5.00
12
Workers’ motivation? 4.25 5.00 4.00
13
Frequency of breaks? 3.55 3.00 3.00
14
Absenteeism and turnover? 4.00 5.00 4.00
15
Poor use of turnover? 3.77 4.00 4.00
16
Pay increases and bonuses? 4.10 5.00 4.00
17
Better management? 4.15 5.00 4.00
18
Job planning?Lack of pre-task planning?
4.36 5.00 5.00
19
4.04 4.00 4.00
20
Lack of work force training? 4.11 5.00 4.00
21
Internal delay (crew interfacing)? 3.65 3.00 4.00
22
Waiting for instructions? 3.57 4.00 4.00
2 Management’s resistance of change 3.70 4.00 4.00
263
324
Supervision delays? 3.60 3.00 4.00
25
Safety (near misses and accidents)? 3.68 5.00 4.00
26
Poor construction methods? 4.03 5.00 4.00
27
Weather conditions? 3.89 5.00 4.00
28
Shortage of skilled labour? 4.06 5.00 4.00
29
Lack of proper tools and equipment? 4.18 5.00 4.50
30
Incentives that reward maintenance of status quo or that reward unproductive employeesAs well as productive ones
3.62 3.00 4.00
SCALE: 1 = No impact; 2 =Low importance; 3 = Moderate; 4 = Important; 5 = Very important’ N/A = Not applicable
SELF-RATED PERCEPTION of PRODUCTIVITY IN CONSTRUCTION SECTOR (con’td)
1 2 3 4 5 Mean Mode Median
31 Ignoring or not asking for employers input? 3.48 4.00 4.00
32 Lack of quality control? 4.03 4.00 4.00
33 Equipment breakdown? 3.93 4.00 4.00
34 Lack of material? 4.13 5.00 4.00
35 Late material fabrication and delivery? 3.69 4.00 4.00
264
36 Congested work areas? 3.34 4.00 4.00
37 Poor drawing or specification? 3.94 5.00 4.00
38 Change orders and rework?Regulatory burdens?
3.68 3.00 4.00
39 3.46 3.00 3.00
40 Inspection delays? 3.38 3.00 3.00
41 Local union and politics? 3.80 4.00 4.00
42 Poor communication between office and field? 4.33 4.00 4.00
43 Project uniqueness (size and complexity)? 3.03 3.00 3.00
44 Theft of material and equipment? 3.86 5.00 4.00
45 Extortion? 3.52 5.00 3.00
SCALE: 1 = No impact; 2 =Low importance; 3 = Moderate; 4 = Important; 5 = Very important’ N/A = Not applicable
265
THE 10 MOST IMPORTANT SELF-RATED PERCEPTION INDICATORS OF PRODUCTIVITY IN CONSTRUCTION SECTOR
1 2 3 4 5 Mean Mode Median
1 Work force skill and experience (Ques11) 4.45 5.00 5.00
2 Job planning (Ques18) 4.36 5.00 5.00
3 Poor communication between office and field (Ques42) 4.33 4.00 4.00
4 Workers’ motivation (Ques12) 4.25 5.00 4.00
5 Lack of proper tools and equipment (Ques29) 4.18 5.00 4.50
6 Better management (Ques17) 4.15 5.00 4.00
7 Lack of material (Ques34) 4.13 5.00 4.00
8 Lack of work force training (Ques20)Pay increases and bonuses (Ques16)
4.11 5.00 4.00
9 4.10 4.00 5.00
10 Shortage of skilled labour (Ques28) 4.06 5.00 4.00
TOTAL
SCALE: 1 = No impact; 2 =Low importance; 3 = Moderate; 4 = Important; 5 = Very important’ N/A = Not applicable
266
Table 13.1.1: RESEARCH QUESTION # 1: Spearman’s rho
ques01 ques11 ques12 ques16 ques17 ques18 ques20 ques28 ques34 ques29 ques42ques01 Correlation Coefficient 1.000 .140 .108 -.073 .137 .270(*) .158 .081 -.030 -.025 .062 Sig. (2-tailed) . .236 .361 .541 .256 .022 .208 .499 .801 .838 .614 N 74 74 73 72 71 72 65 72 72 72 69ques11 Correlation Coefficient .140 1.000 .544(**) .173 .348(**) .212 .372(**) .297(*) .169 .421(**) .069 Sig. (2-tailed) .236 . .000 .145 .003 .074 .002 .011 .157 .000 .573 N 74 74 73 72 71 72 65 72 72 72 69ques12 Correlation Coefficient .108 .544(**) 1.000 -.040 .134 .032 .109 .278(*) .254(*) .388(**) -.024 Sig. (2-tailed) .361 .000 . .739 .268 .793 .387 .018 .032 .001 .843 N 73 73 73 71 70 71 65 72 71 71 68ques16 Correlation Coefficient -.073 .173 -.040 1.000 .194 .143 -.005 -.127 -.013 -.087 -.044 Sig. (2-tailed) .541 .145 .739 . .111 .236 .966 .296 .914 .465 .721 N 72 72 71 72 69 70 64 70 70 72 68ques17 Correlation Coefficient .137 .348(**) .134 .194 1.000 .517(**) .196 .192 .144 .140 .396(**) Sig. (2-tailed) .256 .003 .268 .111 . .000 .120 .114 .237 .250 .001 N 71 71 70 69 71 70 64 69 69 69 67ques18 Correlation Coefficient .270(*) .212 .032 .143 .517(**) 1.000 .220 .238(*) .151 -.027 .345(**) Sig. (2-tailed) .022 .074 .793 .236 .000 . .079 .047 .212 .821 .004 N 72 72 71 70 70 72 65 70 70 70 67ques20 Correlation Coefficient .158 .372(**) .109 -.005 .196 .220 1.000 .319(*) .225 .361(**) .355(**) Sig. (2-tailed) .208 .002 .387 .966 .120 .079 . .010 .077 .003 .005 N 65 65 65 64 64 65 65 64 63 64 62ques28 Correlation Coefficient .081 .297(*) .278(*) -.127 .192 .238(*) .319(*) 1.000 .575(**) .695(**) .277(*) Sig. (2-tailed) .499 .011 .018 .296 .114 .047 .010 . .000 .000 .022 N 72 72 72 70 69 70 64 72 70 70 68ques34 Correlation Coefficient -.030 .169 .254(*) -.013 .144 .151 .225 .575(**) 1.000 .556(**) .454(**) Sig. (2-tailed) .801 .157 .032 .914 .237 .212 .077 .000 . .000 .000 N 72 72 71 70 69 70 63 70 72 70 67
* Correlation is significant at the 0.05 level (2-tailed).** Correlation is significant at the 0.01 level (2-tailed).
267
Based on the statistical test (Spearman rho) which was performed on ‘The 10 most important self-rated perception indicators of?
productivity in construction sector’, the findings revealed that only ‘Job planning’ and ‘Categorization of position was statistically
related. This implies that, hierarchal level that one holds within the construction level is positively related to ‘Job planning’ (cc= 0.27,
Ρ value < 0.05), and not any of the other characteristics identified in the ‘Top 10’ indicators. Based on the contingency coefficient
(0.27 or 27%), the association is a moderately weak one.
268
RESEARCH QUESTION # 2
The statistical test revealed that irrespective of the respondents’ area of specialization in the construction industry, the ‘Top 10 indicators’ are the same. This can have been caused by the sample size (Type II Error – See Appendix II).
RESEARCH QUESTION # 3
The statistical test revealed that irrespective of the respondents’ location of employment in the construction industry, the ‘Top 10 indicators’ remain the same. This can have been caused by the sample size (Type II Error).
ESEARCH QUESTION # 4
The statistical test revealed that irrespective of the respondents’ years of experience in the construction industry, the ‘Top 10 indicators’ remain the same. This can have been caused by the sample size (Type II Error – see Appendix II).
CHAPTER 14
Hypothesis 11: Determinants of the academic performance of students
SOCIO-DEMOGRAPHIC VARIABLES
parent81%
guardian19%
Figure 14.1.1: Characteristic of Sampled Population
Of the sampled population (n=100), 81 percent (n=81) were parents (i.e. biological
parents) compared to 19 percent (n=19) were guardians. (See, Figure 14.1.1)
Predominantly the sampled population was single individuals (45 %, n=45) compared to
39 percent who were married, 12 percent divorced and 4 percent who were remarried
people (See, Table 14.1.1).
Table 14.1.1: Marital Status of RespondentsDetail Frequency Percent
Single 45 45Married 39 39Divorced 12 12Remarried 4 4
Total 100 100
270
Table 14.1.2: Marital Status of Respondents by Gender
gender of
respondents Total Marital status male female
single 521.7%
4051.9%
4545.0%
married 10 29 39 43.5% 37.7% 39.0%
divorced 7 5 12 30.4% 6.5% 12.0%
remarried 1 3 4 4.3% 3.9% 4.0%
Total 23 77 100
Based on Table 14.1.2, 77 percent (n=77) of the respondents were females, of which 51.9
percent (n=40) were single mothers compared to 37.7 percent (29) who were married, 6.5
percent divorced and 3.9 percent (n=3) who had got remarried. Only 23 percent (n=23)
of the sampled population were males, of which approximately 44 percent (n=10) were
married men compared to some 22 percent (n=5) who were single, 30.4 percent (n=7)
divorced and 4.3 percent (n=1) were remarried fathers.
271
Table 14.1.3: Marital Status by Gender by Age Cohort
Gender Marital StatusAge
20 – 30 Yrs
Age
31 – 40 Yrs
Age
Above 40 Yrs
MaleSingle 0 (0.0%) 1 (16.7%) 4(26.7%)Married 1 (50.0%) 3 (50.0%) 6(40.0%)Divorced 1 (50.0%) 2 (33.3%) 4(26.7%)Remarried 0 (0.0%) 0 (0.0%) 1(6.7%)
FemaleSingle 5 (71.4%) 22 (68.8%) 13(34.2%)Married 2 (28.6%) 8 (25.0%) 19(50.0%)Divorced 0(0.0%) 2 (6.3%) 3(7.9%)Remarried 0 (0.0%) 0 (0.0%) 3(7.9%)
Generally the sampled population was from beyond 40 years (53 %, n=53), of which 72
percent (n=38) were females. Of the respondents who were older than 40 years, they
were primarily married men (40%, n=6) and married females (50%, n=19). Only 9
percent of the respondents were younger than 30 years with 71.4 percent (n=5) being
single females compared to no single male of the same age cohort. Approximately 28
percent (n=2) of the respondents who were younger than 30 years were married
compared to 50 percent (n=1) of males (See, Table 14.1.3).
employed80%
unemployed20%
Figure 14.1.2: Employment Status of Respondents
Generally the sampled population was employed (80%, n=80).
272
Table 14.1.4: Marital Status by Gender by Age Cohort
Gender Marital StatusAge
20 – 30 Yrs
Age
31 – 40 Yrs
Age
Above 40 Yrs
Male Employed 2(1000%) 4 (66.7%) 14(93.3%)Unemployed 0 (0.0%) 2 (33.3%) 1(6.7%)
Female Employed 5 (71.4%) 21(65.6%) 34(89.5%)Unemployed 2(28.6%) 11 (34.4%) 4(10.5%)
Of the 80 percent (n=80) of the sampled population who were employed, 90.6 percent
(n=48) were beyond age 40 years, or which 89.5 percent (n=34) were females compared
to 93.3 percent (n=14) who were males. However, only 77.8 percent (n=7) of the people
younger than 31 years were employed with 71 percent being females compared to all the
males being employed (100%, n=2). In regard to the people who were 31 to 40 years at
their last birthday, the employment rate was 65.8 percent. Approximately 66 percent
(n=21) of that age cohort was female compared to 68 percent (n=4) male.
273
Table 14.1.5 Educational Level by gender by age cohorts
Gender Marital StatusAge
20 – 30 Yrs
Age
31 – 40 Yrs
Age
Above 40 Yrs
MaleNone 0 (0.0%) 0 (0.0%) 1 (6.7%)Primary 0 (0.0%) 1 (16.7%) 4 (26.7%)High 1 (50.0%) 4 (66.7%) 2(13.3%)College 0 (0.0%) 0 (0.0%) 2(13.3%)
Tertiary 1 (50.0%) 1 (16.7%) 6 (40.0%)
Female None 0 (0.0%) 3 (9.4%) 0 (0.0%)Primary 2 (28.6%) 8 (25.0%) 6(15.8%)High 3(42.9%) 15 (15.6%) 16(42.1%)
College
Tertiary
0(0.0%)
2 (28.7)
5 (15.6%)
4(12.5%)
7 (18.4%)
9 (23.7%)
The highest level of educational attainment of the sampled population (n=100) was
tertiary with 23 percent (n=23) compared to 38 percent (n=38) who had completed
high/secondary level education, 21.0 percent (n=21) primary, 14 percent (n=14) college
and only 4 percent (n=4) of who had no formal education. Of the seventy-seven percent
(n=77) of the sampled females, the most frequently highest level of formal education had
was secondary (40.3%, n=31) compared to university for the males (34.8%, n=8). Only 4
percent (n=4) of the sampled respondents did not have any formal education, and of this
total, 3.9 percent (n=3) were females compared to 4.3 percent (n=1) of males.
Based on Table 14.1.5, of the 53 percent (n=53) of the sampled who were older than 40
years, 28.3 percent (n=15) had completed university level education, 17.0 percent (n=9)
college, 34.0 percent (n=18) high/secondary, 18.9 percent (10) primary and 1 percent had
no formal education. Generally, in the age cohort 20 to 30 years, males had a higher rate
of completion of high/secondary level school and university level education (50% and
274
50% respectively) compared to females (high - 42.9% and secondary -28.6%). On the
other hand, females had higher completion rate than males in respect to college level (i.e.
people beyond 40 years) and primary (i.e. for people whose ages range from 31 to 40
years).
Table 14.1.6: Income distribution of respondents
Income (in $) Frequency Percentless than 20,000 20 20.0
20,000 - 39,999 20 20.0 40,000 - 59,999 18 18.0 60,000 - 79,999 8 8.0 80,000 - 99,999 10 10.0 100,000 - 119,999 5 5.0 120,000 19 19.0
Less than 69 percent (n=68) of the respondents received income that was lower than
$60,000 per month, with 20 percent (n=20) of them receiving less than $20,000 monthly
and same percent were earning between $20,000 and $39,999 monthly. The median
wage for the sample was between $40,000 to $59,999 with less than 25 percent of the
respondents received incomes which were higher than $100,000 on an average each
month (See, Table 14.1.6)
275
PARENT ATTITUDE TOWARD SCHOOL
Table 14.1.7: Parental Attitude toward SchoolDetail Frequency Percent
Strongly Disagree 45 45Disagree 39 39Undecided 12 12AgreeStrongly Agree
45 5.0
4
Total 100 100
Parental attitude toward the school was generally extraordinarily poor. Based on Table
14.1.7, approximately 84 percent (n=84) of the respondents reported a negative attitude in
respect to the school. Of the 100 respondents, 45 percent viewed the school in an
extremely negative manner compared to 5 percent who reported on the positive extreme.
Only 9 percent (n=9) of the interviewees saw the school in a positive light, with 12
percent (n=12) being unsure (“undecided”).
276
PARENT INVOLVING SELF
Table 14.1.8: Parent Involving SelfDetail Frequency Percent
Strongly Disagree 1 1Disagree 21 21Undecided 47 47AgreeStrongly Agree
431
431
Total 100 100
From the findings in Table 14.1.8, 31.0 percent (n=31) of the respondents reported that
they were involved themselves in the educational well-being of their children. A startling
finding was the high percent of sampled population who indicated that they were
“unsure” of an involvement of self in Parent Teacher Association meetings, assisting
their children with assignment, communicating with their children on school work and
other educational activities. Twenty-two percent (n=22) of the respondents indicated that
they were not involved in the educational development of their children, with 1 percent
reporting that they were absolutely not personally not involvement in the educational
development of their children.
277
SCHOOL INVOLVING PARENT
Table 14.1.9: School Involving ParentDetail Frequency Percent
Strongly Disagree 8 8Disagree 45 45Undecided 33 33AgreeStrongly Agree
140
140
Total 100 100
When the respondents were asked about the schooling involving them in school
activities, 53 percent (n=53) reported no with 8 percent (n=8) of them indicating an
absolute no. Only 14 percent (n=14) of the sampled population cited that they were
invited to be involved in the school’s apparatus with 33 percent (n=33) being unsure of
any such demand. Generally, the sampled population (53%) is reporting that there is a
gap between themselves and the school, with the school requesting little of their
involvement in the educational process of their children.
278
MODEL
Table 14.1.8: Regression Model SummaryDetails Beta Coefficient
Constant 68.751
Dummy Primary School Level Education -22.747*
Dummy High School Level Education -19.995*
Dummy University Level Education. -5.488*
Dummy Income less than $20,000 -12.430*
Dummy Income (1= $40K - $59,999) 7.20*
Dummy Income (1=>$120,000) -6.038*
Dummy Gender (0= males) -4.969*
Dummy Remarried (0= other) -6.009*
Dummy Parent Attitude towardSchool ( 0= negative)
8.737*
Dummy School involving parentsSchool ( 0= low)
-5.183
n 195
R .686
R2 .471
Standard Error 10.19
F statistic 16.378
ANOVA (sign.) 0.000
Model [ Y= β0 + β1x1 +…+ ei ] - where Y represents Academic Performance of the students, β0
denotes a constant, ei means error term and β1 indicates the coefficient of dummy primary level education * x1 where represents the variable primary level of education to βi and xi
* Significant at the two-tailed level of 0.05 (see Appendix V)
279
The findings in Table 14.1.8 (see above) revealed that primary, high and university level
education, gender of respondents, parent attitude towards school, school involving
parents, low income (i.e. income below $20,000), income in excess of $120,000 along
with being remarried are determinants of students’ academic performance. The
relationship between the independent variables (i.e. the determinants) and the dependent
variable (i.e. academic performance) is a statistical one (as the ρ value was less than
0.05). The causal relationship was a relatively strong one (i.e. Pearson’s Correlation
Coefficient = 0.686). Furthermore, approximately 47 percent of the variation in students’
academic performance is explained by a 1 percent change in the determinants. This
means that the regression model explains 47 percent of the total variation in students’
academic performance.
As shown in Table 14.1.8, the regression model, Testing Ho: β=0, with an α =
0.05, indicates that the linear model provides a good fit to the data based on the F value
of (1,700.74, 103.85) 16.378 with a p < 0.05 (p = 0.000).
Generally, without the determinants being held constant, a student will score
68.75 percent on his/her examination. However, if the student’s parent had only
completed primary level education he/she score will decline by 22.75 percent, and if the
parent had completed high/secondary school his/her child score will reduce by 20 percent
compared to a decrease of 5.5 marks if the parent had completed university level
education. Embedded within this finding is the contribution of parents with university
level education compared to other levels of education on a child’s academic performance.
Issues such as income, gender, remarried guardians/parents and school involving
the parents were discovered to decrease students’ performance. From Table 14.1.8, with
280
all other things being held constant, a child’s academic score will decrease by 6 percent if
his/her parent/guardian is remarried, a 5 percent fall in student’s score if school involves
the parents, a reduced score if the parent income is more than $120,000 or less than
$20,000 per month. Another reduction in a child’s score is attributable to the
guardian/parent being female (i.e. approximately 5%). Subsumed in this finding is that
the students with a male parent/guardian score 5% more than children with female
parents/guardians.
The findings further revealed that students’ whose parents have a positive attitude
toward school will score approximately 9% more compared to parent who have a
negative attitude toward the school. Concurrently, a child whose parent/guardian
received between $40,000 and $60,000 per month will score 8.7 % more than students
whose parents/guardians’ income is more $60,000 or less than $40,000. It should be
noted that parents whose incomes are high or lower than $40,000 score approximately
100 % less than children who guardian received $40,000 to $59,999 monthly.
In addition to those variables which were found to be statistically significant (i.e.
ρ value less than 0.05), some issues that initially were entered into the regression model
were discovered to be statistically not significant (i.e. ρ value > 0.05). These factors are
employment status; college trained parents; parents with no formal education; parents
whose income were $20,000 to $39,999, $40,000 to $59,999, $100,000 to $119,999;
divorced, married and single parents and parents involving themselves in their children
educational programme. Hence, the determinants of students’ academic performance of
this sample reads: Students’ Scores = 68.751 + (-22.7) * Parents’ Primary Level
Education + (-20.0) * Parents’ Secondary Level Education + (-5.5) * Parents’ University
281
Level Education + (-6) * Parent who are remarried + (-5.2) * School Involved Parents
(0=low involvement) + (8.8) * Parent Attitude toward school (0=Negative) + (-12.4) *
Parent whose income (less $20,000) + (7.3) * Parent whose income ($40, 000 - $59,999)
+ (-6.0 ) * Parent whose income (beyond $120,000) + (-5.0) * Dummy gender (0=
males).
282
CHAPTER 15
Hypothesis 12: People who perceived themselves to be of the lower social status (i.c. class) are more likely to be in-civil than those of the upper class.
Based on the level of measurement of the variables – dependent (DV), ordinal and the
independent (IV), ordinal. The social researcher has the option of using either (1)
Spearman rho or (2) Cross-tabulations – Chi Square Analysis.
Table 15.1.1: Correlations
Social Status IncivilitySpearman's rho Social Status Correlation
Coefficient1.000
Sig. (2-tailed) . N 216 Incivility Correlation
Coefficient.512(**) 1.000
Sig. (2-tailed) .000 N 216 216
** Correlation is significant at the 0.01 level (2-tailed).
Based on Table 15.1.1, there is a statistical association between incivility and ones
perceived social status (using correlation coefficient of 0.512, Ρ value = 0.001< 0.05).
Furthermore, a positive correlation coefficient, 0.512, indicates that a direct relationship
exists between the DV and the IV. This implies that the higher one goes up the ranked-
ordered social class, the more likely that the individual is less uncivil, which can be
simply put as those within the lower social status are more ‘uncivil’ than those further up
the social ladder. This statistical association is a moderate one using Cohen and
Holliday’s classifications of statistical relationships (Cohen and Holliday 1982). In
283
addition, 26.214% (i.e. cc2 * 100 – 0.512 * .0152 * 100) of the variation in the DV,
incivility, is explained by a change in ones social status.
This could have been analyzed using Chi-Square instead of Spearman’s rho,
based on Chapter 1. Thus, using the former gives this set of analysis.
Table 15.1.2: Cross Tabulation between incivility and social status
Incivility * Social Status Crosstabulation
37 1 12 50
74.0% 2.0% 24.0% 100.0%
37.0% 1.0% 100.0% 23.1%
17.1% .5% 5.6% 23.1%
59 15 0 74
79.7% 20.3% .0% 100.0%
59.0% 14.4% .0% 34.3%
27.3% 6.9% .0% 34.3%
4 86 0 90
4.4% 95.6% .0% 100.0%
4.0% 82.7% .0% 41.7%
1.9% 39.8% .0% 41.7%
0 1 0 1
.0% 100.0% .0% 100.0%
.0% 1.0% .0% .5%
.0% .5% .0% .5%
0 1 0 1
.0% 100.0% .0% 100.0%
.0% 1.0% .0% .5%
.0% .5% .0% .5%
100 104 12 216
46.3% 48.1% 5.6% 100.0%
100.0% 100.0% 100.0% 100.0%
46.3% 48.1% 5.6% 100.0%
Count
% within Incivility
% within Social Status
% of Total
Count
% within Incivility
% within Social Status
% of Total
Count
% within Incivility
% within Social Status
% of Total
Count
% within Incivility
% within Social Status
% of Total
Count
% within Incivility
% within Social Status
% of Total
Count
% within Incivility
% within Social Status
% of Total
1=Strongly agree
2=Agree
3=Disagree
4=Strongly disagree
8
Incivility
Total
1=Lower(Working)
Class2=Middle
Class3=UpperMiddle
Social Status
Total
284
Chi-Square Tests
178.160a 8 .000
203.720 8 .000
27.424 1 .000
216
Pearson Chi-Square
Likelihood Ratio
Linear-by-LinearAssociation
N of Valid Cases
Value dfAsymp. Sig.
(2-sided)
8 cells (53.3%) have expected count less than 5. Theminimum expected count is .06.
a.
Symmetric Measures
.672 .000
.620 .089 7.662 .000
.512 .078 8.709 .000c
.357 .082 5.594 .000c
216
Contingency CoefficientNominal by Nominal
Gamma
Spearman Correlation
Ordinal by Ordinal
Pearson's RInterval by Interval
N of Valid Cases
ValueAsymp.
Std. Errora
Approx. Tb
Approx. Sig.
Not assuming the null hypothesis.a.
Using the asymptotic standard error assuming the null hypothesis.b.
Based on normal approximation.c.
From the Chi-Square Tests table above, there is a statistical association between incivility
(DV) and the perceived social class (IV) of respondents (χ2 (8) = 178.16, ρ value =
0.001< 0.05). In order to establish strength, direction and magnitude of the relationship,
we need to use the Symmetric Measures Table. Based on this Table, given that the
variables are Ordinal, DV and Ordinal, IV, the statistical value which should be used is
the Gamma valuation, 0.620. This value denotes (1) a positive relationship between the
DV and IV; (2) the associate is a moderate one using Cohen and Holliday’s38,39 figures,
and (3) 38.44% of the variation in incivility is explained a by change in ones perceived
social class.
38 Very low, < 0.19; Low, 0.20 – 0.39; Moderate, 0.40 – 0.69; High 0.70 – 0.89; Very High 0.9 – 1.0.39 Bryman and Cramer modified Cohen and Holliday’s work by using Very weak, < 0.19; Weak, 0.20 –
0.39; Moderate, 0.40 – 0.69; Strong 0.70 – 0.89; Very Strong 0.9 – 1.0 (Bryman and Cramer 2005, 219.
285
16. Data Transformation
In order for me to provide an integrative understanding of how the following are possible: Recoding
Dummying variablesAveraging ScoresReverse coding
I will use the Questionnaire below
286
QUESTIONNAIREADVANCED LEVEL ACCOUNTING SURVEY 2004
SECTION 1 CHARACTERISTICS (for all persons)
287
1.1 Is …male or female? О Male О Female
1.2 What is your….at last birthday?
1.3 Where do you live? ____________
1.4 In response to Q1.3, Is the home
О Owned О Rented О Leased О Unsure О Other(specify) ________
1.5 What is your father’s highest level of education?
О No formal education
О Primary/Preparatory school
О All-Age school
О Secondary school
О Vocational/skill training
О Some professional training
О Tertiary (Undergraduate)
О Tertiary (Post-graduate
1.6 What is your mother’s highest level of education?
О No formal education
О Primary/Preparatory school
О All-Age school
О Secondary school
О Vocational/skill training
О Some professional training
О Tertiary (Undergraduate)
О Tertiary (Post-graduate
1.7. What is your perception of your parent(s)/guardian(s) social class?
О Lower class
О Lower middle class
О Middle middle class
О Upper middle class
О Upper class
1.8 Are you currently living with?О Mother only
О Mother and father
О Father only
О Mother and Step-father
О Father and Step-mother
О Other ___________________
288
1.9 Which of the following affect you?
О Migraine О ArthritisО Psychosis О AnxietyО Sickle cell О DiabetesО AsthmaО Heart diseaseО Hard drug addiction –
marijuana, heroine, crack, etc.
О depressionО hypertensionО fit (epilepsy)О numbness of the hand(s) О NoneО Other ________________
1.10 If you answer to Q6.1 is YES, how often in the last three (3) months?
О Always (7-12 weeks)О Sometimes (3-6 weeks)О Occasionally (1-2 weeks)О Rarely (0 to <1 week)О Never (0 week)
1.11 If you answer to Q1.10 is YES, how often in the last six (6) months?
О Always (4-6 months)
О Sometimes (2-3 months)
О Occasionally (1 month)
О Rarely (0 to <4 weeks)
О Never (0 week)
1.12 Do any of your close family member(s) suffering from a major illness?
О Yes О No
1.13 If your response to Q1.12 is Yes, Are you close this family member?
О Yes О No О not really
1.14 If your response to Q1.12 is Yes, How frequently in the last three (3) months?
О Always (11/2 - 3months) О Sometimes (< 3 weeks but > 5weeks)
ОUnsure О Occasionally (less than two weeks) О Never
289
SECTION 2 QUALIFICATION
2.1 What were your grades in the following course(s), specify: tick appropriate response
Subject CXC -General
Grade O’Level Grade A/O Grade
EnglishLanguage
N/A N/A
EnglishLiterature
N/A N/A
Mathematics
General Paper orCommunication Studies
N/A N/A N/A N/A
Principles ofAccounts N/A N/A
290
SECTION 3 ACADEMIC PERFORMANCE
3.1 In Advanced Level, what were your last two (2) tests scores over the past six (6) months? (1) _______________________
(2) _______________________
3.2 In A’ Level Accounting, what were your last two (2) assignments scores over the past six (6) months.
(1) _______________________
(2) _______________________
3.3 What was your lowest score on an Advanced Level Accounting test in the last three (3) months?
(1) __________________________
3.4 Comparing this term to last term, How was your academic performance in A’ Level
Accounting О Better
О Same О Worse
291
SECTION 4 CLASS ATTENDANCE
Read each of the following options, then you are to select the numbered response that best express your choice.
KEY
1- S trongly Disagree; 2 – Disagree; 3 – Neutral; 4 – Agree; 5 – Strongly Agree
1 2 3 4 54.1 I enjoy attending A’ Level Accounting classes 4.2 A’ Level Accounting classes are boring so why
should I attend as this as will destroy my psyche for the other classes
4.3 My Accounts teacher knows nothing so I donot attend
4.4 I attend all the A’ Level Accounts classes in the past because the teacher uses techniques that allow us to grasp the principles of the subject matter
4.5 Whenever its time for A’ Level Accounts classes I become nauseous so I go home
4.6 I wished all the other disciplines, courses, were taught like that of the accounts, I like being there
4.7 I oftentimes wished the A’ Level Accountsclasses never end
4.8 My A’ Level Accounts teacher has impactedpositively on my self concept
4.9 The physical layout of the classroom in which A’ Level Accounts is taught turns me off, so I do not attend
4.10 I will not waste precious time attending A’ Level Accounts classes, when I can spendthis time on other subject(s)
292
SECTION 5 DIETARY INTAKE
5.1 How often do you consume the following per week? Tick your choices
Frequency Breakfast Lunch DinnerSeven timesSix timesFive timesFour timesThree timesTwo timesOne timeNever
SECTION 6 DAILY FOOD INTAKE
6.1 What is your normal food intake for each day; tick your choice(s)? ITEM(S) Pineapple/orange/banana Chicken and parts
Apple/beat root/Grape
Fish, other meats
Carrot Butter/margarine
Cabbage/water Pear
Sweet sop/soar sop Coconut
Turnip/salad/tomatoes Ackee
String beans/string peas/ green peas/broad beans/gongo - PEAS
Rice/oats
Peanuts/cashew Flour/ wheat bread/wheat biscuits
Milk/eggs Cornmeal/wheat/corn
Yam Green bananasIrish/sweet potato(es) Dasheen
293
SECTION 7 INSTRUCTIONAL RESOURCES
Read each of the following options, then you are to select the numbered response that best express your choice.
KEY
1- S trongly Disagree; 2 – Disagree; 3 – Neutral; 4 – Agree; 5 – Strongly Agree
1 2 3 4 57.1 I will not buy an A’ Level Accounting text
7.2 I have a minimum of two (2) of the prescribed reading materials in Accountings
7.3 I am very aware of the required texts needed for the examination in Accounting but I have none
7.4 I visit the library at least once a week in orderto borrow resource materials in Accounting
7.5 The libraries provide pertinent textbooks and journal in Accounting that I use in my preparation of the subject
7.6 My teacher provides little notes on each topic which cannot be used to problem-solve examinations questions
7.7 I have Examiners’ Reports on Advanced level Accounting
7.8 I have never read an Examiners’ Report onAdvanced Level Accounting
7.9 Generally, I revise my notes daily7.10 I have a copy of the Advanced Level Accounting
syllabus 7.11 In the last six (6) months, I have not read the
Advanced Level Accounting Syllabus7.12 Generally, my teacher provides all the solutions to
practiced papers and other questions solved in class
7.13 Generally, I frequently use my textbooksin problem-solving questions
7.14 I am not comfortable using a calculator
294
SECTION 8 SELF-CONCEPT
Read each of the following options, then you are to select the numbered response that best express your choice.
KEY
1- S trongly Disagree; 2 – Disagree; 3 – Neutral; 4 – Agree; 5 – Strongly Agree
1 2 3 4 58.1 I am proud of my present body weight 8.2 I am glad to know I look this good/attractive 8.3 I would like to take plastic surgery to alter a few
aspects of by body 8.4 I am always upset at the accomplishment of others 8.5 I am never angry in being around someone who 8.6 speaks highly of himself/herself 8.7 I am proud of my present body weight 8.8 I am glad to know I look good 8.9 I would like to take plastic surgery to alter a few
aspects of by body 8.10
295
SECTION 9 PHYSICAL EXERCISE
Read each of the following options, then you are to select the numbered response that best express your choice.
KEY
1- S trongly Disagree; 2 – Disagree; 3 – Neutral; 4 – Agree; 5 – Strongly Agree
1 2 3 4 58.1 I enjoy working out (i.e. physical exercise) at least
once per week
8.2 I do not understand why someone wouldwant to become sweaty by exercising
8.3 I just enjoy being physically active 8.4 I do not see the importance of participating in any
form of physical exercise, as other activities appear more important Physical exercising is a crucial aspect of my health programme
8.6 Although physical exercise is good for the Human body, I do not participate because On completion I want to sleep
Now that we have come to the end of this exercise, I would like to expend my deepest appreciation for your co-operation and involvement in this data gathering process – THANK YOU!
296
RECODING A VARIABLE
From the Questionnaire, I will be recoding – Question 4 “What is your mother’s highest level of education?”
In SPSS, Question 4 was coded as
1= Primary/All Age2=Junior High3=Secondary/High4=Technical high5=Vocational6=Tertiary7=None
In order to know how the variables were coded, we need to use the variable view window
297
Instead of the seven categories, I would like to have – 5 categorization – 1=No formal Education; 2= Primary to Junior High (including All Age); 3=Secondary (including Technical High schools): 4= vocational and 5=Tertiary.
Step 1:select TransformStep 2:
select Recode
Step 3:select Into Different variables
298
Step 4:Identify the variable, in this case Education of parents
Use the arrow to take this variable into Input Variable
299
This results from Step4:
q4 is now the variable selected to be recoded
300
Step5Use whatever you want to identify the variable by
301
Step 6:
Select change, which gives this dialogues box‘Recode into Different Variables:
302
In order for the process to be effective, we need to know the old codes following by ‘how we would like the new codes to be. Thus, see the example here:
Old Codes1= Primary/All Age2=Junior High3=Secondary/High4=Technical high5=Vocational6=Tertiary7=None
New Codes1= None2=Primary/All Age - Junior High3=Secondary/High to Technical high4=Vocational5=Tertiary
In order to convert the variables, place the value for the old variable on the Left-hand-side followed by the new value on the right-hand-side, then add (see below)
303
To convert the old 7 to 1, then select add to complete this stage
304
To convert a range of values (for example 1 and 2) – see below
To convert a range of values; step 1:select range
Step 2:Place the lowest value first followed by the last value
Step 3:Place the new value here
Then, do not forget to choose add
305
This is the result, and then continues
306
Having selected continue, this is what results, then choose OK or Paste
307
The next step, is to label the variables
308
Select variable view, then:
Select the left of the values for the recoded variable
309
Step1:
Place the new value here, for example 1
Step 2:
Place 1, then equal, followed by the label of the value, example ‘no formal education’
310
This is ‘what it looks like’
311
Select OK
312
This is to verify what has been done:
313
314
315
Dummying a Variable. Creating a dummy variable apply this rule (k – 1), where k denotes the number of categories. Hence, for this case (2 – 1), which means that we can only dummy once. Where one of the two (males or females) will be given 1 and the other 0.
Initially, these are the code
316
317
Use a label, which will be used identify the dummy variable
318
Select label, this gives ‘compute variable Type and Label
319
Identify the variable you seek to label 1, and implied 0 is not stated
320
Step 1:Select the variable to be dummied, e.g. gender
Step 2:Use the arrow to take it across
Step 3:
this results
321
Step 4:
Select =, then 2, which we want to be saying I and males 0
Choose either OK or Paste
322
Following the OK or the Paste, this results
323
Now, let use see if this process was done and if it as we intended (Descriptive statistics for the dummy variable gender):
324
325
326
Before dummying the variable, e.g. gender, in which we will make 1=female
327
After the process to dummy the variable gender:
328
Dummying a variable that has more that two categories
The example that we will use here is educational level, which has four categories – (1) No formal education; (2) Primary or Preparatory level education,; (3) Secondary level education and (4) Tertiary (or post-secondary) level education.
Step 1 – In order to know the number of dummy variables that are likely to result from this initial variable (educational level), we need to use the formula – k -1. In the formula, k represents the number of categories that constitute the variable education. In this example, if there are categories. Thus, (k-1 = 4-1=3), the number of dummy categories that are possible are 3. It should be noted here, that one of the category which constitute the initial variable educational level will be used as the reference group. The referent unit will be determined based on literature.
Step 2: In this, let us assume that we are seeking to the relationship between educational level of respondents and their wellbeing. Wellbeing is a continuous variable and so, in order to include education within the linear regression model it must be a dummy measure. Therefore, this is what it should like:
Educational levelEdulevel1 1=Primary, 0=Other or OtherwiseEdulevel2 1=Secondary, 0=Other or OtherwiseEdulevel3 1=Tertiary, 0=Other or Otherwise
The reference group is ‘no formal education. The rationale for this choice is the literature that has established that people with more
education have a greater wellbeing. As such, the group that is best suited to be the referent group is ‘no formal education. (Would you like to see how this is done in SPSS? See, below)
329
Reverse Coding
Sometimes within the research process, as is the case in the Questionnaire above - using Section 9, the researcher may want to create a single variable, for example in this case Physical Exercise, from a number of sub-questions around a particular topic. However, he/she is hindered by the differences in direction, for example take Q8.1 – this is a positive statement whereas Q8.2 is negative, thus they cannot be summed as they are not compatible. What is done in such instance is called reverse coding. The researcher will decide of the two directions, which he/she is more comfortable working with. In this case, I will choose the positive, which include Q8.1; Q8.3; Q8.4 with Q8.2; Q8.5; and Q8.6 being negative. Having decided to work with the positive, I must now reverse the codes for Q8.2; Q8.5; and Q8.6, in an effort to attain compatibility. (see the process below, the SPSS approach)
SECTION 9 PHYSICAL EXERCISE
Read each of the following options, then you are to select the numbered response that best express your choice.
KEY
1- S trongly Disagree; 2 – Disagree; 3 – Neutral; 4 – Agree; 5 – Strongly Agree
1 2 3 4 58.1 I enjoy working out (i.e. physical exercise) at least
once per week
8.2 I do not understand why someone wouldwant to become sweaty by exercising
8.3 I just enjoy being physically active 8.4 I do not see the importance of participating in any
form of physical exercise, as other activities appear more important Physical exercising is a crucial aspect of my health programme
8.6 Although physical exercise is good for the Human body, I do not participate because On completion I want to sleep
330
331
Step 1:
select – Transform, Recode, and Into Different Variables
332
Step 2:
Select the variables, which are needed for reverse coding – (the eg here, q8.2; q8.5, q8.6
333
Step 3:
Rename the new variable
Step 4:
State what will be done – reverse coding for q8.2, etc.
Step 5:
Then, select change, each time in step 4 afterq8.2; q8.5, and q8.6
334
Following the completion of this (step 5) the process will look like this
Step 6:
Select Old and New values
335
In order for the researcher to complete the process, he/she needs to know ‘how the variables were coded, initially’ – for example 1- Strongly Disagree; 2 – Disagree; 3 – Neutral; 4 – Agree; 5 – Strongly Agree.Reverse coding means that
Old values New values
1= Strongly Disagree 5=strongly disagree 2 – Disagree 4=disagree 3 – Neutral 3 = Neutral4 – Agree 2=Agree5 – Strongly Agree 1= strongly agree
(See how this is done in SPSS, below)
336
Step 7:Select the old value 1 (this is place in the left-hand window; then write the new value 5, in new value; repeat this process for each base on the old and new values, which are written above
Add is selected, each time a convert is executed
Step 8:Select continue
337
Step9:Select OK or Paste
338
SUMMING CASES:
The issue of summing variables must meet two conditions:(1) Variables must be similar, and(2) If they are not, then use reverses coding
Note: Having reversed the codes for q8.2, q8.5 and q8.6; it now follows that all 6 questions (q8.1 to q8.6) are positive. (see the SPSS steps below)
1 2 3 4 58.1 I enjoy working out (i.e. physical exercise) at least
once per week
8.2 I do not understand why someone wouldwant to become sweaty by exercising
8.3 I just enjoy being physically active 8.4 I do not see the importance of participating in any
form of physical exercise, as other activities appear more important Physical exercising is a crucial aspect of my health programme
8.6 Although physical exercise is good for the Human body, I do not participate because On completion I want to sleep
339
Summing cases in SPSS
(Note in order to sum the cases, we should use those cases such as q8.1, q8.3 and q8.4, which were not reversed along with the reversed once)
Step 1:Select – Transform, and then Compute
340
On carrying out step1, this dialogue box appears
341
Step 2:Type a word or phrase that will represent the combined variable (in this case Total_ ph)
Step 3:Write the label for the event
Step 4:Select continue to move to the next process
342
Step 5:
look for the mathematical operation, sum
Step 6:Select the arrow
Step 6, takes it into the Numeric Expression box (see that output in Step 7, below
343
Step 7:
Having select the arrow, it goes to Numeric Expression- SUM(?,?)
The question mark should be replaced by each variable, followed by a comma. Note no comma should be placed after the last variable
344
Step 8:
Select those variables, which were not recoded in the first class but are apart of the computation of the new composite variable
Step 9:Choose those variables that were reversed coded, and are needed for the composite variable
Step 10:select either OK or Paste
345
This is the final product of step 10
346
What should be done, now is to ‘run’ the frequency (i.e. the descriptive statistics for this new variable, Index of Physical Exercise)
This is the newly created variable, Index of Physical Exercise from the summing and reverse coding processes
What the researcher has created in an index (or a metric variable), which can be reduced by recoding
347
DATA REDUCTION (USING A SUMMED VARIABLE)
The researcher should note that there were five categorizations, from 1= strongly disagree to 5=strongly agree. Thus, to reduce the Index (the summed variable) into five groupings, we should – do a count of the number of values, which constitute the Index. The example here is 16. The approach that I prefer is to divide the 16 by 5, which gives 3.2. This 3.2 indicates that each category should contain a minimum of three values, with one group housing more than three. Before this process can be executed, the researcher should be aware of what constitutes the least value and the largest number. Based on this case, the standard that should be applied is now the values were coded, using the positive coding (i.e. 1= strongly disagree, 2= disagree, 3=neutral, 4= agree and 5=strongly agree). This means that from 5 to 13 would be 1 or strongly disagree in keeping with the coding scheme; 14 to 16, 2 – disagree; hence, 17 to 19, is 3 i.e.– neutral; from 20 to 22 is 4 or agree and strongly agree would have the following numbers – 23, 24, 25, and 27. (see the SPSS process below).
348
DATA REDUCTION (Having computed by hand the categories, use SPSS to recode the new categorization – this will see the variable remaining as Ordinal)
To recode, the calculate values –
Step 1:
select - Transform, Recode, and Into Different Variables
349
Step 2:
Look for the composite variable, which is in the left-hand side dialogue box
Step 3:
Select this arrow, to have the variable placed into the box marked input variable –Output variable box
350
step 4:
write a word for the new variable
step 5:
optional – describe for labeling purposes
step 6:
select change
step 7:
select old and new values, for the recoding exercise
351
Step 8:
Select range
352
step 9:
Based on the index, the old value from the calculation would be from 5 to 13, etc.
Step 10:
Select 1 as the new value, which represent strongly disagree
Step 11:
Having selected the old and new values, then select add to complete the process each time
353
step 12:
Do the same process for all other values, system missing after the last category (5= 23 to 27)
step 13:
Select continue
354
step 14:
go to variable view, in order to label the new variable, then values, followed by the labeling in the Values Label box
355
step 15:
select OK
356
Final stage:
Run the descriptive statistics for the new ordinal variable
357
GOLSSORY
Bivariate r – Bivariate correlation and regression assess the degree of association between two continuous variables (i.e. one independent, continuous and a continuous dependent)
Concept – This is an abstraction that is based on characteristics of a perceived reality
Conceptual (or nominal) definition – this means a statement that encapsulates the particular meaning of a word or concept in a research
Correlation - “Correlation is basically a measure of relationship between two variables (Downie and Heath 1970, 86)
Correlation - “Correlation is use to measure the association between variables” (Tabachnick and Fidell 2001, 53)
Dependent variable – this is the variable with which the study seeks to explain
Eta – This is a measure of correlation between two variables; in which one of the variables is discrete.
Explanation – This denotes relating variation in the dependent variable to variation in the independent variable
Homoscedasticity – Homoscedasticity is a term which is usually related to normality, because when the assumption of normality is attained, in multiple regressions, the association variables are said to be homoscedastic. “For ungrouped data, the assumption of homoscedasticity is that the variability in scores for one continuous variable is roughly the same at all values of another continuous variable” (Tabachnick and Fidell 2001, 79)
Hypothesis – This is a testable statement of relationship, which is derived from a theory
Independent variable – This is the variable that is used explain the dependent variable.
Linearity – This speaks to a straight line relationship between two variables. The issue of linearity holds in Pearson’s Product-Moment Correlation Coefficient, and in multiple linear regressions. In the case of Pearson’s r, linearity is denoted by an oval shaped scatter plot between the DV and the IV. Thus, if any of the variables is non-normal, the scatter plot fails to be oval shaped. Whereas for linear regression, standardized residual when plotted against predicted values, if non-linearity is indicated whenever most of the data-points of the residuals are above the zero line or below the zero line.
358
Logistic Regression – This allows for the prediction of group membership when predictors are continuous, discrete, or a combination of the two. It is used in cases when the dependent variable (DV) is discrete dichotomous variable.
Multiple Regression – “Multiple correlation assess the degree to which one continuous variable (the dependent) is related to a set of other (usually) continuous variables (the independent) that have been combined to create a new composite variable” (Tabachnick and Fidell 2001, 18). Furthermore, it should be noted that multiple regression emphasizes the predictability of the dependent variable from a set of independent variables whereas bivariate correlation speaks to the degree of association between the dependent and the independent variable.
Nonparametric test – A statistical test that requires either no assumptions or very few assumptions about the population distribution.
Operational definition – A specification of a process by which a concept is measured or the measuring rob for a concept
Parameter – A specified number of variables to be found within a population.
Parametric test – A hypothesis testing that is based on assumptions about the parameter values of the population
Pearson’s Product-Moment Correlation, r. -“The Pearson product-moment correlation, r, is easily the most frequently used measure of association and the basis of many multivariate calculations” (Tabachnick and Fidell 2001, 53).
Reliability – This denotes the extent to which a measurement procedure consistently evaluates whatever it is to measure
5% level of significance - “With the use of multivariate statistical technique, complex interrelationship among variables are revealed and assessed in statistical inference. Further, it is possible to keep the overall Type I Error rate at, say 5%, no matter how many variables are tested” (Tabachnick and Fidell 2001, 3)
Null Hypothesis – Speaks of no statistical relationship (or association) between the variables (i.e. dependent and independent variables) that are being tested in a hypothesis.
Validity – this is the extent to which a measurement procedure measures (or evaluates) what it is intended to meaure
Variation – speaks to differences within a set of measurements of a variable
359
REFERENCES
Aitken, A. C. 1952. Statistical Mathematics, 7th ed. New York: Oliver and Boyd.Alleyne, Sylvan and Benn, Suzette L. 1989. Data collection and presentation in social
surveys with special reference to the Caribbean Kingston: Institute of Social and Economic Research.
Babbie, Earl, Halley, Fred, and Zaino, Jeanne. 2003. Adventures in Social Research: Data Analysis Using SPSS 11.0/11.5 for Windows, 5th. London: Pine Forge Press.
Babbie, Earl. 2001. The Practice of Social Research, 9th. New York, U.S.A.: Wadsworth.Behren, Laurence, Rosen, Leonard J., and Beedles, Bonnie. 2002. A Sequence for
Academic Writing. New York, U.S.A.: Longman.Bobko, Philip. 2001. Correlation and Regression: Applications for Industrial
Organizational Psychology and Management, 2nd. London, England: SAGE Publications.
Boxill, Ian, Chambers, Claudia M., and Wint, Eleanor. 1997. Introduction to Social Research with Applications to the Caribbean. Kingston, Jamaica: Canoe Press.
Bryman, Alan and Cramer, Duncan. 2005. Quantitative Data Analysis with SPSS with 12 and 13: A Guide for Social Scientists. East London, England: Routledge.
Burnham, Peter; Gilland, Karin; Grant, Wyn, and Loyton-Henry, Zig. 2004. Research Methods in Politics. New York, U.S.A.: Palgrave MacMillan.
Chicago University. 2003. The Chicago Manual of Style: The Essential Guide for Writers, Editors, and Publishers, 15th. U.S.A.: Chicago University Press.
Chou, Ya-lun. 1969. Statistical Analysis with Business and Economic Applications. U.S.A.: Holt, Rinehart and Winston.
Clarke, G. M. and Cooke, D. 2004. A basic course in Statistics, 5 th ed. New York: Oxford University Press
Cohen, Jacob and Cohen, Patricia. 1983. Applied Multiple Regression/Correlation Analysis of the Behavioral Sciences, 2nd. New Jersey, U.S.A.: Lawrence Erlbaum Associates.
Cohen, L. and Holliday, M. 1982. Statistics for Social Sciences. London, England: Harper and Row. In Bryman, Alan and Cramer, Duncan. 2005. Quantitative Data Analysis with SPSS with 12 and 13: A Guide for Social Scientists. East London, England: Routledge.
Corbett, Michael and Le Roy, Michael. 2003. Research Methods in Political Science: An Corston, Rob and Colman , Andrew. 2000. A Crash Course in SPSS for Windows. Crawford, J., and Chamber, J. 1994. A Concise Course in A-Level Statistics with worked
Examples, 3rd. London, England: Stanley Thornes.Crotty, Michael. 2003. The Foundations of Social Research: Meaning and Perspective in
the Research Process. London, England: SAGE Publications.Dale, Angela, Arber, Sara and Procter, Michael. 1988. Contemporary Social Research.
Doing Secondary Analysis. London, England: Unwin Hyman.
360
Daniel, Wayne, W. 1987. Biostatistics: A foundation for analysis in the health sciences, 4th ed. New York: John Wiley.
Dawson, Catherine. 2002. Practical Research Methods: A User-friendly guide to mastering research techniques and projects. Oxford, United Kingdom: How to Books.
de Vaus, David. 2002. Analyzing Social Science Data. London, England: SAGE Publications.
Evans, Omri I. 1992. Statistics made Simple for Students of Social Sciences and management, 2nd. Kingston, Jamaica: Packer-Evans and Associates.
Flick, Uwe. 2006. An introduction to qualitative research, 3rd. London, England: SAGE Publications.
Francis, A. 1995. Business Mathematics and statistics. London: DP Publications.Freeman, Lincoln. 1965. Elementary applied statistics. New York: John Wiley.Furlong, Nancy, Lovelace, Eugene, and Lovelace, Kristen. 2000. Research Methods and
Statistics: An Integrated Approach. U.S.A.: Harcourt College Publishers.George, Darren, and Mallery, Paul. 2003. SPSS for Windows Step by Step: A Simple
Guide and Reference 11.0 Update, 4th. U.S.A.: Pearson Education.Goode, W. J. and Hatt, P.K. 1952. Methods in social research. New York: McGraw-Hill.Hair, Joseph F. Jr., Anderson, Rolph E., Tatham, Ronald L., Black, William C. 1998.
Multivariate Data Analysis, 5th. New Jersey, U.S.A.: Prentice Hill.Hakim, Catherine. 1987. Research design. Strategies and choices in the desgn of social
research. London: Unwin Hyman.Hill, R. Carter, Griffiths, William E., Judge, George G. 2001. Undergraduate
Econometrics, 2nd. U.S.A.: John Wiley and Sons.Hogg, Robert V., and Tanis, Elliot A. 2006. Probability and Statistical Inference, 7th ed.
New York: Upper Saddle River.Hult, Christine A. 1996. Research and writing in the Social Sciences. U.S.A.: Allyn and
Bacon, Pearson Education.Jackson, Barbara B. 1983. Multivariate Data Analysis: An Introduction. U.S.A.: Richard
D. Irwin.Kendall, Philip C., Butcher, James N., Holmbeck, Grayson N. 1999. U.S.A.: John Wiley
and Sons.King, G., Keohane, R., and Verba, S. 1994. Designing Social Inquiry: Scientific
Inference in Qualitative Research. Princeton, New Jersey, U.S.A.: Princeton University Press.
Kish, Leslie. 1965. Survey Sampling. U.S.A.: John Wiley and Son.Klecka, William R. 1980. Discriminant analysis. London: Sage.Kleinbaum, David G., Kupper, Lawrence L., Muller, Keith E. 1988. Applied regression
analysis and Other Multivariable Methods, 2nd. Boston, U.S.A.: PWS-KENT Publishing.
Larson, Ron, and Farber, Betsy. 2003. Elementary statistics. Picturing the world, 2nd ed. New Jersey: Prentice-Hall.
Lee, Eun S., Forthofer, Ronald N., and Lorimor, Roanld J. 1989. Analyzing Complex Survey Data. U.S.A.: SAGE Publications.
Lewis-Beck, Michael S. 1980. Applied Regression: An Introduction. London, England: SAGE Publications.
361
Magidson, Jay ed. 1978. Analyzing Qualitative/Categorical Data: Log-Linear Models and Latent Structure Analysis, 3rd. Massachusetts, U.S.A.: Abt Associates.
Marsh, Catherine. 1988. Exploring data: An Introduction to data Analysis for Social Scientists. Oxford, United Kingdom: Polity Press and Basil Blackwell.
Maxwell, A. E. 1970. Basic statistics in behavioural research. London: Penguin Education.
McClave, James T., Benson, P. George, and Sincich, Terry. 2001. Statistics for Business and Economics, 8th. New Jersey, U.S.A.: Prentice Hall.
McDaniel, Samuel A. 1999. Your Statistics: The Easy Way. Kingston, Jamaica: Samuel A. McDaniel.
McNabb, David E. 2004. Research Methods for political Science: Quantitative and Qualitative Methods. London, England: M.E. Sharpe.
Miller, Delbert C. 1970. Handbook of research design and social measurement. New York: David McKay .
Miller, Scott A. 1998. Developmental research methods, 2nd. New Jersey, U.S.A.: Prentice-Hall.
Moser, C. A. and Kalton, G. 1971. Survey methods in social investigation. London: Heinemann Educational Books.
Nachmais, David and Nachmias, Chava. 1987. Research Methods in the Social Sciences, 3rd. U.S.A.: St. Martin’s Press.
Neuman, William L. 2003. Social Research methods Qualitative and Quantitative Approahes, 5th. U.S.A.: Allyn and Bacon.
Neuman, William L. 2003. Social Research methods Qualitative and Quantitative Approaches, 5th. U.S.A.: Allyn and Bacon.
Neuman, William L. 2006. Social Research methods Qualitative and Quantitative Approaches, 6th. U.S.A.: Pearson Education.
Norušis, Maria J. 1988. The SPSS Guide to Data Analysis for SPSS/PC+. Chicago, U.S.A.: SPSS Inc.
.Powell, Lawrence A., Bourne, Paul, and Waller, Lloyd. 2007. Probing Jamaica’s
Political Culture: Main Trends in the July-August 2006 Leadership and Governance Survey, Volume 1. Kingston, Jamaica: Centre for Leadership and Governance.
Ramjeesingh, D. H. 1990. Introduction to probability theory, 2nd ed. Kingston: CFM publications.
Ross, Sheldon M. 1996. Introductory statistics. U.S.A.: McGraw-Hill.Roundtree, Kathryn and Laing, Tricia. 1996. Writing by Degrees: A Practical Guide to
Writing Thesis and Research Papers. Auckland, New Zealand: Addison Wesley Longman New Zealand.
Rubin, Donald. 1987. Multiple Imputation for Nonresponse in Survey. New York, U.S.A.: Wiley and Sons.
Rummel, Rudolph J. 1970. Applied Factor Analysis. Evanston, U.S.A.: Northwestern University Press.
Runyon, Richard P., Haber, Audrey, Pittenger, David J., and Coleman, Kay. 1996. Fundamentals of Behavioral Statistics, 8th. New York, U.S.A.: McGraw-Hill.
Stacy, Margaret. 1969. Methods of social research? Pergamon Press.
362
Stevens, James. 1996. Applied Multivariate Statistics for the Social Sciences, 3rd. Mahwah, New Jersey, U.S.A.: Lawrence Erlbaum Associates.
Stevens, James. 1999. SPSS for Windows 8.0 supplement for Applied Multivariate Statistics for the Social Sciences, 3rd. New Jersey.: Lawrence Erlbaum Associates.
Stevens, Stanley S. 1946. “On the Theory of scales of measurement”. Science, 103: 670-680.
Stevens, Stanley S. 1958. “Measurement and Man.” Science, 127: 383-389.Stevens, Stanley S. 1968. “Measurement, Statistics, and Chemapiric view”. Science,
161: 845-856.Tabachnick, Barbara G., and Fidell, Linda S. 2001. Using Multivariate Statistics, 4 th.
MA., U.S.A.: Pearson Education.Thirkettle, G.L. 1988. Weldon’s Business Statistics and Statistical Method, 9 th. London,
England: Pitman.Thirkettle, G.L. 1988. Weldon’s Business Statistics and Statistical Method, 9 th. London,
England: Pitman.Watson, Collin J., Billingsley, Patrick, Croft, D.J., and Hutsberger, David, V. 1993.
Statistics for management and Economics, 5th. MA., U.S.A.: Allyn and Bacon.Welkowitz, Joan, Ewen, Robert B., and Cohen, Jacob. Introductory Statistics for the
Behavioral Sciences, 5th. U.S.A.: Harcourt Brace College.Wilhoit, Stephen W. 2004. A brief guide to Writing from reading, 3rd. U.S.A.: Pearson Willemsen, Eleanor W. 1974. Understanding statistical reasoning. San Francisco,
U.S.A.: W. H. Freeman.Yamane, Taro. 1973. Statistics, An Introductory Analysis, 3rd. New York, U.S.A.:
Harper and Row.
363
APPENDIX I: LABELING NON-RESONPONSES
This may be addressed in any of the two ways:
i) In the event that the variable is a single-digit, the following holds –
For ‘don’t know’ use ‘8’ or ‘-8’In the case the respondent refused to answer, use ‘9’ or ‘-9’If the interviewee used ‘not applicable’ or NAP, use 97 or ‘-97’
ii) In the event that the variable is two-digit, the following holds –
For ‘don’t know’ use ‘98’ or ‘-98’In the case the respondent refused to answer, use ‘99’ or ‘-99’If the interviewee used ‘not applicable’ or NAP, use 97 or ‘-97’
364
APPENDIX II: ERRORS IN DATA
This table should be used in order to establish correctness of a statistical decision
Table: Have We Made the Correct Statistical Decision
REALITY:
STATISTICAL RESEARCHED OUTCOME
Reject Ho Fail to reject Ho
Type I Error40
( α )
Correct Decision
( 1- α )Ho – True
(in the population)
Ho - False
(using the population information)
Correct Decision
( 1- β )Type II Error41
( β )
(See for example de Vaus 2002; Bobko 2001; Tabachnick and Fidell 2001; Willemsen 1974).
Social researcher unlike natural scientists (for example, medical practitioners,
chemists) may not understand the severity and importance of not making a Type II error
because their may not result in physical injury or mortality, but this is equally significant
in social sciences. When a social scientist (for example a pollster) make prediction of say
a particular party winning an election based on Type I error, this may be embarrassing,
when in actuality of the election proves him/her otherwise. On the other hand, if he/she
40 Type I error, α, is the probability of rejecting the null hypothesis when it is true (see for example Steven 1996, 3)41 Type II error, β, denotes the probability of accepting the Ho, when it is false (see for example, Steven 1996, 7)
365
we to fail to predict the results based on the findings, failing to reject Ho, then this is
equally disenchanting for the statistician.
Type I error may be as a result of (1) unreasonable sample size, and/or (2) the
level of the significance, α. Thus, it may be prudent for the researcher to change α
from 0.05 (5%) to 0.10 or 0.15, when the sample size is small (n
≤ 20). It should be noted that, whenever we increase α, we reduce β and vice
versa. With such a possibility, it is in the researcher’s best interest to achieve the right
balance, α and β.
Because a Type II error is so severe, if the researcher knows what this is, then
can establish the statistical power (1 – β), which is the probability of accepting the H1,
when the H0 is false. This is simply, the power of making the right decision.
Furthermore, there is an indirect relationship between the sample size and the
power. Thus, a small sample size is associated with a low power (i.e. probability of being
correct), whereas a large sample size (n ≥ 100), relates to a high power (1 – β).
366
APPENDIX III:
This research, a negative correlation between access to tertiary level education and
poverty status controlled for sex, age, union status, area of residence, household size,
and relationship with head of household, is primarily seeking to determine access to
tertiary level education based on poverty, sex, age of respondents, area of residence,
household size and educational level of ones parents. As such, the positivists’ paradigm
is the most suitable and preferred methodology. Furthermore, the study will test a
number of hypotheses by first carefully analyzing the data through cross tabulation – to
establish relationship, and then, secondly, by removing all confounding variables. After
which, the researcher will use model building in order to finalize a causal model. Hence,
the positivist paradigm is the appropriate choice. The positivists’ paradigm assumes
objectivity, impersonality, causal laws, and rationality. This construct encapsulates
scientific method, precise measurement, deductive and analytical division of social
realities. From this standpoint, the objective of the researcher is to provide internal
validity of the study, which, will rely totally on the scientific methods, precise
measurement, value free sociology and impersonality.
The study will design its approach similar to that of the natural science by using
logical empiricism. This will be done by precise measurement through statistics (chi-
square and modeling – logistic regression). By using hypotheses testing, value free
sociology, logical empiricism, cause-and-effect relationships, precise measurement
367
through the use of statistics and survey and deductive logical with precise observation,
this study could not have used the interpretivists paradigm. As the latter seeks to
understand, how people within their social setting construct meaning in their natural
setting which is subjective rather than the position taken in this research – an objective
stance. Conversely, this study does not intend to transform peoples’ social reality by way
of empowerment but is primarily concerned with unearthing a truth that is out there and
as such, that was the reason for the non-selection of the Critical Social Scientist
paradigm.
METHODS
A secondary data set (Jamaica Survey of Living Conditions – JSLC) from the Planning
Institute of Jamaica and Statistical Institute of Jamaica was used for the analysis of the
variables. Data were analyze using SPSS (Statistical Packages for the Social Sciences)
12.0. Firstly, prior to the bivariate analyses that were done, univariate frequency
distributions were done so as to pursue the quality of the specified variables. Some
variables were not used because, the non-response rate was high (i.e. >20%) or the
response rate was low (i.e. < 80%). In addition, before a number of variables were
further used in multivariate analysis, because they were skewed, first, they were logged to
attain normality. Secondly, the researcher selected ages that were greater than or equal to
17 years, because this is the minimum age at which colleges and university accept
entrants. Thirdly, the independent variables were chosen based on their statistical
significance from a bivariate analysis testing and on the literature. Next, logistic
368
regression analysis was performed in order to identify the determinants of access to
education of poor Jamaicans.
Chi-square analysis is used in determining whether any meaningful association exist
between choiced variables so that will be to construct a model in regard to the poor’s
ability to access tertiary level education. Variables that are found significant will be used
in the regression modeling equation. Table 4.(i) and 4 (ii) provides an overview of the
variable under discussion, after which cross-tabulations are presented in setting a premise
for the model in Table 4.0.
CONCEPTUAL DEFINITION
Access – According to UNESCO “Access means ensuring equitable access to tertiary
education institutions based on merit, capacity, efforts and perseverance”. For this study,
the variable of access to post-secondary education is conceptualized as the number of
persons beyond age 16 years who are attending and have attended universities and
colleges, highest level of examination passes of post 16 year-olds, number of schooling
years attending of people who are older than 16 years, and approval of loans from the
Students’ Loan Bureau (SLB). Hence, Access to tertiary education will be measured
based on: (1) one half of the highest level of examination passed and one half of the
school attending. The primary reason behind this is due to the number of missing cases
or valid responses for persons who are applied to the loans from SLB. Where less than 1
percent of the sampled population has received grants from SLB, or no more than 5
percent applied for SLB grants or loans.
369
GENERAL HYPOTHESIS
There is a negative correlation between access to tertiary level education and poverty
controlled for sex, age, area of residence, household size, and educational level of parents
SPECIFIC HYPOTHESES
Ho: Reduction in poverty does not result in greater access to tertiary level education;
Ha: Reduction in poverty results in greater access to tertiary level education;
Ho: If one is poor, gender does not influence access to tertiary level education;
Ha: If one is poor, gender influences access to tertiary level education;
Ho: Poor people who reside in rural zones have less access to tertiary level education than those in urban zones ;
Ha: Poor people who reside in urban zones have greater access to tertiary level education than those in rural zones;
Ho: there is a positive association between age of respondents and access to tertiary level education;
Ha: there is a negative association between age of poor respondents and access to tertiary level education;
Ho: there is a positive association between typologies of relationship with head of household and access to tertiary level education;
Ha: there is a positive association between typologies of relationship with head of household and access to tertiary level education;
370
Ho: there is a direct relationship between increasing household size and access to tertiary level education;
Ha: there is an indirect relationship between increasing household size and access to tertiary level education;
OPERATIONALIZATION AND DATA TRANSFORMATION
DEPENDENT VARIABLE
Access to tertiary level education: First, two variables are used to construct this variable
(i.e. highest examination passed, b24, and school attending, b21). Secondly, highest
examination passed is transformed into two categories – (1) access - 3+ CXC passes and
beyond are considered to be matriculation requirement for some tertiary level institution,
and (2) no access. School attending is categorized into (i) none tertiary (i.e. secondary
level and below) and (ii) tertiary (i.e. vocational institutions, other colleges and
universities. Thirdly, a summative function is used to convert the two named variables
and then finding one half of each. Finally, the indexing technique is used to finalize the
variable, access to tertiary level education. Despite the importance of grants from
Students’ Loan Bureau (SLB), the response rate is less than 6 percent, d10b8, in one
instance and in another less than 2 percent, d10b8. With this being the case, loans and-or
grant from the SLB are not used in this study because of the non-response rate of in
excess of 94 and-or 98 percent.
371
INDEPENDENT VARIABLES:
Part B, question 21 “What type of school did… [Name] ….last attends. This is an ordinal variable which when recoded was given a value of “0” for primary education, “1” for secondary and a value of “2” for tertiary level education;
Popquint: This ordinal variable dealt with the five (5) quintiles; poverty is recoded as Poor for quintiles 1 and 2, Lower Middle Class for quintiles 3, Upper Middle Class 4, and Rich for quintile 5. Following this, these are dummied for the regression analysis;
The variable Union Status is a nominal variable, given to question 7 on the Household Roster; it is grouped as was (see Appendix I) in addition to none being included as apart of single. After which each option is dummied for the purpose of the linear regression modeling;
Household size is logged in order to remove some degree of its skewness for regression;
Area: Initially this variable is a nominal one which reads: Kingston Metropolitan Area, Other Towns, Rural and 4 and 5. First, from the frequency distribution there were two categories 4 and 5 that are that the researcher placed into Kingston Metropolitan Area (group 1). Following this process, each of the response was dummied in order for appropriateness in the regression model. Where for KMA “1” denotes KMA and “0” other localities; for Other Towns, “1” represents Other Towns and “0” indicates any other area of residence; for Rural – “1” means rural zones and “0” implies residence outside of the rural classification;
From the Household roster, Round 16, the question, Sex, dichotomous variable) (1) Male, (2) Female, was recoded as Gender, (0) Female (1) Male;
The variable relationship to head of household is a nominal variable with the following categorization: Head, spouse, child of spouse, great grand child, parent of head/spouse, other relative, helper/domestic and other not relative. The variable relationship to head of household, relatn, is dummied for the reason of the regression analysis. The dummy is for each category- where for example
i) head of household – “1” for head and “0” for not head;ii) spouse – “1” for spouse and “0” for not spouse;iii) child of spouse – “1” for child of spouse and “0” for not; iv) great grand child – “1” for great grand child and “0” for not;v) parent of head/spouse – “1” parent of head/spouse and “0” for not;vi) helper/domestic – “1” for helper and “0” for not;vii) other not relative – “1” for other not relative and “0” for not.
372
Age: From the age restriction of tertiary institution on its entrants, the researcher selects the minimum age of 16 years in order to construct an access model of tertiary education. With this complete, the variable is logged because of its skewness. The age variable is people’s ages from 16 years onwards.
The interval variable, Age, located on the Household Roster, is logged (i.e. natural log) in order to reduce its skewness for the multiple linear regression model.
373
APPENDIX IV: EXAMPLE OF AN ANALYSIS PLAN
The Statistical Packages for the Social Sciences (SPSS) was used to analyze the data.
Cross tabulations was be used to ascertain the relationship between the dependent and the
independent variables. The method of analyses was Pearson’s correlation testing that
determine if any relationship existed between the variables. Contingency coefficient was
be used to determine the strength of any relationship that may exist between variables.
The level of significance used is alpha=0.05, at the 95 percent confidence level (CI).
374
APPENDIX V: ASSUMPTIONS IN REGRESSION
Regression Model:
Parameter (population)
Yi = α + β1X1 + β2X2 + β3X3 + β4X4 + β5X5 + β6X6+ …+ βnXn + Єi
Statistic (sample)
Yi = a + b1X1 + b2X2 + b3X3 + b4X4 + b5X5 + b6X6+ …+ bnXn + ei
In order to use ‘a’ and ‘bs’ to accurately infer of the true population values, α, β, the
following assumptions will be made of ‘a’ and ‘bs’:
(Note: α or a denotes a constant; β1 … βn – where B1 refers to the coefficient of the variable X1 and like).
Assumptions of regression
1 No specification error(a) the relationship between Xi and Yi is linear;(b) no germane independent variables are exclusive from the model;(c) no irrelevant independent variables were included
2 No measurement error – the IVs and DV are accurately measured;
3 Assumptions in regard the error term:
zero mean E(Єi) = 0 – the expected value of the error term E(Єi), for each observation, is zero;
Homoskedasticity E(Є2i) = 62 – the variance of the error term is construct
for all values of xi; no autocorrelation E(Єi Єj) = 0, (i≠j) – the error terms are uncorrelated; the independent variable is uncorrelated with the error term E(Єi Xi) = 0; normality – the error term, Єi, is normally distributed
(See for example, Lewis-Beck 1980; Stevens 1996; Bryman and Cramer 2005; Blaikie 2003; Tabachnick and Fidell 2001; Kleinbaum, Kupper and Muller 1988)
375
APPENDIX VI: STEPS IN ‘RUNNING’ CROSSTABULATIONS
Figure: Appendix VI
STEP TWELVE
Analyze the output
STEP TWELVE
Analyze the output STEP ONE
Assume bivariate
STEP ONEAssume
bivariate
STEP TWO
Select Analyze
STEP TWO
Select Analyze
STEP THREE
Select descriptive statistics
STEP THREE
Select descriptive statistics
STEP FOUR
select crosstabs
STEP FOUR
select crosstabs
STEP FIVEin row place either DV or
IV
STEP FIVEin row place either DV or
IVSTEP SIXin column
vice versa to Step 5
STEP SIXin column
vice versa to Step 5
STEP SEVEN
select statistics
STEP SEVEN
select statistics
STEP EIGHTselect x2,
contingency coefficient and
Phi
STEP EIGHTselect x2,
contingency coefficient and
Phi
STEP NINE
select cells
STEP NINE
select cells
STEP TENin percentage, select – row, column and
total
STEP TENin percentage, select – row, column and
total
STEP ELEVEN
select paste or ok
STEP ELEVEN
select paste or ok
HOW TO
RUN CROSS TABULATIONSin
SPSS?
HOW TO
RUN CROSS TABULATIONSin
SPSS?
376
In order to illustrate the steps in Figure Appendix VI, I will use the hypothesis, “There is a statistical association between ones state of general happiness and the gender of the respondents”
(The variables are general happiness, dependent, and gender, independent)
Step 1:
Select analyze
377
Step 2:
Select ‘Descriptive statistics’
378
Step 3:
Select Crosstabs…
379
On selecting Step 3,this dialogue box will open
380
Step 4:
From the left-hand side, select the variable that you would like to be in the row(s), I prefer the dependent in this section but there is no rule as to where this should go
381
Step 5:
From the left-hand side, select the variable that you would like to be in the column(s), I prefer the independent in this section but there is no rule as to where this should go. However, if the independent variable is place in the row, then the independent goes in the column
Step 6:
Select ‘Statistics’ – this is where the statistical tests are for crosstabs…
382
On selecting Step 6, this dialogue box opens
Step 7:
Choose the appropriate ‘statistics’ – based on the types of variables, and the number of categories of within each variable
Step 8:
Select continue, then ‘cell’- (i.e. which is at the end of the dialogue box
383
Step 9:
There is no rule embedded in stone that you should select ‘row’, ‘column’ and ‘total’ as this is dependent on the researcher. Some researcher chooses what is needed; and this is based on where the independent variable is. If the independent variable is placed in the column, then what are really needed are the column and total percentages. On the other hand, if it is in the ‘row’ then row and total percentages are need and nothing else.
Step 10:Select ‘continue’, and either ‘OK’ or ‘Paste’ from Crosstabs dialogue box-
384
Final Output – this is on completion of the ten steps above. (See the entire ‘Final Output, below
385
FINAL OUTPUT
Case Processing Summary
Cases
Valid Missing Total
N Percent N Percent N PercentGeneral Happiness * Respondent's Sex 1504 99.1% 13 .9% 1517 100.0%
General Happiness * Respondent's Sex Cross tabulation
Respondent's Sex Total
Male Female General Happiness
Very Happy Count206 261 467
% within General Happiness
44.1% 55.9% 100.0%
% within Respondent's Sex
32.5% 30.0% 31.1%
% of Total 13.7% 17.4% 31.1% Pretty Happy Count 374 498 872 % within
General Happiness
42.9% 57.1% 100.0%
% within Respondent's Sex
59.1% 57.2% 58.0%
% of Total 24.9% 33.1% 58.0% Not Too Happy Count 53 112 165 % within
General Happiness
32.1% 67.9% 100.0%
% within Respondent's Sex
8.4% 12.9% 11.0%
% of Total 3.5% 7.4% 11.0%Total Count 633 871 1504 % within
General Happiness
42.1% 57.9% 100.0%
% within Respondent's Sex
100.0% 100.0% 100.0%
% of Total 42.1% 57.9% 100.0%
386
Chi-Square Tests
Value df
Asymp. Sig. (2-sided)
Pearson Chi-Square 7.739(a) 2 .021Likelihood Ratio 7.936 2 .019Linear-by-Linear Association
4.812 1 .028
N of Valid Cases1504
a 0 cells (.0%) have expected count less than 5. The minimum expected count is 69.44.
Symmetric Measures
Value Approx. Sig.Nominal by Nominal Phi .072 .021 Cramer's V .072 .021 Contingency
Coefficient.072 .021
N of Valid Cases 1504
a Not assuming the null hypothesis. b Using the asymptotic standard error assuming the null hypothesis.
(The social researcher having got the output from the Cross Tabulations, see above, needs to know the figures which are appropriate for his/her usage. I have said already that we will always analyze with the independent variables, which means:
NOTE:
χ value is 7.739 (it is taken from the chi-square test table); df (degree of freedom) is 2 (in the chi-square test table); ρ value , 0.021, is taken from the Symmetric measure table and it is the Approx. sig).
The case processing summary has a number of vital information: (1) Total sampled population (that is, the number of people interviewed for this study) 1,517 whereas the number of cases which are used for this cross tabulation is 1,504 (i.e. the valid cases)
I have been emphasizing that we use the independent values for the analysis of the cross tabulations. See below (using the information in the cross tabulation
χ2 = 7.739
Ρ value = 0.021
n = 1,504, the number of cases used for the cross tabulation
387
APPENDIX VII – Appendix 7- Steps in running a trivariate cross tabulation
place independent variables in
column
place dependent variable in
row determine the independent variables
determine the dependent variable
operationalize each variable
conceptualize each variable
Identify variables from
hypothesis
The hypothesis
select the appropriate
statistics
select the necessary percentage
run the SPSS command
388
There is a positive relationship between ones perceived social status and income, and that
this does not differ based on gender?
Step 1 – identification of the variables with the hypothesis – social status, income and
gender (note that there are three variables for this hypothesis unlike if it were social status
and income, thus this question is a trivariate cross tabulation)
Step 2 – define and conceptualize each variable (for this purpose, I will assume that the
variables are already conceptualized and operationalized, hence the substantive issue is
the ‘running of the cross tabulation’
Step 3 – determine the dependent and the independent variables (dependent – social
status; independent variables – income and gender)
Step 4 – End – ‘Running the cross tabulations’ – (see illustrations below)
389
Select ‘Analyze’
390
Select‘analyze’ then ‘descriptive statistics’
391
Having selected‘analyze’ and ‘descriptive statistics’, then you choose‘crosstabs..’
392
393
For this purpose, I will begin with entering the dependent variable first (i.e. entering this with the row space)
394
After which, I will enter the independent variable second (i.e. entering this with the column space)
When has just occurred is called, bivariate analysis, using cross tabulations. To continue this into a trivariate relationship, I will enter the third (control variable) in the final entry box. (see example, below)
395
This process illustrates what is referred to trivariate analysis, using cross tabulations(see final steps below)
396
Selecting the Appropriate statistical test
397
Selecting the necessary cell values42
42 In the spaces below the percentage, there is absolutely no need to select ‘row’, ‘column’ and ‘total’ as the appropriateness of this lies in which position the independent variable is placed. Thus, if the independent variable is placed in the column, then what is needed is the column percentage; and if the independent variable is in the row, then we need the ‘row percentage’. Hence, I have only chosen all three because of formatily.
398
The Final Selection, before ‘running the SPSS’ command
Gender is the control variable, hence, this becomes trivariate analysis
399
FINAL OUTPUT IN SPSS, PART I
Output:Summary of the association
Number of cases used for the association
400
FINAL OUTPUT IN SPSS, PART II
Ρ value for female, 0.003
Ρ value for male, 0.000
χ2 = 150.00
‘df’ is the degree of freedom
401
APPENDIX VIII – WHAT IS PLACED IN A CROSSTABULATION TABLE, USING THE ABOVE SPSS OUTPUT?
Bivariate relationships between general happiness and gender (n= 1,504)
GENDER χ 2 = 7.739
Male Female
Number (Percent) Number (Percent)
Ρ value
0.021
GENERAL HAPPINESS:
Very Happy 206 (32.5) 261 (30.0)
Pretty Happy 374 (59.1) 498 (27.2)
Not Too Happy 53 (8.4) 112 (12.9)
Ρ value = 0.021 < 0.05
402
APPENDIX IX– How to run a regression in SPSS?43
43 Before we are able to run a linear regression, ensure that the metric variables are not skewed. Note a linear regression can also be done without using all metric variables. You could dummy, some. The rule for dummy a variable is K – 1. It should be noted that k denotes the number of categories within the stated variable.
STEP TWELVE
Analyze the output
STEP TWELVE
Analyze the output
STEP ONE
Identify all the variables
STEP ONE
Identify all the variables
STEP TWOdetermine the DV and
the IVS
STEP TWOdetermine the DV and
the IVS
STEP THREE
Select analyze
STEP THREE
Select analyze
STEP FOUR
select regression, then linear
STEP FOUR
select regression, then linear
STEP FIVEplace the DV in the space
marked dependent
STEP FIVEplace the DV in the space
marked dependent
STEP SIXplace the IVs in the space for marked
Indepenent(s)
STEP SIXplace the IVs in the space for marked
Indepenent(s)
STEP SEVEN
select statistics
STEP SEVEN
select statistics
STEP EIGHT
choose descriptive, collinearity diagnostics
STEP EIGHT
choose descriptive, collinearity diagnostics
STEP NINE
select plots
STEP NINE
select plots
STEP TENselect Z
RESID for Y; and Z PRED
for X
STEP TENselect Z
RESID for Y; and Z PRED
for X
STEP ELEVEN
select paste or ok
STEP ELEVEN
select paste or ok
HOW TO
RUN A REGRESSION
MODEL
HOW TO
RUN A REGRESSION
MODEL
403
APPENDIX X– RUNNING REGRESSION IN SPSS
Assume that the hypothesis is “Public expenditure on education and health determines level of development” – variables – public expenditure on education; public expenditure on health, and levels of development (which is measured using HDI). For this example, the dependent variable is levels of development (using HDI) and the independent variables are (1) public expenditure on education and (2) public expenditure on health.
Select Analyze
404
Step 1:Select Analyze
Step 2:Select Regression
Step 3:Select Linear
405
Step 4:Select Dependent variable, from the list of variables
Step 5:Select Dependent variable , Human Development
406
Step 7:Select Independent variable(s) - Public Exp. on Edu
Step 6:Select Independent variable(s), from the list of variables
407
Select Public Exp. on Health
408
Step 8:Select statistics
Step 9:Select – ‘descriptive’ …
409
410
FINAL OUTPUT
Correlations
Correlations
1 .413** .435**
. .000 .000
115 114 106
.413** 1 .395**
.000 . .000
114 165 142
.435** .395** 1
.000 .000 .
106 142 145
Pearson Correlation
Sig. (2-tailed)
N
Pearson Correlation
Sig. (2-tailed)
N
Pearson Correlation
Sig. (2-tailed)
N
PUBLIC EXPENDITUREON EDUCATION ASPERCENTAGE OF GNP(HDR 1994)
HUMAN DEVELOPMENTINDEX: 0 = LOWESTHUMAN DEVELOPMENT,1 = HIGHEST HUMANDEVELOPMENT (HDR,1997)
1990: TOTALEXPENDITURE ONHEALTH ASPERCENTAGE OF GDP(HDR 1994)
PUBLICEXPENDITU
RE ONEDUCATION
ASPERCENTAGE OF GNP(HDR 1994)
HUMANDEVELOPMENT INDEX:0 = LOWEST
HUMANDEVELOPM
ENT, 1 =HIGHESTHUMAN
DEVELOPMENT (HDR,
1997)
1990: TOTALEXPENDITU
RE ONHEALTH ASPERCENTAGE OF GDP(HDR 1994)
Correlation is significant at the 0.01 level (2-tailed).**.
This is the Pearson Moment Correlation Coefficient (0.395)
Level of significance(Ρ value = 0.000, which is written as
0.001)
411
Variables Entered/Removedb
1990:TOTALEXPENDITURE ONHEALTHASPERCENTAGE OFGDP (HDR1994),PUBLICEXPENDITURE ONEDUCATION ASPERCENTAGE OFGNP (HDR1994)
a
. Enter
Model1
VariablesEntered
VariablesRemoved Method
All requested variables entered.a.
Dependent Variable: HUMAN DEVELOPMENTINDEX: 0 = LOWEST HUMAN DEVELOPMENT, 1 =HIGHEST HUMAN DEVELOPMENT (HDR, 1997)
b.
Model Summaryb
.490a .240 .225 .213970Model1
R R SquareAdjustedR Square
Std. Error ofthe Estimate
Predictors: (Constant), 1990: TOTAL EXPENDITUREON HEALTH AS PERCENTAGE OF GDP (HDR 1994),PUBLIC EXPENDITURE ON EDUCATION ASPERCENTAGE OF GNP (HDR 1994)
a.
Dependent Variable: HUMAN DEVELOPMENT INDEX:0 = LOWEST HUMAN DEVELOPMENT, 1 = HIGHESTHUMAN DEVELOPMENT (HDR, 1997)
b.
Coefficient of determination (R2 = 0.240)
412
ANOVAb
1.472 2 .736 16.072 .000a
4.670 102 .046
6.141 104
Regression
Residual
Total
Model1
Sum ofSquares df Mean Square F Sig.
Predictors: (Constant), 1990: TOTAL EXPENDITURE ON HEALTH ASPERCENTAGE OF GDP (HDR 1994), PUBLIC EXPENDITURE ON EDUCATION ASPERCENTAGE OF GNP (HDR 1994)
a.
Dependent Variable: HUMAN DEVELOPMENT INDEX: 0 = LOWEST HUMANDEVELOPMENT, 1 = HIGHEST HUMAN DEVELOPMENT (HDR, 1997)
b.
Coefficientsa
.351 .060 5.811 .000
.033 .012 .257 2.707 .008 .825 1.212
.033 .010 .322 3.392 .001 .825 1.212
(Constant)
PUBLIC EXPENDITUREON EDUCATION ASPERCENTAGE OF GNP(HDR 1994)
1990: TOTALEXPENDITURE ONHEALTH ASPERCENTAGE OF GDP(HDR 1994)
Model1
B Std. Error
UnstandardizedCoefficients
Beta
StandardizedCoefficients
t Sig. Tolerance VIF
Collinearity Statistics
Dependent Variable: HUMAN DEVELOPMENT INDEX: 0 = LOWEST HUMAN DEVELOPMENT, 1 = HIGHEST HUMANDEVELOPMENT (HDR, 1997)
a.
Linear Multiple Regression formula - Y44 = a + b1 X1 + b2X2 + ei
(Levels of Development = 0.351 + 0.033* Public Exp on Edu. + 0.033 * Public Exp. on Health)
44 where Y is the dependent variable, and X1 to X2 are the independent variables; with b1 and b2 being coefficients of each variable
Constant, a, 0.351
b1, coefficient of X1, i.e. Public Exp. on Edu. is 0.033
b2, coefficient of X2, i.e. Public Exp. on Health, is 0.033
ANOVA, analysis of variance, with an F test that is significant 0.000
413
-4 -2 0 2 4
Regression Standardized Predicted Value
-3
-2
-1
0
1
2
Reg
res
sio
n S
tan
da
rdiz
ed
Resid
ual
Dependent Variable: HUMAN DEVELOPMENT INDEX: 0 = LOWEST HUMAN DEVELOPMENT, 1 = HIGHEST HUMAN
DEVELOPMENT (HDR, 1997)
Scatterplot
This aspect of the textbook was only to show how a linear regression in SPSS is done, but in order for us to analysis this, this is already done above.
414
APPENDIX XIa – INTERPRETING STRENGTH OF ASSOCIATION
This section is not universally standardized, and as such, the student should be cognizant that this should not be construed as such. Thus, what I have sought to do is to provide some guide as to the interpretation of the value for Phi, or Cramer’s V, or Contingency Coefficient just to name a few:
InterpretingPhi, Lambda, Cramer’s
V, Contingency Coefficient, et al.
Very weak: 0.00 – 0.19
Weak:0.20 - 0.39
Moderate:0.40 – 0.69
Strong:0.70 – 0.89
Very strong:0.90 – 1.00
415
APPENDIX XIb – INTERPRETING STRENGTH OF ASSOCIATION
Over the years, I have come to the realization that the aforementioned valuations on the strength of statistical correlations can be modified to:
InterpretingPhi, Lambda, Cramer’s V,
Contingency Coefficient, et al.
InterpretingPhi, Lambda, Cramer’s V,
Contingency Coefficient, et al.
Weak:0.00 – 0.39
Weak:0.00 – 0.39
Moderate:0.40 - 0.69Moderate:0.40 - 0.69
Strong:0.70 – 1.00Strong:
0.70 – 1.00
416
APPENDIX XII – SELECTING CASES
Sometimes a researcher may need information on a specific variable. The example here
is, let us say I need information on only males. I could select cases for males.
In this case 1=males, so
417
Step 1:select data
Step 2:choose – select cases
418
Step 3:select if
419
Step 4:select gender
Step 5:select arrow
Step 5:Take this here
420
step 6:Choose =, then the value for the which you need to select, in this case 1, which is for males
421
Step 7:select continue
422
Step 8:select OK or Paste
423
It should be noted that having selected cases for males, any information that is forthcoming would be those for only males, the selected cases. To undo this process (see below)
The result will be something that looks like this, where the select cases are marked (meaning information for only males
424
APPENDIX XIII – ‘UNDO’ SELECTING CASES
Step 1:select dataStep 2:
Choose select cases
425
Step 4:select all cases
426
Final step
Choose OK or Paste, which then remove the markers
427
APPENDIX XIV – WEIGHTING CASES
Sometimes within your research, you may decide to weight the cases owing to sampling
issues or insufficient cases to name a few examples. See below for this process:
The example here is we have decided to weight the cases by 10 (see Illustration below).
Step 10:place the
weight in the
section marked Frequency var.
variable
Step 9:choose weight
cases by, on the right hand side
Step 8:select the word,
weight in weight cases
section
Step 7:
select weight cases Step 6:
select data
Step 5:
select OK or Paste
Step 4:In the numeric
expression, type 10 (i.e. the weight value)
Step 3: In the Target variable, write
the word weight
Step 2:
select compute
Step 1:select
Transform
Weighting cases
428
Step 1& 2: select Compute, then Transform
429
Steps 4 &5: In the Target variable, write the word weigh
430
Step 6: Type the value for the weight, in this case 10.
Step 7: select either OK or Paste
431
Following Step 7, it takes you here
432
With this box, observe what I will do with the weight
433
Step 8:Select weight cases by
434
This is referred to as the arrow
435
The dataset is now weighted by 10
436
APPENDIX XV – ‘UNDO’ WEIGHTING CASES
Step 1:select data and then weight cases
437
step 2:look for the word weight on the left hand side, window
438
This is what would have existed from the process of weighting the cases, so in order to undo this, see the final set below
439
Final step:select Do not weight cases, then either OK or Paste
440
In the event, the researcher wants to calculate the average or the mean value of say a group of variables. In this case, I would like the find the average score for two test scores. (Variables to be used are – Questions
3.1 In Advanced Level, what were your last two (2) tests scores over the past six (6) months? (1) _______________________
(2) _______________________
Step 1:Select Transform
Step 2:Select Compute
441
Use a phrase or word to identify the averaged score
Detailed the variable, which is used to identify the variable
442
Select the mean, which is used to calculate the average score for number of variables
443
select, the arrow, which results in
444
Step 2:
Separate each variable that will be used by a comma
Step 1:Select each variable from this section, then use the arrow
Step 3:
Select OK or Paste
445
The following will be done to ‘run’ the descriptive statistics for the new variable, called averaged scores
446
APPENDIX XV – Statistical and/or mathematical Symbolism
µ - mu – Population mean
α - alpha – level of significant; probability of Type I error
θ - sigma -
β - beta - probability of Type II error
1 - β - power
Σ - summation – total of a set of observation (i.e. data points)
Ν - population (i.e. parameter) – total of all observations of a population
n - sample (i.e. statistic) – total of all observations of a sub-set of a population
Φ - phi - statistical test, which is used in the event of dichotomous variable
Ŷ - predictor of Y
± - plus and/or minus
< - less than
> - greater than
γ - gamma
≤ - less than or equal to
≥ - greater than or equal to
≠ - not equal to
≈ - approximately equal to
H1 - alternative hypothesis (i.e. Ha)
H0 - null hypothesis
r - Pearson’s moment correlation coefficient
r2 - coefficient of determination (i.e. strength of a linear relationship)
λ - lambda
Δ - delta (i.e. incremental change)η - etaρ - rho
447
χ2 - chi-square
APPENDIX XVI – Converting ‘string’ data into ‘numeric’ data
Sometimes a researcher may not have entered the data him/herself, and so the data
entry operator may use ‘string’ to enter the data in SPSS instead of numeric. From
entering the data as ‘string’ it prevents further manipulation of the as the data are not
considered as numbers but rather letter (see example below).
Before the researcher begins with any form of data analysis he/she should check to ensure that the data are entered as ‘numeric’ and not ‘string data. This is found in the ‘variable view’ window to the end of the SPSS window (see below)
448
Having established that data were entered as ‘string’ data, the researcher can use any of
the following options:
Apparatus One
(i) Use – for example ‘a20’ on each occasion that the variable will be used for any form
of analysis (see Figure 1); or
Apparatus Two
(ii) Convert the ‘string’ into ‘numeric’ data (see Figure 2).
In the forthcoming pages, I will seek to provide detailed information on how the
processes of converting ‘string’ into ‘numeric’ data’ are achieved using option II.
449
CONVERTING FROM ‘STRING’ TO ‘NUMERIC’ DATA45
Figure 1: CONVERTING FROM ‘STRING’ TO ‘NUMERIC’ DATA: When the data were entered as numbers, only. (See illustration below, the SPSS form)
45 There are instances, when the researcher uses a combination of ‘letters and ‘numbers’. In this case we use Figure 2 instead of Figure 1(See figure 2, below).
View the Variable View -
which is at the bottom of the SPSS
– Data Editor Window?
Pursue the Data View, to
establish ‘how data were entered?’
Then, select the right-hand side to the ‘string’
the option ‘numeric’. Then
OK.
If the data were entered as,
numbers but the researcher
selected ‘string’
Return to ‘Variable View’, and then go to the variable in
question …
STARTING POINT
END,HERE.
450
Step 1select to the right-hand side of this box
APPARATUS ONE
451
Step 2:
Having selected the right-hand side to the string for the variable, it produces this dialogue box. Remove the mark from ‘string’ to numeric. (See illustration, below).
452
(Note: The process that has just ended is an illustration of how we address converting ‘string’ data to ‘numeric’ data, if the initially data were entered as number but the data entry clerk had selected ‘string’ in Type instead of ‘Numeric’. (See below, how the combination is handled).
By select ‘Numeric’, we have deselected ‘string’
Step 3:To execute the command, we select ‘OK’
453
CONVERTING FROM ‘STRING’ TO ‘NUMERIC’ DATA
Figure 2: CONVERTING FROM ‘STRING’ TO ‘NUMERIC’ DATA: When the data were entered as numbers and letters.
View the Variable View -which is at the bottom of the SPSS – Data
Editor Window?
Pursue the Data View, to establish ‘how
data were entered?’
Select ‘Old and New Values’
If the data were entered as,
numbers and letters but the researcher
selected ‘STRING’
Select ‘Transform’,
‘Recode’, then go to ‘Into
same variable’
Leave all the numeric values, and then select the letter in the
form it was type – SEE END
In old value type the ‘letter’, in
New value type the number, then
OK.
START
TEND
454
APPPARATUS TWO
Step 1: Run the frequency for the variable labeled ‘string’. In this case, the variable is a20.
455
Note:
From all indications, the clerk entered 1, 2, 3, 4, 5, and N in the data view. This is the reason for this output. Thus, this ‘string’ can be converted to numeric by (see illustration below).
456
Steps 1 to 3:
Select ‘Transform’, ‘Recode’, and ‘Into Same Variables…’
457
Step 4 and 5:
Identify the variable on the left-hand side (i.e. the dialogue box), then use the arrow to take it into the space marked ‘Variable’
458
This is the result from executing steps 4 and 5.
Step 6:
Now the next step is to select ‘Old and New Values…’
459
The researcher needs to understand that the conversion is not for the numeric variables
that are present within the data set but for the letter ‘N’, as this was mistakenly recorded
by the data-entry clerk. Thus, we are seeking to correct the error. (See below).
Step 7:
The mistake was using capital ‘n’ instead of no, which was coded as two. Note whatever is used in the first instance must be entered herein. (See page 399, N).
Step 8:
Initially, what the clerk should have been entered was 2; instead he/she used N. Thus, now, we select New Value and type the number 2.
Step 9:
In order that this command can be recorded, we need to select ‘Add’, which takes it into the dialogue box marked ‘Old→New’. On completion, you should select ‘continue’ then ‘OK’ which will then execute the command.
460
461
This is the output for the variable that had a combination of ‘string’ and ‘numeric’ data pre the conversion exercise.
On completion of the steps carried out earlier, this is the result of what the variable looks like post the exercise. There is no more ‘N’ of 44 case, it is now in two, which has increased by 44 cases (i.e. the frequency of two was 464, with the additional 44 cases it becomes 508.
Having used the steps above, the researcher will then perform the final step by converting the variable from ‘string’ to ‘numeric’ data. using Apparatus One.
462
APPENDIX XVII – Running Spearman
Figure: Steps to following to performing Spearman’s ranked ordered Correlation
Step 4: Use either OK or paste to execute the command chosen in step 3
Highlight and choose the ordinal variables from the left-hand-side, then use the arrow between left-hand and right-hand side to select the variables in the dialogue box on the right hand side that was empty
Step 3:
Step 2:In order to run a an ordinal against an ordinal variable, you should deselect Pearson and choose Spearman
Step 1:Select
Analyze→ correlate→ bivariate
Steps in running Spearman rho
Step 3:
Highlight and choose the ordinal variables from the left-hand-side, then use the arrow between left-hand and right-hand side to select the variables in the dialogue box on the right hand side that was empty
Step 1:
Select Analyze → correlate
→ bivariate
463
Step 1:
Select analyze, then correlate and followed by bivariate…
464
Step 2:
By default the computer shows Pearson, in order to run a an ordinal against an ordinal variable, you should deselect Pearson and choose Spearman
465
Step 3: Highlight and choose the ordinal variables from the left-hand-side, then use the arrow between left-hand and right-hand side to select the variables in the dialogue box on the right hand side that was empty
Step 4:
Use either OK or paste to execute the command chosen in step 3
466
Final Output from the entire step executed above
Given that there is no relationship from a noted sig. ( 2-tailed) that is more than 0.05, correlation coefficient is not used as there is no association to establish strength and/or direction
The sig. (2-tailed) of 0.704 is used to state whether a relationship exists at the 0.05 level of significance
467
APPENDIX XVIII – Running Pearson
Figure: Steps to following to performing Pearson’s Product moment Correlation
Step 4: Use either OK or paste to execute the command chosen in step 3
Highlight and choose the ordinal variables from the left-hand-side, then use the arrow between left-hand and right-hand side to select the variables in the dialogue box on the right hand side that was empty
Step 3:
Step 2:
Select a set of metric variables, which are normally distributed
Step 1:Select
Analyze→ correlate→ bivariate
Steps in running Pearson
Step 3:
Highlight and choose the metric variables from the left-hand-side, then use the arrow between left-hand and right-hand side to select the variables in the dialogue box on the right hand side that was empty
Step 1:
Select Analyze → Correlate
→ Bivariate
468
Step 1:
Select analyze, then correlate and followed by bivariate…
469
Step 2:
By default, the computer shows Pearson, this should be left alone
470
∙
Age
Income
471
Pearson
Income
Age
472
APPENDIX XIX – CALCULATING sampling errors from sample sizes
Students should be aware that despite the scientificness of statistics, the discipline recognizes that by seeking to predict events (behavioural or otherwise), there is a possibility of making an error. This is equally so when deciding on a particular sample size.
se = z√ [(p %( 100-p %)] √ s
Symbols and their meanings:
se = sampling error (i.e. the percentage of error that the researcher is prepared to accept or tolerate)s = sample size (or n)
z = the number relating to the degree of confidence you wish to have in the result: (note 95% CI, z= 1.96; 99% CI, z=2.58; and 90% CI, z=1.64)
p = an estimated percentage of people who are into the group in which you are interested in the population
In order to illustrate the usage of the above formula, we will give an example. Here for example, assume that from a sample of 500 respondents (s or n), 20% of people will vote for the PNP/JLP in the upcoming elections (p – percentage or proportion). What is the sampling error, using a 95% confidence level?
se = 1.96√(20(80)) √ 500
Interpretation of the results:
The result from the formula is 3.5% (this can either be positive or negative). The value denotes, ergo, that based on a sample of 500 Jamaicans, we can be 95% sure that the true measure (e.g. voting behaviour) among the whole population from which the sample was drawn will be within +/-3.5% of 20% i.e. between 16.5% and 23.5%.
473
APPENDIX XX – CALCULATING sample size from sampling error(s)
One of the fundamental requirements of executing social (or natural science) research is selecting a sample. The researcher must decide on how many people (or subjects or participant) that she/he would like to survey, interview or speak with in regard a particular subject matter. In quantitative studies, the researcher must select from a population (i.e.) a subpopulation (sample) with which s/he is interesting to garner germane information. There are two formulae that are available to the researcher.
Formula One
n = (z / e) times 2
Symbols and their interpretations:
n = the sample size
z = the value for the level confidence level. Researchers frequently use a 95% confidence level, but this is not carved in stone. Other confidence levels can be used such as 99% and its ‘z’ is 2.58; 95% confidence with a ‘z’ value of 1.96; ‘z’ = 1.64 for 90% confidence and 1.28 for 80% confidence.
e = the error you are prepared to accept, measured as a proportion of the standard deviation (accuracy)
For a better understanding of this situation, we will use an illustration. The example here is, assume that we are estimating mean weight of a women in Lucea, Hanover, and that we wish to identify what sample size to aim for in order that we can be 95% confident in our result. Continuing, let us assume that we are prepared to accept an error of 10% of the population standard deviation (previous research might have shown the standard deviation of income to be 8000 and we might be prepared to accept an error of 800 (10%)), we would do the following calculation:
n = 2(1.96 / 0.1)
Therefore s = 384.16. As such, we should use 385 people.
Because we interviewed a sample and not the whole population (if we had done this we could be 100% confident in our results), we have to be prepared to be less confident and because we based our sample size calculation on the 95% confidence level, we can be
474
confident that amongst the whole population there is a 95% chance that the mean is inside our acceptable error limit. There is of course a 5% chance that the measure is outside this limit. If we wanted to be more confident, we would base our sample size calculation on a 99% confidence level and if we were prepared to accept a lower level of confidence, we would base our calculation on the 90% confidence level.`
Formula Two
n = z 2 (p (1-p)) e2
Symbols and their interpretations:
n = the sample size
z = the number relating to the degree of confidence
p = an estimate of the proportion of people falling into the group in which you are interested in of the population
e = the proportion of error that the researcher decides to accept
We will use a hypothetical case of voters to illustrate the use of this formula, which is different from Formula One. If we assume that we wish to be 99% confident of the result i.e. z = 2.85 and that we will allow for errors in the region of +/-3% i.e. e = 0.03. But in terms of an estimate of the proportion of the population who would vote for the PNP/JLP candidate (p – proportion and not party abbreviation), if a previous survey had been carried out, we could use the percentage from that survey as an estimate. However, if this were the first survey, we would assume that 50% (i.e. p = 0.05) of people would vote for candidate X and 50% would not. Choosing 50% will provide the most conservative estimate of sample size. If the true percentage were 10%, we will still have an accurate estimate; we will simply have sampled more people than was absolutely necessary. The reverse situation, not having enough data to make reliable estimates, is much less desirable.
In the example:
s = 2.582(0.5*0.5) = 1,8490.032
475
This rather large sample was necessary because we wanted to be 99% sure of the result and desired and desired a very narrow (+/-3%) margin of error. It does, however reveal why many political polls tend to interview between 1,000 and 2,000 people.
476
APPENDIX XXI – Sample sizes and their sampling errors
One thing that must be kept in mind when doing research that there is truth that errors
are ever present with sampling or for that matter equally existing in census data. With
this recognition, the researcher must now plan what is an acceptable sampling error that
she/he wants from a certain sample size. Thus, the choice of a sample size should not be
arbitrary but it should be based on – (i) the degree of accuracy that is required from the
selected sample size, and (ii) the extent with which there is a variation in the population
with regard to the principal features of the study. We will now provide a listing of sample
sizes and their appropriate sampling error, assuming that we are using the 95% level of
confidence (i.e. confidence level - CI).
Table 1: Sample errors and their appropriate sample sizes, using a CI of 95%46
Sample Error (in %) Sample Size Sample Error (in %) Sample Size1.0 10000 6.0 2771.5 4500 6.5 2372.0 2500 7.0 2042.5 1600 7.5 1783.0 1100 8.0 1563.5 816 8.5 1384.0 625 9.0 1234.5 494 9.5 1105.0 400 10.0 1005.5 330
Interpretation: This is simple, do not be scared, as 1.0% which is beneath sample error corresponds to a sample size of 10,000 respondents (or subject or participants or interviewed). Continuing, if one were to select a sample size of 277 participants for a survey, using 95% confidence level, then she/he is expected to have a sample error 6.0%. It should be noted that Table 1 above, assumes a 50/50 split for the sample size (i.e. this should be used if the researcher is unsure what the proportion of population might be that she/he intends to study).
46 In attempting to make this text simple, we have sought to provide the easy way to understand complex materials. Thus, the calculation of Table above can be done by inputting the figures (the sample size 10,000 and 50% sample proportion in space provided on (http://www.dssresearch.com/toolkit/secalc/error.asp), and no figure should be placed in total population, because this is in keeping with the assumption that the researcher does not know this. Note 50% produces the largest likely variation.
477
APPENDIX XXII – Sample sizes and their sampling errors
Table 1: Sample errors and their appropriate sample sizes, using a CI of 95%Sample Error (in %) Sample Size47 Sample Error (in %) Sample Size0.6 10000 3.4 2770.8 4500 3.6 2371.1 2500 3.9 2041.4 1600 3.9 2001.7 1100 4.2 1782.0 816 4.5 1562.2 625 4.8 1382.5 494 5.0 1232.8 400 5.3 1103.1 330 5.6 100
Factors which are used in determining a sample size
1) the degree of accuracy required for the sample; and
2) the extent to which there is variation in the population concerning the key characteristics of the study
47 Table 1 above, assumes a 90/10 split for the sample size (i.e. we are assuming that the sample represents a 10% of the population - the proportion of population is 10%).
478
APPENDIX XXIII – If conditions
In order that we will be able to make to grasp the understanding of this ‘If conditionalities’ in
research, we will present a frequency tables of tow univariate factors – (i) gender and (ii) age of
the sampled group.
Table 1: Gender of the respondents
Frequency Percent Valid Percent
Cumulative Percent
Valid MALE 59 43.4 43.4 43.4 FEMALE 77 56.6 56.6 100.0
Total 136 100.0 100.0
Table 2: The age distribution of the sampled population
Frequency Percent Valid PercentCumulative
PercentValid 16 25 18.4 18.5 18.5 17 51 37.5 37.8 56.3 18 40 29.4 29.6 85.9 19 13 9.6 9.6 95.6 20 3 2.2 2.2 97.8 21 1 .7 .7 98.5 22 1 .7 .7 99.3 25 1 .7 .7 100.0 Total 135 99.3 100.0Missing System 1 .7Total 136 100.0
To effectively reduce this to micro simplicity, we will be seeking to carryout a command, which
is to ascertain young adults (i.e. respondents who are at most 16 years at their last birthday).
479
If conditionality (or If condition) are a set of mathematical formulae with which the researcher
will write as a programme that upon completion, the computer (using SPSS) will generate the
commands which were given it.
In order to bridge the challenge of this apparatus to you the reader, we will perform the task
through a serious of step.
Steps 1
→ Go to the SPSS menu bar, where you will see a number of words including ‘File’.
Select the ‘File’ by ‘clicking’ on that option
480
Steps 2
→ Now you would be within the ‘File’ menu bar, and so your next step is to select ‘New’
followed by the word ‘syntax’. It is through this widow that the mathematical formula will be
store and manipulated.
481
Steps 3
→ Because you have selected ‘New’ and ‘syntax’, a program will that is called the
‘syntax’ will now appear (see display below)
482
Steps 4
→ Note that our objective is to construct a program with which the computer on the given
instruction will create a variable called young adults (i.e. respondents who are at most 16 years
of age at their last birthday).
483
In order to understand why we have written these jargons, you need to know the end objective. This is a variable which denotes young adults (<or = 16 yrs.).
With this in mind, the next step is to write If (the variable which houses gender - i.e. q1 and the value for male – i.e. 1 then and (or &) which is the symbol that speaks to the desire overlap between being young and male) followed by the name of the new variable – i.e. young adults, equals a value which represents young men. On completion of each
expression, a period should follow – ‘.’
The same process is carried out for the young female, with a few modifications. These changes are necessary as 2 is the valuation for the female within q1. The next adjustment is the valuation for ‘Young adults’ which must be different from the value given to the males. Hence, this is the why it was called 2 indicate the new label. The final command that is used is the now ‘execute’ followed by period. If you are to highlight and ‘run’ this expression the computer will give you a table with young male ‘1’ and females ‘2’.
Running the Command
484
Comparing the result to ascertain the truthfulness of the operation
Table 3: Young_Adults_1
Frequency Percent Valid PercentCumulative
PercentValid 1.00 16 11.8 64.0 64.0 2.00 9 7.4 36.0 100.0 Total 25 19.1 100.0Missing System 111 80.9Total 136 100.0
Note carefully- using the age distribution that only 25 respondents are approximately 16 yrs. old.
Table: Age at last birthday
Frequency Percent Valid PercentCumulative
PercentValid 16 25 18.4 18.5 18.5 17 51 37.5 37.8 56.3 18 40 29.4 29.6 85.9 19 13 9.6 9.6 95.6 20 3 2.2 2.2 97.8 21 1 .7 .7 98.5 22 1 .7 .7 99.3 25 1 .7 .7 100.0 Total 135 99.3 100.0Missing System 1 .7Total 136 100.0
485
Students should be cognizant that cross tabulation can be used to verify the authenticity of the
mathematical formula (see below)
486
APPENDIX XXIV – The meaning of the ρ value
The ρ value speaks to the likelihood that a particular outcome may have occurred by chance.
Thus, ρ = 0.01 level of significance, means that there is a 1 in 100 probability that the result may
have happened by chance or a 99 in 100 probability that the outcome is a reliable finding.
Furthermore, ρ = 0.05 is a 1 in 20 probability (or 5 in 100) probability that the observed results
may have appear by chance. Another matter is that a significance level of 0.05 to 0.10, indicates
a marginal significance. Social scientists have generally used the rule of thumb of 0.05 level of
significance to indicated statistical significance. Thus, when the level of significance falls
below 0.05 (e.g. 0.01, 0.001, 0.0001, etc), the smaller the numeric value the greater the
confidence of the researcher in speaking about his/her findings (i.e. the findings are valid).
I would like for reader to note here that in the social environment (i.e. in particular social
sciences), nothing is ever “proved”. This position is not the same in the natural sciences (or
physical sciences) as phenomena can be “proved” but in the social space, it can be demonstrated
or supported at a certain level of significance (or likelihood). Again, the smaller the ρ value, the
greater is the likelihood that the findings are valid.
487
APPENDIX XXV – Explaining Kurtosis and Skewness
Skewness is a statistically measure that is used by statisticians and researchers
to evaluate the distribution of a data. It measures the degree of a distribution of
values divide the symmetry around the mean. The value for skewness may be
more than zero (i.e. 0) or less than zero; where a value of zero (0) indicates a
symmetric or evenly balanced distribution. A value of zero is ideal and in social sciences
the realistic values will more likely be ± 1, ±2 or ± ≥3; and a skewness value between ±1
is considered excellent for most social scientists, but some argue that a value between
±2 is also acceptable. The issue of acceptability speaks of value without which no
modification is required as it can be used as indicating normality. However, in this text
we will use between ±1; and any value more 1 or less than negative 1 is unacceptable as
this indicates high skewness.
Kurtosis evaluates the “peakness” or the “flatness” of a frequency distribution (or frequency curve). Kurtosis’ value is indicate a similarly to skewness as zero (0) means
normality. However, this is idealistic and so the acceptable reality is between ±1, which
is considered an excellent mark of normality, and so social scientists cite that this can be
between ±2. Nevertheless, in this text we will use between ±1; and any value more 1 or
less than negative 1 is unacceptable as this indicates high skewness.
488
APPENDIX XXVI – Sampled Research Paper I
Health Determinants: Using Secondary Data to Model Predictors of Wellbeing of Jamaicans
Paul Andrew Bourne48
Department of Community Health and Psychiatry, Faculty of Medical Sciences
The University of the West Indies at Mona, Jamaica
Brief synopsis
This study broadens the operational definition of wellbeing from physical functioning (or health
conditions) to include material resources and income. Secondly, it seeks to provide a detail
listing of predisposed variables and their degree of influence (or lack of) on general wellbeing.
48 Correspondence concerning this article can be by email: [email protected] or by telephoning (876) – 841-4931 or by mail to Department of Community Health and Psychiatry, Faculty of Medical Sciences, The University of the West Indies, Mona-Jamaica.
489
Abstract
Objective. During 1880-1882 life expectancy for Jamaican males was 37.02 years and 39.80 for their female counterparts and 100 years later, the figures have increased to 69.03 for males and 72.37 for females. Despite the achievements in increased of life expectancies of the general populace and the postponement of death, non-communicable diseases are on the rise. Hence, this means that prolonged life does not signify better quality life. Thus, this study seeks to examine the quality of life of Jamaicans by broadening the measure of wellbeing from the biomedical to the biopsychosocial and ecological model Method. Secondary data was used for this study. It is a nationally representative sample collected by the Statistical Institute of Jamaica and the Planning Institute of Jamaica in 2002. The total sample is 25,018 respondents of which the model used 1,147. Data was stored and analysed using the Statistical Packages for the Social Sciences (SPSS). Multivariate regression was used to test the general hypothesis that wellbeing is a function of psychosocial, biological, environmental and demographic variables.Results. The model explains 39.3 percentage of the variance in wellbeing (adjusted r2). Among those 10, the 5 most significant determinants of wellbeing in descending order are average number of persons per room (β = -0.254, ρ < 0.001); area of residence (1=KMA), (β = -0.223, ρ < 0.001); area of residence (1=Other Towns), (β = -0.209, ρ < 0.001); and lastly age of respondents (β = -0.207, ρ < 0.001). Those five variables accounted for 27.2 percentage of the model, with average occupancy and area of residence (being KMA) accounted for 7 percentages each.Conclusion. This study has shown that wellbeing is indeed a multidimensional concept. The paper has proven that the determinants of wellbeing include psychosocial, environmental and demographic variables.
490
Introduction
Many scholars such as Erber (1), Brannon and Feist (2) have forwarded the idea that it is
germane and timely for us to use a biopsychosocial construct for the measurement of quality of
life. But neither Erber nor Brannon and Feist have proposed a mathematical model that can be
used to evaluate this worded construct. This is also similar to and in keeping with the broad
definition given by the WHO in 1946 (3), and later promulgated by Dr. George Engel (4-8).
However, in 1972, Grossman (9) filled this gap in the econometric analysis to formulate a
measurement for health. This was later expanded by Smith and Kington (10,11). Despite the
premise set by Grossman, Smith and Kington used physical functioning in their definition of
health, which again is a narrow approach to the concept of health and wellbeing. Grossman’s
model which was further enhanced by Smith and Kington did not provide us with the relative
contribution of each of the determinants of wellbeing. On the other hand, a study by Hambleton
et al (12) in Barbados, decomposed the predictors of self-reported health conditions, and found
that 38.2% of the variation in health status can be explained by some predisposed variables. Of
the variation explained, ‘current health status’ account for 24.5%, lifestyle risk factors, 5.8%,
current socioeconomic factors, 2.5% and historical conditions, 5.4%. The composition of the
aforementioned groups were (i) Historical indicators – education, occupation, childhood
economic situation, childhood nutrition, childhood health, number of childhood diseases; (ii)
Current socioeconomic indicators – income, household crowding, currently married, living
alone; (iii) Lifestyle risk factors – body mass index, waist circumference, categories of diseases,
smoking, exercise and (iv) current Disease indicators – number of illness, number of symptoms,
geriatric depression, number of nights in hospitals, number of medical contacts in 4-month
period. Again, while Hambleton et al’s work provided explanations that determinants of
491
wellbeing expand beyond ‘current disease conditions’ to lifestyle practices and socioeconomic
factors using ‘physical functioning’ (i.e. health conditions) in conceptualizing health. This is not
in keeping with the WHO expanded definition (3). Such an approach focuses on the mechanistic
result of the exposure to certain pathogen which results in ‘disease-causing conditions’.
The WHO’s definition has been widely criticized for being elusive and immeasurable
because the concept is too broad. On the other hand, the traditional view of the Western culture
is such that health means the ‘absence of diseases’ (Papas, Belar & Rosensky (13). However, in
the 1950, a psychiatrist, Dr. Engel (4-8), began promoting what he referred to as the
biopsychosocial model. He believed that the treatment of mental health must be from the
perspective of the body (i.e. biological conditions), mind (i.e. psychological) and sociological
conditions. Engel believed that the psychological, biological and social factors are primarily
responsible for human functioning. He forwarded the thought that these are interlinked system
in the treatment of health care, which is compared to the interconnectivity of the various parts of
the human body. Engel believed that when a patient visits the doctor, for example, for a mental
disorder, the problem is a symptom not only of actual sickness (biomedical), but also of social
and the psychological conditions. He, therefore, campaigned for years that physicians should use
the biopsychosocial model for the treatment of patient’s complaints, as there is an
interrelationship among the mind, the body and the environment. He believed so much in the
model that it would help in understanding sickness and provides healing that he introduced it to
the curriculum of Rochester Medical School (14, 15). Medical psychology and psychopathology
was the course that Engel introduced into the curriculum for first year medical students at the
University of Rochester. This approach to the study and practice of medicine was an alternative
paradigm to the biomedical model that was popular in the 1980s and 1990s, and is still popular in
492
Jamaica in 2007. In writing about wellness and wellbeing, there are no studies in Jamaica that
can definitely state that these are the determinants of wellbeing, or quality of life. Dr. Pauline
Milbourn Lynch (16), Director of Child and Adolescent Mental Health in the Ministry of Health
in 2003, argued that wellness is “a balance among the physical, spiritual, social, cultural,
intellectual, emotional and environmental aspects of life” but, there is no research that put all of
these conditions together, and show their relationship with wellbeing. As such, a model was
constructed which will be in keeping with the concept of the biopsychological model. This study
seeks to examine the quality of life of Jamaicans by broadening the measure of wellbeing and to
ascertain possible factors that can be used to predict wellbeing from a biopsychosocial and
environmental approach as against the traditional biomedical model (i.e. biological conditions or
the absence of pathogens).
Theoretical Framework
The overarching theoretical framework that is adopted in this study is an econometric model that
was developed by Grossman (9), quoted in Smith and Kington (10), which reads:
Ht = ƒ (Ht-1, Go, Bt, MCt, ED) ……………………………………… (2)
In which the Ht – current health in time period t, stock of health (Ht-1) in previous period, Bt –
smoking and excessive drinking, and good personal health behaviours (including exercise – Go),
MCt,- use of medical care, education of each family member (ED), and all sources of household
income (including current income)- (see Smith and Kington 1997, 159-160). Grossman’s model
further expanded upon by Smith and Kington to include socioeconomic variables (see Equation
3).
Ht = H* (Ht-1, Pmc, Po, ED, Et, Rt, At, Go) …. ……………………… (3)
493
Equation (i.e. Eq.) (2) expresses current health status H t as a function of stock of health
(Ht-1), price of medical care Pmc, the price of other inputs Po, education of each family member
(ED), all sources of household income (Et), family background or genetic endowments (Go),
retirement related income (Rt ), asset income (At,)
Among the limitations in the use of the biopsychology model that is use by Smith and
Kington are psychological conditions and ecological variables. This study is equally limited by
many of the variable used in Eq. (2) because data from this study is based Jamaica Survey of
Living Conditions (JSLC) and Labour Force Survey (LFS) were not primarily intended for this
purpose. The JSLC is a national cross-sectional study which collects data for general policy
formulation and so we will not be able to track the individuals over time in order to establish a
former health status (17). The updated JSLC and LFS do have information – such as
preventative lifestyle behaviour – exercise, family background, and not-smoking. The JSLC, on
the other hand, collects data on crime and victimization, environment conditions and household
size, room occupancy, gender and age of respondents, which were all important for this modified
model from that use by Smith and Kington in Equation 3.
W=ƒ ( Pmc , ED, Ai , En, G, M, AR, P, N, O, Ht, T, V,S, HS) ………… (4)
Wellbeing of Jamaican W, is the result of the cost of medical care (Pmc), the educational
level of the individual, ED, age of the respondents, the environment (En), gender of the
respondents (G), marital status (M), area of residents (AR), positive affective conditions (P),
negative affective conditions (N), average number of occupancy per room (O), home tenure,
(Ht), land ownership(proxy paying property taxes), T, crime and victimization, V, social support,
S, seeking health services, HS.
Method and Data
494
This research uses secondary data [JSLC, 2002)] that is a joint publication of the
Planning Institute of Jamaica (PIOJ) and the Statistical Institute of Jamaica (STATIN). Its
purpose is to divulge the efficiency of public policy on the Jamaican economy. The survey was
carried out between June-October, 2002; it is a subset of the Labour Force Survey (i.e. ten
percent). Of a population of 9,656 households, the sample size used for the JSLC was 6,976
households. The instrument (i.e. questionnaire) was categorized based on demographic
characteristics, household consumption, education, health, social welfare and related
programmes, housing and criminal victimization.
Based on interpretability and parsimony, the best model was obtained using the entry
method, which involved entering all the variables in block in a single step. To assess how well
the model fits the data, the F test was used. A single multiple regression model was used to fit
the data, which is the Wellbeing (W) of Jamaicans. We examined the statistical importance of
each predictor using squared value of the partial correlation coefficients. All the predisposed
variables were added to the model at once, and the enter technique was used to ascertain those
variables that are statistically significant determinants with associated 95% confidence intervals
(CIs).
495
Results
Demographic characteristics
Respondents’ background
The total sample was 25,018 of which there was 49.3% males (n=12,332) compared to 50.7%
females (12,675). The average age of the sample was 29 years (± 21 years), with the median age
being 24 years. Decomposing age by gender reveals that the average age for females (29 yrs. ±
22 yrs.) was marginally greater than that of males (28 years ± 22 yrs). The mean overall
wellbeing of Jamaicans is low (4 out of 14), with the mode being 4.5. Wellbeing is a composite
variable constituting material resources (MR) and health conditions (H). It is calculated as
follows: W = ½ ∑ MR – ½ ∑ Hi. Where higher values denote more wellbeing. The index ranges from a
low of -1 to a high of 14. Scores from 0 to 3 denotes very low, 4 to 6 indicates low; 7 to 10 is moderate
and 11 to 14 means high wellbeing.
Furthermore, the majority of the sample was never married (67.3%, n=10,813) followed by
married (25.2%, n=4,050), widowed (5.6%, n=905), separated (1.2%, n=185) and lastly those
who are divorced (0.8%, n=123). Marginally more males are in each group within the marital
status category than females except in ‘widowed’ and separated. (See Table 1.1.1).
Predisposed Factors in Wellbeing Model
In this section of the paper, the General hypothesis will be tested:
W=ƒ (Pmc, ED, Ai , En, G, M, AR, P, N, O, Ht, T, V,S, HS)………………………….(1)
Of the 14 predisposed factors that were tested (see Eqn. 1), 10 came out be predictors of
wellbeing. Among those 10, the 5 most significant determinants of wellbeing in descending
order are average number of persons per room (β = -0.254, ρ < 0.001); area of residence
(1=KMA), (β = -0.223, ρ < 0.001); area of residence (1=Other Towns), (β = -0.209, ρ < 0.001);
496
and lastly age of respondents (β = -0.207, ρ < 0.001). (See Table 1.1.2). Based on the signs
associated with the unstandardaized coefficient, area of residence, positive affective conditions,
individual’s educational attainment and marital status are positively associated with wellbeing,
with the others being negatively relating to wellbeing. Those that are not factors of wellbeing
are as follows: (1) seeking health care (β = 0.014, ρ > 0.05); (2) gender ((β = 0.015, ρ > 0.05); (3)
crime and victimization ((β = 0.030, ρ > 0.05), and (4) house tenure ((β = -.003, ρ < 0.05). (see
Table 1.1.2).
Continuing, the model explains 39.3% (i.e. adjusted r2) of the variance in wellbeing. One
may argue that the unexplained variation is significantly more than the explained variation and
so the model is useless. But, the finding in this study is in keeping with Hambleton’s et al.’s
research which was conducted on elderly persons in Barbados in 2005 (Hambleton and his
colleague 12). They found that 38.2% of the variance in predisposed variables can explain the
variance in wellbeing of elderly Barbadians.
W=ƒ ( Pmc , ED, Ai , En, G, M, AR, P, N, O, Ht, T, V,S, HS)…………………………(1)
Hence from the equation [1] above, we derived a linear model with only the predisposed
variables that are significant:
W= 1.922+ 0.197Pmc + 1.091AR 2 + 1.698 AR 3 – 0.633 En + 0.341 M1 + 0.560 M2 + 0.240 ED 2
+ 1.700 ED3 + 0.210S – 0.691O + 0.606 T + 0.105P -0052N-0.022 Ai + ei
Interpreting the linear model:
It follows that with all else being constant, the minimum wellbeing of a Jamaican is 2 (i.e.
1.922), which means that the overall wellbeing of this individual would be very low. With the
497
referent group being living in rural Jamaica, the coefficient of 1.091 for AR 2 denotes that people
with dwell in the Kingston Metropolitan Area has a greater wellbeing by this coefficient. The
interpretation for AR 3 is similar to that of AR 2, with the exception that those who residence in
Other Town have a higher wellbeing when compared to those who live in rural Jamaica.
Continuing, from the coefficient of area of residence, the highest wellbeing is experienced by
those to dwell in Other Towns. The same reasoning is applicable to individual’s educational
attainment, 0.240 ED 2 + 1.700 ED3. It should be note here that the wellbeing of someone who
has tertiary level education is significant more than that of individual with primary and below
education, and that this is substantially greater when compared to someone who has only attained
secondary level education.
Based on the coefficient for En (i.e. environment), an individual’s will decrease by 0.633
units because of the living in an environment with natural disaster, and toxins. Hence, the same
interpretation can be used for Age (i.e. Ai), positive affective conditions, P, and negative
affection conditions, N, land ownership, T, cost of health care, Pmc,, and those who have social
support, S. The difference in these cases would be based on a reduction or an increased, which is
dependent on sign of the coefficient (negative or positive respectively).
Limitations to the Model
This model W=ƒ ( Pmc , ED, Ai , En, G, M, AR, P, N, O, Ht, T, V,S, HS) + ei is a linear function
W= 1.922+ 0.197Pmc + 1.091AR 2 + 1.698 AR 3 – 0.633 En + 0.341 M1 + 0.560 M2 + 0.240 ED 2
+ 1.700 ED3 + 0.210S – 0.691O + 0.606 T + 0.105P -0052N-0.022 Ai + ei
therefore we are unable to distinguish between the wellbeing of two individuals who have the
same typology and wellbeing of an individual that may change over short time intervals that does
not affect the age parameter. As such in attempting to add further tenets to this model in order
498
that it is able to fashion a close approximation of reality, the following modifications are been
recommended.
Each individual’s wellbeing will be different even if that person’s valuation for quality of
life is the same as someone else who share similar characteristics. Hence, a variable P
representing the individual should be introduced to this model in a parameter α (p). Secondly,
the elderly’s wellbeing is different throughout the course of the year and so time is an important
factor. Thus, we are proposing the inclusion of a time dependent parameter in the model.
Therefore, the general proposition for further studies is that the linear function should
incorporate α (p, t) a parameter depending on the individual and time.
Summary
For this study, wellbeing is indeed a multidimensional concept. The paper has proven that the
determinants of wellbeing include psychosocial, environmental and demographic variables,
which is in keeping with the literature (3-12, 15, 18-20). This is a departure form the biomedical
model that emphasizes ‘dysfunction’ or diseases. The most fundamental assumption of this
model is the ‘absence of diseases’ means a healthy individual or a population. This implies that
reduced quality of life is only associated with increased illnesses. As early as 1946, the WHO
gave a definition of health which is an extensive one when this was compared to the traditional
operational definition (3). Because some scholars argue that this definition was too broad, it may
be the reason behind the Grossman’s model in 1972 (9, 10). Grossman used econometric
analysis to show some of the predisposed predictors of health. This was later expanded on by
Smith and Kington in 1997 (10), and later applies in a study on the elderly in Barbados by
Hambleton et al. (1) between 1999 and 2000. All those operational definition of wellbeing used
499
‘dysfunctions (or health conditions). The current study expanded on the operational definition of
wellbeing, and provides a list of determinants of wellbeing along with their degree of influence.
Based on the results of the model in Tables 1.1.2 and Table 1.1.3, we now have a model
that guide public health practitioners, and health professional in their policy formulation and
treatment of patient care.
In concluding, the general quality of life of the Jamaicans is a function of: area of
residence, cost of health care, psychological conditions- positive and negative affective
conditions, educational level, marital status, age and average occupancy per room, property
ownership, and social support. Therefore, treating an individual for illnesses, injuries, degrees of
injury is just a fraction of the components of those things that constitute their health and by
extension their wellbeing. It would have been good to include among those mentioned factors –
religion, and lifestyle practices such as smoking, alcohol consumption, exercise and diet within
the general model but this a limitation of the dataset. However, what is presented here are some
of the predisposed factors that determine the quality of life of a Jamaican. The elderly, despite
enjoying the company of their grandchildren and other family members, are not satisfied with the
invasion of their private spaces by large family size. This is further borne out in the fact that
positive psychological condition was the fourth most important determinant of quality of life.
Within this context, with the dearth of literature that has shown that biological ageing is directly
associated with increasing frailty and physical ailments, it should come as no surprise that the
cost of the health care was ranked third. The direct relationship between individual wellbeing
and cost of health care (i.e. β = 0.184) speaks to the literature that states that the ‘good health-
care’ can be bought. In that, the more wealth and individual has, the more he/she will be able to
purchase better health-care (i.e. medication, practitioners, skilled technicians, specialized care
500
and long-term care and so on), a gift that is not made available to the poor. The PIOJ and
STATIN reports have provided information on Jamaicans that the poverty has a geographic bias.
In that, poverty is substantially a Rural Zone phenomenon, and so it comes as no surprise that
‘Area of Residence’ happens to be the second most critical determinant of wellbeing. This
means that the elderly who resides in KMA has a higher probability of having a higher quality of
life than his/her counterpart who dwells in Other Towns and more so than those who live in
Rural Areas.
501
Reference
1. Erber J. Aging and older adulthood. New York: Waldsworth, Thomson Learning; 2005
2. Brannon L, Feist J. Health psychology. An introduction to behavior and health 6 th ed. Los Angeles: Thomson Wadsworth; 2007.
3. World Health Organization, WHO. Preamble to the Constitution of the World Health Organization as adopted by the International Health Conference, New York, and June 19-22, 1946; signed on July 22, 1946 by the representatives of 61 States (Official Records of the World Health Organization, no. 2, p. 100) and entered into force on April 7, 1948. “Constitution of the World Health Organization, 1948.” In Basic Documents, 15 th ed. Geneva, Switzerland: WHO, 1948.
4. Engel G. A unified concept of health and disease. Perspectives in Biology and Medicine 1960; 3:459-485.
5. Engel G. The care of the patient: art or science? Johns Hopkins Medical Journal 1977a; 140:222-232.
6. Engel G. The need for a new medical model: A challenge for biomedicine. Science 1977b; 196:129-136.
7. Engel G. The biopsychosocial model and the education of health professionals. Annals of the New York Academy of Sciences 1978; 310: 169-181
8. Engel, GL. The clinical application of the biopsychosocial model. American Journal of Psychiatry 1980.; 137:535-544.
9. Grossman M. The demand for health- a theoretical and empirical investigation. New York: National Bureau of Economic Research; 1972. In: Smith JP, Kington R. Demographic and economic correlates of health in old age. Demography 1997;34:159-170.
10. Smith, J. P., and R. Kington. 1997a. Demographic and economic correlates of health in old age. Demography 1997a; 34:159-170.
11. Smith JP, Kington R. Race, socioeconomic status, and health in late life. Quoted in L. G. Martin and B.J. Soldo. Racial and ethnic differences in health of older American, ed. Washington, DC: National Academy Press; 1997b.
12. Hambleton IR, Clarke K, Broome Hl, Fraser HS, Brathwaite F, Hennis AJ. Historical and current predictors of self-reported health status among elderly persons in Barbados. Rev Panam Salud Publica 2005; 17:342-353.
13. Papas RK, Belar CD, Rozensky RH. The practice of clinical health psychology: Professional issues. In: Frank RG, Baum A, Wallander JL, eds. Handbook of clinical health psychology (vol 3: 293-319. Washington, DC: American Psychological Association; 2004.
14. Dowling AS. Images in psychiatry: George Engel. 1913-1999. http://ajp.psychiatryonline.org/cgi/reprint/162/11/2039 (accessed May 8, 2007); 2005.
15. Brown TM. The growth of George Engel's biopsychosocial model. http://human-nature.com/free- associations/engel1.html. (accessed May 8, 2007); 2000.
16. Lynch P. Wellness. A National Challenge. Kingston: Grace, Kennedy Foundation Lecture 2003; 2003.
502
17. Planning Institute of Jamaica, (PIOJ) and Statistical Institute of Jamaica, (STATIN). Jamaica Survey of Living Conditions, 2002. Kingston: PIOJ and STATIN.
18. Longest BB, Jr. Health Policymaking in the United States, 3 rd ed. Chicago: Health Administration Press.
19. Bourne, P. Determinants of well-being of the Jamaican Elderly. Unpublished thesis, The University of the West Indies, Mona Campus; 2007a.
20. Bourne, P. Using the biopsychosocial model to evaluate the wellbeing of the Jamaican elderly. West Indian Medical J, 2007b; 56: (suppl 3), 39-40.
503
Table 1.1.1: Percentage and (count) of Marital Status by Gender of respondents
Details
Gender of Respondents
Males Females
Marital Status
Married 25.7 (2007) 24.7 (2043)
Never Married 69.4 (5421) 65.2 (5392)
Divorced 0.8 (64) 0.7 (59)
Separated 1.1 (85) 1.2 (100)
Widowed 3.0 (234) 8.1 (671)
Total 100 (7811) 100 (8235)
504
Table 1.1.2: A Multivariate Model of Wellbeing of Jamaicans Model
Dependent variable: Wellbeing of Jamaicans
Independent variables: Unstandardized coefficient
Standardardized coefficient
Constant 1.922Physical Environment -0.633* -.167*Positive Affective Conditions .105* .131*Negative Affective Conditions -.052* -.085*lnCost of medical (Health) care 0.197* 0.128*Area of Residence 2 (1=KMA) 10.91* .233*Area of Residence 3 (1=Other Towns) 1.698* .209*Age -0.022* -0.207lnAverage occupancy per room -0.691* -0.254*marstatus1 (1=Divorced, separated, widowed) 0.341* 0.075*marstatus2 (1=Married) 0.561* 0.141*House Tenure -0.081Land Ownership 0.606* 0.145*Crime 0.008Edu_Level2 (1=Secondary) 0.240* 0.061*Edu_Level3 (1=Tertiary) 1.700* 0.156*Dummy gender (1=male) 0.060Seeking Health care 0.055Social Support 0.210* 0.054*N= 1146R = 0.634Adjusted R2 = 0.393Error term = 1.5 F statistics [18,1128] = 42.126ANOVA = 0.001* significant p value < 0.05
505
Table 1.1.3: Decomposing the 39.3% of the variance in Wellbeing of Jamaicans, using the squared partial correlation coefficientVariables Percentage
Average occupancy per room 7.0
Area of residence (1=KMA) 7.0
Area of residence (1=Other Towns) 6.4
Individual’s educational attainment (1=Tertiary) 3.4
Individual’s educational attainment (1=Secondary) 0.5
Psychological state – Positive Affective conditions
- Negative Affective conditions
2.4
1.0
Age of respondents 3.4
Marital status – (1=married)
- (1=separated, widowed, divorced)
1.0
0.5
Physical environment 3.4
Cost of health care 2.4
Property ownership (excluding owing a house) 2.9
Social support 0.5
506
APPENDIX XXVI – Sampled Research Paper II
Factors that Predict Public Hospital Health Care Facilities Utilization in Jamaica: Are there Differentials of Health Care Hospital Care Facility Utilization By Income Quintiles and Area of
Residence?
Paul Andrew Bourne49
Department of Community Health and PsychiatryFaculty of Medical Sciences, Mona, Kingston &, Jamaica W.I.
49 Corresponding author: Paul Andrew Bourne can be contacted at the Dept of Community Health and Psychiatry, Faculty of Medical Sciences, The University of the West Indies, Mona, Jamaica. Or by emailing [email protected] or telephoning 876-467-6990.
507
AbstractObjective: Health is a crucible component in any discussion on development, and public-private hospital health care utilization accommodates this mandate of governments. The aim of the current study is to examine factors that account for people’s public hospital health care facilities utilization in Jamaica, and to ascertain whether is a difference between public hospital care utilization and income quintile and area of residence.
Method: The current study has extracted a sub-sample of 1,936 respondents from a national survey of 25,018 respondents. The sub-sample constitutes those respondents who had indicated visits to public hospital facilities for health care or private hospital health care facilities owing to self-reported ill-health. It is taken from a larger cross-sectional survey which was conducted between June and October 2002. It was a nationally representative stratified probability survey of 25,018 respondents. The data were collected by a comprehensive self-administered questionnaire, which was primarily completed by heads of households on all household members. The questionnaire is adopted from the World Bank’s Living Standards Measurement Study (LSMS) household surveys and was modified by the Statistical Institute of Jamaica with a narrower focus and reflects policy impacts. Chi-square, t-test and analysis of variance (ANOVA) were used for bivariate relationships, and logistic regression was used to explain factors that determine who attended public hospital health care facilities.Findings: The current findings revealed that 6 factors determine 35.6% of the variability in visits to public hospital health care facilities utilization in Jamaica. Two major findings from this study are 1) health seeking behaviour and health insurance coverage are the two most significant factors that determine public hospital health care facilities utilization, and that 2) the two aforementioned factors and positive affective conditions inversely correlate with public hospital health care facility utilization. In addition to the above, there is no statistical difference between the utilization of public hospital health care facilities and area of residence while lower income quintile becomes the greater public hospital health care facilities utilization has been.Conclusion: The demands for public hospital health care facility utilization in Jamaica are primarily based on inaffordability and low perceived quality of patient care. The issue of low quality of patient care speaks not to medical care, but to the customer service care offered to clients. The greater percentage of Jamaicans who access private health care is not owing to plethora of services, higher specialized doctors, more advanced medical equipment, or low, but this is due to social environment – customer service and social interaction between staffers and clients- and physical milieu – more than one person per bed sometimes, uncleansiless of the facilities. Keywords: Public-private hospital health care utilization, Public health care demand, Health care facility utilization, Jamaica
508
Introduction
Health is a crucible component in development. The health status of a people does not only mean
personal development; but also greater economic development for the nation. As healthier people
are more likely to produce greater output than those who are ill, Accounting for higher
productivity and efficiency. Ill/injury means in-voluntary absenteeism which accounts again for
lowered production. A substantial part of a country’s Gross Domestic Product (GDP) per capita
each year is loss to illnesses. The WHO has forwarded that between 3 and 10 years of life is loss
owing to illnesses (1,2), suggesting that illness reduces not only output by quality of life. Hence,
it is not important for observed length of life (ie. life expectancy), but it is imperative to take into
consideration loss years owing to illness which means the measure of importance will be health
life expectancy. And so, the public health facility can accommodate this mandate of
governments. While private health care facilities supply a demand for health care, the average
citizen in many countries is unable to afford the medical expenditure of those facilities and so the
public care facility is not only the access of the average person is the bedrock upon which the
health care system of the society relies.
Public-private hospital health care utilization in Jamaica for over the last 11-years (1996
to 2006) has been narrowing, suggesting that economic wellbeing of population has been falling
as the economic cost of survivability has been increasing and this explain the narrowing gap
seeing in the hospital health care facility utilization (Figure 1). It is noted in the data that there is
decline in medical care seeking behaviour of Jamaicans in 2006 from 70% to 66% in 2007 (In
Table 2). Although there is an increasing demand of public hospital health care facilities
utilization by those who seek medical care (Table 1), within the context of an increase in self-
reported illness (by 3.3%) coupled with the dialectic of reduction in medical care seeking
509
behaviour, and decline in public health utilization (including clinics, Table 1), there is still a
positive sign as there was increase in health insurance coverage (from 21.2% in 2007 over 18.4%
in 2006).
In 2007 inflation increased by 194.7% over 2006 and accounts for this narrowed gap
between public and private utilization of health care in Jamaica. The exponential increase in
inflation (194.7%) has accounted for higher cost of living of Jamaicans and has rationalized the
decline in private health utilization and the switching to public health care utilization (Table 3).
Furthermore, this goes to the core of the drastic reduction in the bed occupancy at public hospital
health care facilities in 2004 over 2003 (by 33.7%), suggesting that the poor’s medical care
seeking behaviours are significantly affected in tough times. This is further accounted for in the
fact that data on private facilities utilization for those in the poorest quintile fell by 36.1% in
2007 over 1991 and 37.1% for those in the poor quintile over the same period, while there was
an increase in public facilities utilization for those in the poorest quintile (by 29.8%) and by
53.6% for those in poor quintile for the same period.
Inflation is not the only economic impediment that is affecting health care utilization in
Jamaica, as looking at the data on remittances which accounted for the single largest foreign
exchange receipt in the nation, this fell by 7.7% in 2007 over 2006 (Figure 2). The poor and the
poorest were the most affected by the decline in remittances as rate was 22.1% and 16.9%
respectively. Despite the reduction in remittances in Jamaica, 41.8% of Jamaican received
monies this way, which means that a 7.7% decline of those people whom received remittance
affect some 206,522 Jamaicans which include the most vulnerable such as the poor, children,
unemployable elderly and youths. When inflation is coupled with reduction in remittances, given
that remittance substantially contribute to the economic income for the poor and the poorest
510
quintile more than the other upper quintiles, this mean that health and health seeking behaviour
in the poor-to-the-poorest people will take a back seat to consumption expenditure on food and
non-alcoholic beverages (3).
Comparatively there has been a marginal increase in private health care facilities
utilization by 6.5% of those in the wealthiest quintile, a substantial increase (by 31%) for those in
the wealth quintile (quintile 4), and a mild decline by 0.47% for those in quintile 3 (middle
quintile). Nevertheless, there is a 3.9% increase in public health care facilities utilization for
those in the wealthiest quintile, while the middle to wealth quintiles showed increases. Therefore,
emerging from these findings is a particular social profile of people who attend public health
care facilities in Jamaica as in excess of 62% of those in middle-to-wealthiest quintiles attended
private health care facilities compared to 66% and more of those in the poor-to-poorest quintile
(Table 3).
In 2007, 50.7% of those in the poorest quintile indicated that they were unable to afford
to seek health care for ill/injury compared to 36.7% of quintile 2, 34.4% in quintile 3, 21.4% in
quintile and 7.1% of those in the wealthiest quintile. Adults sometimes may not attend medical
facilities for care, but they will take their children because they are protective of them. This is
revealing about affordability as in 2007, 51.7% of those in the poorest quintile indicated that they
sought medical care for their children (0-17 years), 52.7% in quintile 2, 61.2% in quintile 3,
61.8% in quintile 4 and 67.6% in the wealthiest quintile. Is in-affordability an issue in medical
care utilization for those in the poorest to poor quintiles?
The mean annual amount spent on ‘food and beverage’ in 2002 by those in the poorest
quintile was 50.4 per cent compared to 38.1 per cent of those in the wealthiest quintile. The mean
annual amount expended on the same in 2006 rose by 3.6 per cent for those in the former
511
quintiles compared to reduction of 0.1 per cent for those in the latter group. (3). Medical
expenditure which is a constituent of non-consumption expenditure was 2.2% for those in the
poorest quintile (in 2006) compared to 13.5% of wealthiest quintile. The economic well-being of
the poor and the poorest in the population has become even more graved as this is reflected in the
inflation rate as it increased by 3 times for 2007 over 2006 (4). While the down turn the United
States economy in particular the Jamaica economy has more than one-half since 2006 (growth in
GDP at Constant (1996) prices in 2006 2.5 per cent and 1.2 per cent in 2007), those in the
poorest quintiles are hard hit by this economic recession, explaining the rationale for the
switching to home care or more public care.
All the aforementioned arguments omit area of residence, suggesting that this is the same
across geographical boundaries. Poverty has been decline since 1991 from 44.6%, when inflation
rate was at the highest in the history of the nation (80.2%), to 9.9% in 2007. However, rural
poverty which was 71.3% in 2007 saw an 8.5% increase over 2006 (65.7%) within the economic
environment of a drastic increase in inflation, cost of living and prices of non-consumption items
such as medical care. When we take into consideration the reduction of remittance by 8.7% in
2007 over 2006 (42.3%) and fact that 67% of the elderly (people age 60+ years) dwell in rural
zones, remittance represents not only an income but economic living. Is this Accounting for any
of the narrowing of the gap between public-private hospital health care facility utilization? And
what are the factors which explain public hospital care facilities utilization in Jamaica? This is
the first study in the English speaking Caribbean and in particular Jamaica to seek to examine
conditions that explain public hospital health care facility utilization. Hence, the aim of the
current study is to examine factors that account for choice of public hospital care facilities
512
utilization and to ascertain whether there is a difference between public hospital care utilization
and income quintile and area of residence.
Method
The current study extracted a sub-sample of 1,936 respondents from a national survey. The sub-
sample constitutes those respondents who indicated having visited public and private hospital
health care facilities for medical treatment owing to ill-health. The sample is taken from a larger
cross-sectional survey, which was conducted between June and October 2002. It was a nationally
representative stratified probability survey of 25,018 respondents. The sample (N=25,018 or
6,976 households out of a planned 9,656 households) was drawn, using a 2-stage stratified
random sampling technique, involving a Primary Sampling Unit (PSU) and a selection of
dwelling from the primary units. The PSU is an Enumeration District (ED), which constitutes a
minimum of 100 dwellings in rural areas and 150 in urban zones. An ED is an independent
geographic unit that shares a common boundary. This means that the country was grouped into
strata of equal size based on dwellings (EDs). Based on the PSU, a listing of all the dwellings
were made and this became the sampling frame from which a Master Sample of dwellings were
compiled and which provides the frame for the labour force. The survey adopted was the same
design as that of the labour force.
The national survey was a joint collaboration between the Planning Institute of Jamaica
and the Statistical Institute of Jamaica. The data were collected by a comprehensive self-
administered questionnaire, which was primarily completed by heads of households on all
household members in Jamaica. The questionnaire was adopted from the World Bank’s Living
Standards Measurement Study (LSMS) household surveys and was modified by the Statistical
513
Institute of Jamaica with a narrower focus and reflects policy impacts. The instrument assessed:
(i) general health of all household members; (ii) social welfare; (iii) housing quality; (iv)
household expenditure and consumption; (v) poverty and coping strategies, (vi) crime and
victimization, (vii) education, (viii) physical environment, (ix) anthropometrics measurement and
Immunization data for all children 0-59 months old, (x) stock of durable goods, and (xi)
demographic characteristics.
Data were stored and retrieved in SPSS 15.0 for Windows. The current study is
explanatory in nature. Descriptive statistics were forwarded to provide background information
on the sampled population. Following the provision of the aforementioned demographic
characteristics of the sub-sample, chi-square analyses were used to test statistical association
between some variables; t-test statistics and analysis of variance (ie ANOVA) were also use to
examine the association between a metric dependent variable and either a dichotomous variable
or non-dichotomous variable respectively. Logistic regression was used to examine the statistical
association between a single dichotomous dependent variable and a number of metric or other
variables (Empirical Model). In order to test the association between a single dichotomous
dependent variable and a number of explanatory factors simultaneously, the best technique to use
was logistic regression.
Empirical Model
Given a plethora of factors that simultaneously affect health care visits, the use of bivariate
analyses will not capture this reality. Therefore, in order to capture those factors that influence
visits to public hospital health care facility, we used a logistic regression instead. The regression
model examines several factors that might affect visits to public health care facilities.
514
The data source was from the Jamaica Survey of Living Conditions of 2002 on health,
consumption, social programme, physical environment, education, public-private hospitalization
utilization, and crime and victimization. The rationales for the use of 2002 data were (1) it was
the second largest national representative survey that was conducted in the history of data
collection by the Statistical Institute of Jamaica and the Planning Institute of Jamaica to assess
policy impacts (25,018 respondents), and (2) it was inclusive of issues on crime and
victimization, and physical environment that were not in the post-2002 survey, nor the preceding
years. Although there are more recent data (2004 to 2007), these have excluded many of the
factors that are present in the 2002 data ( that is physical milieu, crime, victimization and mental
health), and wanting to establish factors that influence health care, we needed more possible
factors that less as well as crime and victimization as these are crucible issues that have been
facing the country increasingly since 2002.
Ergo, the 2002 consist of more possible factors that determine people’s decision to visit
public hospital health care facilities utilization compared to private hospital health care facilities
utilization. Explanatory factors include psychological factors conditions self-reported health
insurance coverage; area of residence; educational level; and other variables. The basic
specification for the model was:
VPHCFi = ƒ (αjiDEMi, βjiPSYi, ƏPmci, πSSi, γjiHSBi, εi) (1)
Where VPHCFi is visits to public or private hospital health care facilities of person i is a
function of demographic vector factors, DEMi; psychological factors of person i, PSYi, medical
expenditure, Pmc; social support of individual i, SSi; health seeking behaviour of person i, HSBi; εi
is the residual term. Αji, βji, γji, are coefficient vectors for person i of variables j and Əi, π, are
coefficient of vector for person i. VPHCFi is a binary variable, where 1= self-reported visits for
515
public hospital health care facilities for medical care and 0=self-reported visits to private hospital
health care facilities. [I am not so clear on this sentence].
Measure
Public Hospital Health Care Utilization variable measures the total number of self-reported cases
of visit to either public hospital health care facilities or private hospital health care facilities in
the last 4-weeks ( whereby the survey period is used as the reference point). Public Hospital
Health utilization was dummied to read 1=visits to public hospital health care facilities, and
0=private hospitals health care facilities.
Income Quintile Categorization. This variable measures the per capita population income
quintile that each individual is categories. There are 5 categories, from the poorest to the
wealthiest income quintile. For the purpose of the regression analysis, the variable was
measured as:
1= Middle Quintile, 0=otherwise
1=Two Wealthiest Quintiles, 0=otherwise
The referent group is the two poorest income quintiles
Crowding. This is the total number of persons living in a room with a particular household.
, where represents each person in the household and r is is the number of
rooms excluding kitchen, bathroom and verandah.Age: This is a continuous variable in years, ranging from 15 to 99 years.
Positive Affective Psychological Condition: Number of responses with regards to being
optimistic about the future and life generally.
Negative Affective Psychological Condition: Number of responses from a person on having loss
a breadwinner and/or family member, loss of property being made redundant, failure to meet
household and other obligations.
516
Private Health Insurance Coverage (or Health Insurance Coverage) proxy Health Seeking
Behaviour is a dummy variable which speaks to 1 if self-reported ownership of private health
insurance coverage and 0 if did not report ownership of private health insurance coverage.
Health Seeking Behaviour. Visits to health care practitioners outside of illnesses, dysfunctions,
and injuries. This is a binary variable where 1 = self-reported seeking medical care and 0 = not
reporting seeking medical care
ResultsThe sub-sample for the current study was 1,936 respondents of which 39.4% were males
(N=762) and 60.6% females (N=1,174), suggesting that females are 1.5 times more likely to seek
medical care from public or private hospitals compared to males. The findings (indicated in
Table 4) revealed that marginally more Jamaicans who visited hospital facilities for medical care
went to public facilities (53%, N=1,021). In addition to the aforementioned issues, 56%
(N=1,086) of the sample reported health care insurance coverage compared to 44% (N=850) who
did not. The mean age of the sample was 44 years (SD=27.5 years). Some 45% of the
population were never married (N=671), 36% married (N=532), and 20% were divorced,
separated or widowed. Furthermore, Table 4 reveals that two-thirds of the population dwelt in
rural Jamaica, 22% (N=424) in Other Towns and 12% Kingston Metropolitan area (N=223).
On the matter of the psychological state of Jamaicans, this was classified into two main
conditions - positive and negative psychological conditions. The mean negative psychological
conditions of population was 4.9 (out of 16, SD=3.3), suggesting that the negative psychological
conditions of the population was low. On the other hand, the mean value for the positive
affective psychological condition of the population was 3.2 (out of 6, SD = 2.4) indicating that
positive affective conditions of the population was moderate (Table 4).
517
The examination between public-private hospital health care facility utilization and area of
residence found no statistical correlation between the two aforementioned variables – χ 2(2)
=0.385, ρ-value=0.825 > 0.05 – (Table 5). The no correlation between the two conditions
indicates that Jamaicans, irrespective of their places of abode attended public-private hospital
health care facilities for care of ill-health. (Table 5)
A cross tabulation between visits to health care facilities and per capita population income
quintile showed a statistical association - χ 2(4)=157.024, ρ-value <.001. The findings revealed
that people in the poorest income quintile was 2.4 times more likely to visit public health care
facilities compared to those in the wealthiest per capita income quintile; people in the poorest
income quintile was 1.5 times more likely to visit public facilities compared to those in the
second wealthiest quintile. However, the findings revealed that those in the second poorest
income quintile indicate no statistical difference themselves and those in the middle income
quintile - quintile 3 (Table 6). Nevertheless, people in the poorest income quintile were 1.3 times
more likely to visit public facilities compared to those in the middle income quintile. There is a
substantial difference between those who visit public health institutions, who are in the poorest
income quintiles (73.8%, N=251) and those in the second poorest income quintile (58.4%,
N=208). Embedded in the aforementioned finding is the increase in switching from public to
private hospital health care facilities the more income quintile shifts to the wealthiest category
(Table 6). The aforementioned findings, raise concern about the extent of public-private hospital
health care expenditure
518
Of the sample (N=1,707), 912 people visited private hospital health care facilities and reported
that they spent on average $2,977.41 (SD=$4,053.01) compared to $1,376.12 (SD=$2,547.93,
N=1,019) for a visit to a public hospital care facility, suggesting that those who attend private
hospital health care institutions spent about 2.2 times more than those who visit the public
hospital health care facilities. Using t-test analysis, there is a difference between expenditure on
public hospital health care and private hospital health care – t10.5 [1929] = ρvalue < 0.001.
Using analysis of variance (ANOVA), generally, it was found that a statistical association exists
between negative psychological conditions and per capita income quintile (F statistic [4, 1926]
=28.793, ρ-value< 0.001). (Tables 7.1 – 7.2). Further investigation of the negative affective
conditions by per capita quintile revealed that there is no difference between the negative
affective psychological conditions of those in three bottom quintiles (quintiles 1 to 3), ρ-value >
0.05 (Table 7.2). In addition to the aforementioned issue, there is no difference between the
negative psychological state of people in quintiles 3 and 4 (ρ-value>0.05) and quintiles 1, 2 and
3, indicating that negative affective conditions can be classified into 3 groups (1) high for those
in quintiles 1, 2 and 3; (2) moderate for quintile 4 and (3) low for those in quintile 5. However
those classified in quintile 5 has the lowest negative affective conditions compared to those in
the other quintiles (ρ-value<0.001). Embedded in this finding is that as people move to the
wealthiest quintile, they experience less negative trauma such as the loss of breadwinner, owing
to abandonment, death or incarceration, crop failure, redundancy, loss of remittances, inability to
meet household expenses, and less hopeless about the future.
519
There is statistical association between positive affective psychological conditions and per capita
income quintile - F statistic [4, 1492] =12.366, ρ-value< 0.001. (Table 8.1). Further
examination of the two aforementioned variables revealed that there is no statistical difference
between the positive affective psychological conditions for those in quintiles 1 and 2; and
between quintile 2 and quintiles 3 and 4. Hence the statistical difference in positive affective
conditions is between those who are classified into two poorest quintiles and those in the wealthy
quintiles (Table 8.2).
Overall, there are statistical differences among health care expenditure of rural, urban and
periurban residences in Jamaica – F-statistic [2, 1928] = 4.902, ρvalue < 0.001. Rural area
dwellers spent on an average $2,009.98 (SD=$2,999.88, N=1286) per visit on medical care
compared to peri-urban residents who spent $2,593.13 (SD=$4,587.67, N=423) and $1,963.68
was spent by urban dwellers (SD=$3,188.31, N=222). Further examination revealed that there is
a difference between the medical expenditure made by rural residence and those in other towns –
p value <0.05. The former on an average spent $583.17 less than those in other towns.
However, there are no statistical differences between medical expenditure of urban residents and
that of rural dwellers (ρvalue >0.05) and other towns (ρvalue >0.05).
Empirical Results
The regression analytic model was established in order to simultaneously examine a number of
explanatory variables’ impact on those who attend public hospital health care facilities for ill-
health. Table 6 and Table 7 provide information on empirical model (Eq (1)) and in the process
answers the suitability of the model ( Table 6), while Table 7 answers to the question of which of
520
the variables are factors and their importance. Before embarking on the report of the regression
model which contains all the predisposed variables and which those that are statistical significant
(ie pvalue<0.05), we will examine the ‘goodness’ of fit of the data in regard to the model.
Table 6 reports a ‘classification of visits to hospital health facilities owing to ill-health’
and contained examination of observed compared to predicted classification of the dependent
variable (that is visits to hospital health care facilities in due to negative health). Of the 1,051
respondents that were used to establish the model (using the principle of parsimony, that is only
those variables that have a pvalue < 0.05 will be used in the final model), 73% (N=767) were
correctly classified: 71.6% (N=374) of those who visit private hospital health care facilities for
care owing to illnesses or injuries and 74.3% (N=393) of those who visited public hospital health
care institutions for treatment of dysfunctions or injuries. Therefore, the data is a ‘good’ fit for
the model (ie. 73% were correctly classified).
Table 10 contained the answers the empirical model (Eq. (1))
VPHCFi = ƒ (αjiDEMi, βjiPSYi, ƏPmc, πSSi, γjiHSBi, εi) (1)which shows that 35.6% of the variability in visits to health facilities for care are affected by a
number of factors- Chi-square (24) = 326.58, p-value < 0.001, -2Log likelihood = 1130.37. Of all
the demographic variables contained in the current study, only total expenditure was found to be
a factor of visits to public hospital health care facilities for ill-health (Wald statistic=4.458;
OR=1.00: 1.00, 1.00). The cost of medical care was directly related to reason for patients’ visits
to public hospital health care facilities for treatment against ill-health (Wald statistic=13.959;
OR=1.00: 1.00, 1.00) likewise was the positive statistical relationship between social support and
visits to health care facilities (Wald statistic=13.419; OR=1.741: 1.29, 2.34). A direct
association was observed between negative affective psychological conditions and visits to
521
public hospital health care facilities. This suggested that more the patients/individuals are
impacted upon by the loss of a breadwinner, crop failure, redundancy, loss of remittances.
On the other hand, people who have access to private health insurance coverage (Wald
statistic=89.35; OR=0.134: 0.089, 0.204), visited a health practitioners for non-ill checks (Wald
statistic=72.07; OR=0.494: 0.419, 0.581), and a positive affective psychological conditions
(Wald statistic=4.74; OR=0.931: 0.874, 0.993) are more likely not to attend public hospital
health care facilities. These issues are all preventative and optimistic measures which are directly
related with switching away from public to private hospital health care facilities. Embedded in
these findings (based on Table 5.2) is the fact that optimistic in the study are those in the middle
to the upper class. This study has shown that there is no distinction between the positive affective
psychological conditions of those patients who are classified in the middle to the wealthiest
class, but there is a difference between the aforementioned group and those in the poor classes
(ie. quintiles 1 to 2 – poorest to poor classes).
Therefore, in addressing the issue of using self-reported health (subjective health or
wellbeing) to evaluate health (or wellbeing), it is imperative to note that there is an old
cosmology that forwards that subjective assessment of health (self-reported health) is not a good
measurement to apply to health or wellbeing. In this section of the study that discourse will not
be examined as it will be done in the discussion; however, we must briefly compare and contrast
self-reported visits to public facilities data collected by the Planning Institute of Jamaica and the
Statistical Institute of Jamaica (in Jamaica Survey of Living Conditions, JSLC) and actual data
collected by the Ministry of Health Jamaica for the period of 1996 and 2004.
Using actual visits to public facilities (in Ministry of Health, Jamaica Annual Report) and
that of self-reported visits to the same institutions, the data revealed that generally the statistics
522
as collected by the Planning Institute of Jamaica and the Statistical Institute of Jamaica (in
Jamaica Survey of Living Conditions, JSLC) reveals health status and conditions of Jamaicans.
Based on Table 9, in 1997, the actual visits to public facilities were 33.1% as reported by the
Ministry of Health and the self-reported figure for the same period was 32.1% (in JSLC). The
difference between the actual and the subjective visits was 1%, which has no statistical
difference. Some eight years post 1997 (2004), another comparison was made to assess whether
the self-reported data is still good to use to proxy not only perception but reality of hospital
health care facility utilization in Jamaica. The figures were 52.9% for actual visits and 46.8% for
subjective visits. This indicates that in 2004 Jamaica marginally report lower visits to facilities
(6.1%) than the data published by the Ministry of Health. Despite the under reporting of health
visits to public facilities in 2004 in Jamaica, there is no statistical difference between the year
and the figures by the aforementioned institutions – χ 2(4) =157.024, ρ-value <0.05
Conclusion
Health seeking behaviour ( ownership of private health insurance coverage and visited a
health practitioners for non-ill checks) is the most important factor that determines visits to
public health facilities or private health facilities for care for illnesses (or injuries). Following the
value of health seeking behaviour is the cost of medical care; reinforcing the reality for financial
inability among people is it lower class, middle class or upper class will see a switching from
private to public facilities for ill-treatment. In continuing this discourse, social support is directly
related to visits to public hospital health care facilities and so offers some explaining for the large
number of people visiting the said institutions to support the patients who visit for treatment of
negative health conditions. Again the positive association that exists between expenditure and
visits to public facilities further reinforces the point that the more people spent which is the less
523
income they have for saving and further speaks about the poor, they will be less likely to visit
private hospital health care facilities. The poor who are less hopeful about the future (unlike
those in the middle class) are more optimistic because of financial stability and are ergo able to
access private hospital health care because of expenditure of private health care does intimate
better health care, which they are willing to pay for.
Table 11: Public Hospital Facility Visits (using the JSLC and Ministry of Health Jamaica) By 1997 and 2004
Public Facilities in Jamaica
Year Actual Visits, MOH1 Self-reported Visits, JSLC% %
1997 33.1 32.1
2004 52.9* 46.8
Source: Ministry of Health Jamaica and the Jamaica Survey of Living Conditions (JSLC)χ 2(4) =0.083, ρ-value > 0.05
1 The Percentages of Actual visits were computed by Paul Andrew Bourne*Preliminary data were used to calculate this percentage
Discussion
In view of life expectancy for both genders in Jamaica (71.3 for males and 77.1 for females) (5),
this study indicates that health status of the populace are high as life expectancy means living or
denying the odds of disease causing pathogens. In order for a populace to defy the odds of
morality or to delay it, the following life expectancy precursors must be considered; namely:
healthy lifestyle behaviour or levels of health seeking behaviour, and hospital health care facility
must meet universal health standard. The foregoing suggests that health seeking behavior and
hospital health care facility utilization, plays a crucial role in embracing such reality. In 2007,
524
Jamaicans sought less medical care for ill-health by 4% over 2006 (70%) They reported more
health conditions over the same period (15.5% in 2007 and 12.2% in 2006). Although this is
suggesting that they are using more home (or herbal) remedy, It leaves concern about health
care facilities utilization and factors that may be Influential.
Data on health care facilities utilization in Jamaica have been reported on and so this
paper is seminal.. Over the last 2 decades (ending 2007), Jamaicans preference for private
hospital health care facility utilization has been lower, narrowing towards public facility
utilization. Within the global economic climate which is Accounting for the lowered remittances
(3), people must spend more for increased consumption goods while at the same time,
maintaining good health. The World Health Organization (WHO), in recognizing the role of
income on health, postulated that the unfinished agenda for health, poverty remains the main
item (6), thus suggesting that poverty means increased hunger, malnutrition and by extension ill-
health. This study evidences that there is a correlation between public-private hospital health
care facility utilization and per capita income quintiles which is inkeeping with the literature (6-
17). The data showed that 74% of those in the poorest quintile used public facilities compared to
31.3% of those in the wealthiest quintile. Embedded in the hospital health care facility
utilizations are socio-demographic characteristic (social standing) of demanders. Some 2.8 (≈3)
more people of the poorest quintile attended public facilities than private facilities, and that 2.4
more of the poorest than the wealthiest people attended the former than the latter facilities.
The typological of hospital health care facility utilization in the nation is a reflection of
inability (ability) and than inflation (increase prices) will substantially lower the poorest demand
for medical care. It is well established in the literature that income affects health, and lower
income direct correlates with poor health (7), which was reinforced in a study conducted by
525
Powell, Bourne and Waller (8) who found that the those in the lower subjective social class
reported the least health status. Those in the poorest income quintile are more concerned and able
to primarily have difficulty purchasing the necessary nutrients from the required foods groups,
and this accounts for their high consumption of public facilities, owing to low cost medical
services. This study found that the cost of medical care strongly correlated with public hospital
health care facility utilization, and further explains this potency as it was revealed that the more
people spending, the more they will attend public facility. An individual who spends more has
less income to save as well as use for medical expenditure that account for increased utilization
of private facility with movement along the rung of per capita income quintile.
With less income coupled with more spent on consumption items, health seeking medical
behaviour becomes less. Within this reality, the negative correlation between health seeking
behaviour and public hospital health care facility utilizations expected as public facility demand
is strongly correlated with income or affordability of health care. Private facility consumption
depends on one’s ability to pay the cost for the care, and it is this which bars the poorest from
highly accessing this facilities. This study has revealed that public hospital health care facility
utilizations substantially demanded by the poorest and those who are experiencing negative
affective conditions and positive affective psychological conditions.
Studies have shown that one psychological state affects his/her health (18-21). This was
further refined into negative and positive affective conditions (18, 21,22). Being positive
directly correlated to health as people who entertain positive affective conditions are more likely
to view like a more optimistic manner and this enhance their health status. In seeking to unearth
‘why some people are happier’ Lyubomirsky (21) approached this study from the perspective of
positive psychology. She noted that, to comprehend disparity in self-reported happiness between
526
individuals, “one must understand the cognitive and motivational process that serves to maintain,
and even enhance happiness and transient mood’ (21). Using positive psychology, Lyubomirsky
identified comfortable income, robust health, supportive marriage, and lack of tragedy or trauma
in the lives of people as factors that distinguish happy from unhappy people, which was
discovered in an earlier study by Diener, Suh, Lucas and Smith (23). In an even earlier study by
Diener, Horwitz and Emmon (24), they were able to add value to the discourse of income and
subjective well-being. They found that the affluent (those earning in excess of US 10-million,
annually) self-reported well-being (personal happiness of the wealthy affluent) was marginally
more than that of the lowly wealthy.
Studies revealed that positive moods and emotions are associated with well-being (20) as
the individual is able to think, feel and act in ways that foster resource building and involvement
with particular goal materialization (21). This situation is later internalized, causing the
individual to be self-confident from which follows a series of positive attitudes that guide further
actions (25). Positive mood is not limited to active responses by individual, but a study showed
that “counting one’s blessings,” “committing acts of kindness”, recognizing and using signature
strengths, “remembering oneself at one’s best”, and “working on personal goals” all positively
influence well-being (25, 26). Happiness is not a mood that does not change with time or
situation; hence, happy people can experience negative moods (27,28).
This takes the study to the next area, psychological conditions and per capital income
quintile. Those with negative psychological conditions are from the lower class (poor), and
studies have shown that there is a correlation between health and psychological conditions. Now,
additional issues have emerged from this study as poor are negative and attend public facility
more than those at the greater per capita income quintile. On the other hand, those who are more
527
likely to report positive affective psychological conditions are greater for those at the highest
level of the income quintile, the findings also show that those who attend private facility are
experience greater positive conditions. It follows that public facilities in Jamaica service and
service quality are more in keeping with particular psychological state and subjective social
class. Hence, private facilities are not only more expensive but the service that it affects is in
keeping with the high social standings of its clients, and the reverse is equally the case for public
facilities staffers and their clients.
In summary, the demands for public hospital health care facility utilization in Jamaica are
primarily based on in affordability and low perceived quality of patient care. The issue of low
quality of patient care speaks to not medical care, but to the customer service care offered to
client. The greater percentage of Jamaicans who access private health care is not owing to
plethora of services, higher specialized doctors, more advanced medical equipment, or low, but
this is due to social environment – customer service and social interaction between staffers and
clients- and physical milieu – more than one person per bed sometimes, uncleansiless of the
facilities. These issues accommodate for the lowly particular persons visiting public and private
facilities for medical care.
Acknowledgement
The researcher would like to extend sincere gratitude to staff of the documentation centre at the
Sir Author Lewis Institute of Social and Economic Studies, Faculty of Social Sciences,
University of the West Indies, Mona, Jamaica for making available the dataset from which this
study was based.
528
Reference
1. World Health Organization, (WHO). WHO Issues New Healthy Life Expectancy Rankings: Japan Number One in New ‘Healthy Life’ System. Washington D.C. & Geneva: WHO; 2000.
2. WHO. Healthy life expectancy. Washington DC: WHO; 2003.3. Planning Institute of Jamaica (PIOJ) & Statistical Institute of Jamaica (STATIN).
Jamaica Survey of Living Conditions, 1988-2008. Kingston: PIOJ & STATIN; 1988-2008.
4. Planning Institute of Jamaica (PIOJ). Economic and Social Survey of Jamaica, 2007. Kingston: PIOJ; 2008.
5. STATIN. Demographic Statistics, 2004, 2008. Kingston: STATIN; 2006, 2008.6. WHO. The World Health Report 1998: Life in the 21st Century. A Vision for All.
Geneva: WHO; 1998.7. M. Marmot. The influence of Income on Health: Views of an Epidemiologist.
Does money really matter? Or is it a marker for something else? Health Affairs 21 (2002), pp.31-46.
8. L.A. Powell, P. Bourne, L. Waller. Probing Jamaica’s Political Culture, vol. 1: Main Trends in the July-August 2006 Leadership and Governance Survey. Kingston: Centre of Leadership and Governance, Department of Government, University of the West Indies, Mona; 2007.
9. Hambleton IR, Clarke K, Broome HL, Fraser HS, Brathwaite F, Hennis AJ. 2005. Historical and current predictors of self-reported health status among elderly persons in Barbados. Rev Pan Salud Public. 2005;17:342-352.
10. Smith JP, Kington R. 1997. Demographic and economic correlates of health in old age. Demography. 1997; 34(1):159-170.
11. Grossman M. The demand for health- a theoretical and empirical investigation. New York: National Bureau of Economic Research; 1972.
12. Benzeval, M., K. Judge, and S. Shouls. 2001. Understanding the relationship between income and health: How much can be gleamed from cross-sectional data? Social policy and Administration.
13. Case, A. 2001. Health, Income and economic development. Prepared for the ABCDE Conference, World Bank, May 1-2, 2001. http://www.princeton.edu/~rpds/downloads/case_economic_development_abcde.pdf (accessed 6 June 2001).
14. Diener, E., E. Sandvik, L. Seidlitz, and M. Diener. 1993. The relationship between income and subjective well-being: Relative or absolute? Social Indicator Research 28:195-223.
15. Kawachi, I. 2000. Income inequality and health. In social epidemiology, ed. L. K. Berkman and I. Kawachi. New York: Oxford Univeristy Press.
16. Kawachi, I., B. P. Kennedy, K. Lochner, and D. Prothrow-Stitch. 1997. Social capital, income inequality, and mortality. American Journal of Public Health 87:1491-1498.
529
17. Rojas, M. 2005. Heterogeneity in the relationship between income and happiness: A conceptual-referent-theory explanation. Journal of Economic Psychology 1-14.
18. E. Diener. Subjective wellbeing. Psychological Bulletin 95 (1984), pp. 542-575. 19. Kashdan, T. B. 2004. The assessment of subjective well-being (issues raised by
the Oxford Happiness Questionnaire). Personality and Individual Differences 36:1225–1232.
20. Leung BW, Moneta GB, McBride-Chang C. 2005. Think positively and feel positively: Optimism and life satisfaction in late life. International Journal of Aging and human development 2005; 61:335-365.
21. Lyubomirsky, S., L. King, and E. Diener. 2005. The benefits of frequent positive affect: Does happiness lead to success? Psychological Bulletin, 6, 803-855.
22. E. Diener. Subjective Well-Being: The Science of Happiness and a Proposal for a National Index. American Psychological Association 55 (2000), pp. 34-43.
23. Diener, Ed, Suh, M., Lucas, E. and Smith, H. 1999. Subjective well-being: Three decades of progress. Psychological Bulletin, 125:276-302.
24. Diener, E., J. Horwitz, and R. A. Emmon. 1985. Happiness of the very wealthy. Social Indicators Research 16:263-274.
25. Sheldon, K., and S. Lyubomirsky. 2006. How to increase and sustain positive emotion: The effects of expressing gratitude and visualizing best possible selves. Journal of Positive Psychology 1:73-82.
26. Abbe, A., C. Tkach, and S. Lyubomirsky. 2003. The art of living by dispositionally happy people. Journal of Happiness Studies 4:385-404.
27. Diener, E., and M.E.P. Seligman. 2002, Very happy people. Psychological Science 13: 81–84.
28. Lyubomirsky S. Why are some people happier than others? The role of cognitiveand motivational process in well-being. American Psychologist 2001; 56:239-249.
530
Figure 1: Public-Private Health Care Utilization in Jamaica (in %), 1996-2002, 2004-2007Source: Taken from Jamaica Survey of Living Conditions, various issues
531
Figure 2: Remittances By Income Quintiles and Jamaica (in Percent): 2001-2007Source: Extracted from the Jamaica Survey of Living Conditions, 2007
532
Table 1: Discharge, Average Length of Stay, Bed Occupancy and Visits to Public Hospital Health Care Facilities, 1996-2004Year Discharge Average Bed Occupancy Visits to Public Facility
Length of Stay Rate1996 145,656 5.7 56.1 546,9331997 153,101 5.8 57.3 598,0041998 158,851 5.5 58.0 634,7921999 163,714 5.1 52.2 6547462000 173,700 4.9 74.9 643,1012001 171,963 6.0 84.6 667,3212002 173,614 6.9 80.2 695,2392003 179,322 6.4 84.5 746,8442004 182,053 6.8 56.0 775,7272005 NI NI NI NI2006 NI NI NI NI2007 NI NI NI NISource: Ministry of Health, Jamaica, Planning and Evaluation Branch, various issuesNI No information available
533
Table 2: Inflation, Public-Private Health Care Service Utilization, Incidence of Poverty, Illness and Prevalence of Population with Health Insurance (in per cent), 1988-2007
Year Inflation Public Private Prevalence Illness Health SeekingMean
Utilization Utilization of poverty Insurance Medical Care Days of Coverage Illness
1988 8.8 NI NI NI NI NI NI NI1989 17.2 42.0 54.0 30.5 16.8 8.2 54.6 11.41990 29.8 39.4 60.6 28.4 18.3 9.0 38.6 10.11991 80.2 35.6 57.7 44.6 13.7 8.6 47.7 10.21992 40.2 28.5 63.4 33.9 10.6 9.0 50.9 10.81993 30.1 30.9 63.8 24.4 12.0 10.1 51.8 10.41994 26.8 28.8 66.7 22.8 12.9 8.8 51.4 10.41995 25.6 27.2 66.4 27.5 9.8 9.7 58.9 10.71996 15.8 31.8 63.6 26.1 10.7 9.8 54.9 10.01997 9.2 32.1 58.8 19.9 9.7 12.6 59.6 9.91998 7.9 37.9 57.3 15.9 8.8 12.1 60.8 11.01999 6.8 37.9 57.1 16.9 10.1 12.1 68.4 11.02000 6.1 40.8 53.6 18.9 14.2 14.0 60.7 9.02001 8.8 38.7 54.8 16.9 13.4 13.9 63.5 10.02002 7.2 57.8 42.7 19.7 12.6 13.5 64.1 10.02003 13.8 NI NI NI NI NI NI NI2004 13.7 46.3 46.4 16.9 11.4 19.2 65.1 10.02005 12.6 NI NI NI NI NI NI NI2006 5.7 41.3 52.8 14.3 12.2 18.4 70.0 9.82007 16.8 40.5 51.9 9.9 15.5 21.2 66.0 9.9Source: Bank of Jamaica, Statistical Digest, Jamaica Survey of Living Conditions, Economic and Social Survey of Jamaica, various issuesNote: Inflation is measured point-to-point at the end of each year (December to December), based on Consumer Price Index (CPI)
NI – No Information Available
534
Table 4Demographic Characteristic of Sampled Population (in N and per cent), N=1,936
N Percent
SexMale 762 39.4Female 1174 60.6
Income Quintile CategorizationTwo Poorest Quintiles 696 36.0Middle Quintile 376 19.4Two Wealthiest Quintiles 864 44.6
Marital StatusMarried 532 35.5Never married 671 44.8Divorced 20 1.3Separated 25 1.7Widowed 250 16.7
Visitors to hospital health care facilitiesPrivate hospital 915 47.3Public hospital 1021 52.7
Private Health Insurance CoverageNo 1086 56.1Yes 850 43.9
Area of residenceRural areas 1289 66.6Other Towns 424 21.9Kingston Metropolitan area 223 11.5
Educational LevelPrimary and below 563 39.4Secondary or post-secondary 813 56.9Tertiary 53 3.7
Age (Mean ± SD) 43.99 ± 27.458Crowding (Mean ± SD) 1.7431 ± 1.26568Negative Affective Psychological condition (Mean ± SD) 4.9182 ± 3.272Positive affective Psychological condition (Mean ± SD) 3.15 ± 2.436
535
Table 5Public Hospital Health Care Facility Utilization by Area of Residence (in percentage), N=1,936
Hospital Utilization
Area of Residence
TotalRural Areas Other Towns KMA
Private
46.9 48.6 47.1 47.3
Public 53.1 51.4 52.9 52.7
Total 1289 424 223 1936
χ 2(2) =0.385, ρ-value=0.825 > 0.05
536
Table 6Public Hospital Health Care Facility Utilization By Per Capita Population Income Quintile (in per cent), N=1,936
Hospital Utilization
Per Capita Population Quintile
Poorest 2.00 3.00 4.00 Wealthiest Total
Private
26.2 41.6 41.2 51.7 68.8 47.3
Public
73.8 58.4 58.8 48.3 31.3 52.7
Total 340 356 376 416 448 1936
χ 2(4) =157.024, ρ-value <0.001
537
Table 7.1Descriptive Statistics of Negative Affective Psychological Conditions and Per capita Income Quintile
Income Quintile N Mean
Std. Deviatio
nStd.
Error
95% Confidence Interval Lower Bound Upper Bound
1.00=Poorest 338 5.7840 2.89747 .15760 5.4740 6.09402.00 355 5.6507 3.17061 .16828 5.3198 5.98173.00 375 5.1627 3.28954 .16987 4.8286 5.49674.00 415 4.6940 3.07402 .15090 4.3974 4.99065.00=Wealthiest 448 3.6875 3.39306 .16031 3.3725 4.0025Total 1931 4.9182 3.27172 .07445 4.7722 5.0642
F statistic [4, 1926] =28.793, ρ-value< 0.001
Table 7.2: Multiple Comparison of Negative Affective Psychological Condition by Per Capita Income Quintile(Tukey HSD)
(I) Per Capita Population Quintile
(J) Per Capita Population Quintile
Mean Difference (I-J) Std. Error Sig. 95% Confidence Interval
Lower BoundUpper Bound
Lower Bound Upper Bound Lower Bound
1.00=Poorest 2.00 .13332 .24177 .982 -.5268 .7934 3.00 .62136 .23861 .070 -.0301 1.2728 4.00 1.09005(*) .23309 .000 .4536 1.7265 5.00 2.09652(*) .22921 .000 1.4707 2.7223
2.00 1.00 -.13332 .24177 .982 -.7934 .5268 3.00 .48804 .23558 .233 -.1552 1.1313 4.00 .95673(*) .23000 .000 .3288 1.5847 5.00 1.96320(*) .22606 .000 1.3460 2.5804
3.00 1.00 -.62136 .23861 .070 -1.2728 .0301 2.00 -.48804 .23558 .233 -1.1313 .1552 4.00 .46869 .22667 .235 -.1502 1.0876 5.00 1.47517(*) .22267 .000 .8672 2.0831
4.00 1.00 -1.09005(*) .23309 .000 -1.7265 -.4536 2.00 -.95673(*) .23000 .000 -1.5847 -.3288 3.00 -.46869 .22667 .235 -1.0876 .1502 5.00 1.00648(*) .21675 .000 .4147 1.5983
5.00=Wealthiest 1.00 -2.09652(*) .22921 .000 -2.7223 -1.4707 2.00 -1.96320(*) .22606 .000 -2.5804 -1.3460 3.00 -1.47517(*) .22267 .000 -2.0831 -.8672 4.00 -1.00648(*) .21675 .000 -1.5983 -.4147
The mean difference is significant at the .05 level.
538
Table 8.1: Descriptive Statistics of Total Positive Affective Psychological Conditions and Per Capita Income Quintile
Per Capita Income Quintile
N MeanStd.
Deviation Std. Error95% Confidence Interval
Lower Bound
Upper Bound
1.00=Poorest 243 2.4156 2.66056 .17068 2.0794 2.75182.00 273 2.8059 2.50786 .15178 2.5070 3.10473.00 278 3.2230 2.29752 .13780 2.9518 3.49434.00 313 3.2843 2.39504 .13538 3.0180 3.55075.00=Wealthiest 386 3.6943 2.21795 .11289 3.4723 3.9163Total 1493 3.1500 2.43610 .06305 3.0264 3.2737
F statistic [4, 1492] =12.366, ρ-value< 0.001
Table 8.2: Multiple Comparisons of Positive Affective Conditions by Per Capita Income QuintileTukey HSD
(I) Per Capita Population Quintile
(J) Per Capita Population Quintile
Mean Difference (I-
J) Std. Error Sig. 95% Confidence Interval
Lower BoundUpper Bound
Lower Bound Upper Bound Lower Bound
1.00=Poorest 2.00 -.39022 .21165 .349 -.9683 .1878 3.00 -.80738(*) .21075 .001 -1.3830 -.2318 4.00 -.86871(*) .20518 .000 -1.4291 -.3083 5.00 -1.27866(*) .19652 .000 -1.8154 -.7419
2.00 1.00 .39022 .21165 .349 -.1878 .9683 3.00 -.41716 .20448 .247 -.9756 .1413 4.00 -.47848 .19873 .114 -1.0213 .0643 5.00 -.88844(*) .18978 .000 -1.4067 -.3701
3.00 1.00 .80738(*) .21075 .001 .2318 1.3830 2.00 .41716 .20448 .247 -.1413 .9756 4.00 -.06132 .19778 .998 -.6015 .4788 5.00 -.47128 .18878 .092 -.9868 .0443
4.00 1.00 .86871(*) .20518 .000 .3083 1.4291 2.00 .47848 .19873 .114 -.0643 1.0213 3.00 .06132 .19778 .998 -.4788 .6015 5.00 -.40996 .18254 .164 -.9085 .0886
5.00=Wealthiest 1.00 1.27866(*) .19652 .000 .7419 1.8154 2.00 .88844(*) .18978 .000 .3701 1.4067 3.00 .47128 .18878 .092 -.0443 .9868 4.00 .40996 .18254 .164 -.0886 .9085
The mean difference is significant at the .05 level.
539
Table 10: Logistic Regression: Predictors of Public Hospital Health Care facility utilization in Jamaica, N=1,049
Explanatory variables
β coefficient
Std. Error
WaldStatistic ρ-value
OR95.0% C.I.
Lower Upper
Retirement Income -.613 .397 2.376 .123 .542 .249 1.181 Household Head -.367 .728 .255 .614 .693 .166 2.886 Cost Health Care .000 .000 13.959 .000 1.000 1.000 1.000 Health Insurance -2.007 .212 89.352 .000 .134 .089 .204 Other Towns .183 .196 .875 .350 1.201 .818 1.765 KMA .033 .357 .008 .927 1.033 .514 2.079 Social supp .555 .151 13.419 .000 1.741 1.294 2.343 Crowding .119 .109 1.194 .275 1.126 .910 1.394 Crime Index .021 .013 2.672 .102 1.021 .996 1.048 Landownership -.226 .173 1.699 .192 .798 .568 1.120 Environment -.283 .208 1.855 .173 .754 .502 1.132 Gender .010 .167 .004 .951 1.010 .728 1.402 Negative Affective .070 .026 7.084 .008 1.072 1.019 1.129 Positive Affective -.071 .033 4.738 .029 .931 .874 .993 Number of males in house .083 .089 .869 .351 1.086 .913 1.293 Number of females in
house.128 .095 1.834 .176 1.137 .944 1.369
Number of children in house
.011 .078 .020 .889 1.011 .868 1.178
Assets owned -.043 .035 1.504 .220 .958 .894 1.026 Age -.004 .004 .728 .393 .996 .988 1.005 Total Expenditure .000 .000 4.458 .035 1.000 1.000 1.000 Health Seeking Behaviour -.706 .083 72.077 .000 .494 .419 .581 Constant 3.654 .896 16.640 .000 38.616
Model Chi-square (df=21) = 326.58, p-value < 0.001-2Log likelihood = 1130.37Nagelkerke R-square=0.356Overall correct classification = 73.0% (767)Correct classification of cases of public utilization =74.3% (N=393)Correct classification of cases of not public utilization (private) = 71.6% (N=374)
540
Table 3Hospital Health Care Facility Utilization (Using Jamaica Survey of Living Conditions Data) By Income Quintile (in per cent), 1991-
1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2004 2006 2007
Public Quintile1=Poorest 57.8 48.8 57.5 54.1 49.4 54.8 44.5 59.1 61.0 55.7 67.6 73.4 70.9 71.0 75.02 43.3 41.8 36.9 34.9 25.3 42.7 39.9 49.0 46.3 44.3 53.5 57.5 53.6 51.1 66.53 29.0 28.8 29.3 17.0 22.7 32.8 37.3 40.7 37.5 41.3 32.1 58.6 57.3 50.6 22.14 35.8 27.1 20.6 25.6 21.7 29.5 26.3 35.1 37.7 44.6 35.3 46.5 36.7 27.5 27.05=Wealthiest 20.6 12.3 16.5 15.7 16.8 11.9 12.4 17.2 15.4 12.8 24.4 30.9 27.6 21.7 21.4
PrivateQuintile1=Poorest 34.4 46.3 32.3 41.2 47.1 40.4 49.1 35.5 34.7 38.7 29.3 22.8 26.8 24.3 22.02 52.9 48.4 58.7 57.0 66.3 54.1 51.1 45.0 50.3 53.8 38.7 37.5 35.7 42.3 33.33 64.5 65.9 62.2 77.0 69.7 62.5 51.8 56.6 59.8 48.8 62.9 37.4 35.7 42.9 64.24 53.1 65.4 74.2 72.2 68.0 63.8 62.5 58.3 57.1 48.8 59.1 46.3 55.6 65.4 69.65=Wealthiest 73.8 78.1 82.5 81.5 80.0 84.6 80.0 78.4 75.4 78.4 66.5 52.5 65.1 73.9 78.6Source: Jamaica Survey of Living Conditions, various issues (a joint publication of the Planning Institute of Jamaica and the Statistical Institute of Jamaica)
541
APPENDIX XXVI – Sampled Research Paper III
Is there a Shift in Voting Behaviour Taking Place In Jamaica?
Paul A. Bourne
542
Abstract
Objective: One of the pillows upon which ‘good’ democracy is built is one’s right to change governments through the autonomous process of voting. Voting behaviour of Jamaicans dates back to 1944. After 1944 to 1971, voting behaviour was analyzed by way of the electoral data. Stone (1992; 1989; 1981; 1978a, 1978b; 1974), on the other hand, has shown that opinion survey can be effectively used to predict an election by way of knowing the profile of the electorates. Since Stone’s (1993) study no one has sought to update and evaluate the voting behaviour of Jamaicans. Plethora of literature exists in the past on voting behaviour using the electoral system and survey opinion polling; but with the PNP being in power for more than the two terms that we have come accustomed, is there a shift taking place in voter preference, or is democracy under siege? This paper seeks to update the knowledge reservoir on contemporary Jamaican voters, 2007Method: This study utilizes data taken from two surveys that were administered by the Centre of Leadership and Governance (CLG), University of the West Indies, Mona-Jamaica, in July to August 2006 and May 2007. For each survey, the sample was selected using a multistage sampling approach of the fourteen parishes of Jamaica. Each parish was called a cluster, and each cluster was further classified into urban and rural zones, male and female, and social class. The final sample was then randomly selected from the clusters. The first survey saw a sample of 1,338 respondents, with an average age of 34 years and 11 months ± 13 yrs and 7 months. On the second survey, 1,438 respondents aged 18 years and older were interviewed, with a sampling error of approximately ± 3%, at the 95% confidence level (i.e. CI). The results that are presented here are based solely on Jamaicans’ opinion of their political orientation. Descriptive statistics will be used to analyze the data.Findings: The current survey (May 2007) indicates that PNP still retains a 3 percent lead (36.2% PNP to 33.2% JLP) among eligible voters. However, a substantial narrowing has occurred since August 2006, when the comparable figures were 53% PNP and 23.1% JLP. This represents a 10% net increase for JLP, and a 17% decrease for PNP. Approximately 67% of the respondents to the May 2007 survey perceived themselves to be in the “working class” (i.e. the lower class), 27% in the “middle class”, 4% within the “upper-middle” class, and 2% “upper class”. Although the survey shows PNP with a slight advantage in the vote across all of the social classes, that advantage tends to be weakest and most vulnerable among the lower class (36.7% PNP, 34.7% JLP), who make up approximately two-thirds of voting age adults. The PNP’s advantage is somewhat stronger among middle class voters (35.6% PNP, 31.2% JLP), and is strongest among the ‘upper-middle’ and ‘upper’ class voters (44.3% PNP, 31.1% JLP). Furthermore, from the May 2007 survey, 41% of the males identified with PNP and 42% with JLP, whereas for females 42% identified with PNP and only about 35% with JLP--a substantial gender difference in party preference. Women also are less satisfied with the two-party system generally, with 22% opting for “something else”, as compared with 17% among males. The May survey also indicates about a 3 percent difference in anticipated voting patterns. Of those who indicated a choice of either PNP or JLP in the coming election, the males were about evenly split at 50.6% JLP / 49.4% PNP. However, among women, 53.5%
543
said they would vote for PNP and 46.5% for JLP -- a 7-point difference. Women also appear to be less satisfied with the performance of their existing MPs. When asked ‘How satisfied are you that the MP from this constituency listens to the problems of the people?’, 12% of the May 2007 sample said they were ‘satisfied’, 54% said ‘sometimes’ and 35% indicated ‘dissatisfied’. Of those who reported being ‘satisfied’, 51.0% were males and 49.0% were females. However of the ‘dissatisfied’, 46% were males with 54% being females. In terms of how they intend to vote in the coming election, among ‘youth’ 30.8% say they will vote for PNP, 26% for JLP, and 34.7% say they will not be voting. The figures are much closer for middle-aged adults, with 38.7% saying they will vote for PNP and 36.3% for JLP. Among the elderly, there is a ten-point spread, with 48% for PNP and 38% for JLP. Levels of non-voting are highest among youth, with 34.7% saying they “will not vote”, compared to 19.8% among middle-aged adults, and 10% among the elderly. Conclusion: Voting behaviour is not, and while people who are ‘undying’ supporters for a party may continue to voting one way (or decides not to vote); the vast majority of the voting populace are more sympathizers as against being fanatics. With this said, voting behaviour is never stationary but it is fluid as water and dynamic as the social actions of man. Generally, people vote base on (i) charismatic leadership; (ii) socialization - earlier traditions; (iii) perception of direct benefits (or disbenefits); (iv) associates and class affiliation; (v) gender differences, and that there is a shift-taking place in Jamaican landscape. Increasingly more Jamaicans are becoming meticulous and are moving away from the stereotypical uncritical and less responsive to chicanery. Education through the formal institutions and media are playing a pivotal function in fostering a critical mind in the public.
544
Introduction
Since its transition from the colonial system to independent self-government,
Jamaica is one of the few countries in the global South that has entertained a competitive
party system (Stone 1978). There had been a regular transference of power between the
two dominant political parties, the Peoples National Party (PNP) and the Jamaica Labour
Party (JLP). But with the PNP having been in power since 1989, Jamaica may be seeing a
shift in voter preference, or a larger transition in their democratic process. Stone’s (1993)
study was the last study which sought to incorporate the Caribbean into the extant
literature on democratic theory by analyzing the voting behaviour of Jamaicans. In the
subsequent elections under universal suffrage (1944 to 1971), voting behaviour was
analyzed by way of the electoral data. Stone (1992; 1989; 1981; 1978a, 1978b; 1974)
demonstrated that opinion survey can be effectively used to predict an election by way of
knowing the profile of the electorates. Dearth of literature exists in the past on voting
behavior in Jamaica using the electoral system and survey opinion polling; since Stone’s
(1993) study no one has sought to update and evaluate the voting behaviour of Jamaicans.
Using data taken from two surveys that were administered by the Centre of Leadership
and Governance (CLG)50, University of West Indies, Mona-Jamaica, this paper seeks to
update the knowledge reservoir on Jamaican voters in 2007, pending a very critical
upcoming election period.
Until the late 1980s, no political party has had more than two terms in office in
Jamaica (Stone 1978b). There had been a regular transference of power between the two
dominant political parties: the ‘left’ oriented Peoples National Party (PNP) and the
50 The Centre for Leadership and Governance was launched in November 2006 within the Department of Government, UWI, Mona-Jamaica, to develop governance structure, encourage student participation, and provide policy based research activities for parliamentarians.
545
capitalist oriented Jamaica Labour Party (JLP).51 Stone (1978) argued that the continuous
changing of the political directorates was a hallmark of a healthy democratic system. The
victory of the PNP in 1989 changed this cycle; following that victory, the party won four
consecutive general elections, something that has come as a surprise to many political
pundits. This change signals a paradigm shift from what constitutes a “healthy”
democracy. The Peoples National Party (PNP) has accomplished an unprecedented feat,
having been in power for the past 15 years; therefore, an analysis of voting behaviour is
needed in order to understand what has changed this two party competitiveness that once
existed in Jamaica. But to what extent can we assess people’s support of democratic
freedom from their voting behaviours? If a people continue to democratically elect the
same party, it could be construed as a change occurring within the political culture. 52
One of the particular features of Jamaican political culture is the class affiliations
of the two dominant parties. It can been argued that the “lower” and “middle” classes of
Jamaican are predominantly oriented towards the PNP while Jamaica’s “upper” class is
generally affiliated with the JLP. Each of the main political parties in Jamaica, the JLP or
the PNP, will amass support from various social classes because of programmes that they
employ. For example, when the Michael Manley administration (PNP) took the decision
to introduce free education in the 1970s, maternal leave for pregnant women, “crash
programme work” for the working class, this resonated with the working and middle
51 Despite the fact that the political affectation of the PNP has changed since its original installation, the party is still associated with social democratic principles.52 Space does not allow for a thorough examination of Jamaica’s political culture, nor is such an examination the thrust of this paper, but it is important to offer some thoughts on political socialization as it relates to this study. It has been argued that the political culture of a society is tied to its socialization, which is a consensus of beliefs, customs, preconception and a certain orientation among its members (see Powell, Bourne and Waller 2007). In this paper, political socialization will refer to the process by which Jamaican’s develop their partisan attitudes and affiliations. It would be dangerous to assert that the socialization process, the process by which people form their beliefs and customs, is owed entirely to the family unit. Recognizing the role that the family plays in locating people within larger structures like class, it is the contention of this paper that education too plays a pivotal role in political socialization.
546
classes in Jamaica. The JLP through Sir Alexander Bustamante has equally contributed
to the perspective of the particular classes. When Bustamante took the position to die
rather than leaving the sugar workers, it resonated with the working class of the day, and
could justify his victory at the poll following that showing. An important consideration of
this study will be the class composition of the voters surveyed.
This study borrows from Stone’s (1978) previous usage of opinion polling to
determine voting behaviour. What was unique about Stone’s work is that he was aware of
the limitations of empiricism, and therefore sought to explain the “swings” in electoral
outcomes via a political economy framework (Edie 1997). The likelihood of a Jamaica
Labour Party (JLP) win or the continuance of current PNP administration, which in and
of itself would be furthering a neoteric history of voting behaviour in this country,
requires careful analysis beyond aggregate numbers. Indeed, the association between
factors such as gender, and age, and their impact on voting behaviour and voter
numeration will be important considerations in this paper as well. Therefore, one of the
objectives of this study is to examine the differences in voting behaviour by gender. A
second objective is to evaluate whether there are differences in support for the two main
political parties across age groups and social classes.
One of the challenges of such a study is the static use of self-reported data as a
yardstick to assess future decisions of people. Human behaviour is fluid, and so any
attempt to measure this in the long-term might be futile. Nevertheless, we will attempt
here to unearth some salient characteristics of the Jamaican voters as well as to provide a
more in-depth understanding of a probable outcome of the next general elections. While
this study is not concerned with furthering the epistemological framework that Stone
547
relied on, we recognize that the survey research technique could offer tremendous
insights on Jamaica’s voting behaviour in the forthcoming elections. This study should
offer some grounds on which to compare and contrast the voting behavioural patterns of
Jamaicans currently and perhaps in the future, and to understand those factors that are
likely to influence non-voters.
Originally, political economists used electoral data to provide rich information on
aggregate voting patterns by regions (Stone 1978; Lipset and Rokkan 1967). The study of
voting behaviour emerged out of the electoral data, but this only offer scholars and non-
academics alike an aggregate perspective on the actual voting patterns by geographic
space (Stone1974; 1978b). A comparison between electoral statistics and sample survey
method, is that the former is not able to probe the meaning systems of people, their
attitudes, perceptions, moods, expectations, political behaviour that justify their actions
(or inactions). On the side of the delimitation of electoral statistics, it is primarily past
events with subdivision concerning socio-demographic and psychological conditions of
people. Therefore, this approach whilst offering invaluable information on the
ideographic, cross-national and comparative patterns of voting, and equally providing a
contextual background on the political milieu from which the voters are drawn is limited
in scope. As voters are not only influenced by those conditions, but also impacted upon
by socio-psychological and economic conditions (Stone 1974), the need was there for a
method that would capture those tenets, which is the ‘political sociology of voting’.
It follows then that when Professor Carl Stone introduced sample survey method
in the political landscape to probe people’s voting behaviour it was a first for the nation
(Stone 1973, 1974, 1978b). The sample survey method allows for a more detailed
548
analysis of voting behaviour, by way of those demographic, socio-economic and political
factors that influence the choices of voters. The sample survey method allows for the use
of the social structure model in seeking to investigate voting behaviour. Among the
advantages of the use of the survey method is its ability to predict behaviour, provide
association (or the lack thereof), it is high in ability to generalize, can be used for
national, regional and international comparison among other nations. With this approach,
Stone was able to consecutively predict all the winners for the general elections between
1970 and 1994. The social structure model places emphasis on social conditions such as
social class as predictors of voting behaviours. In this paper, the author will only address
age, gender and class as predictors of voting behaviour, because the survey with which
this analysis will be made possible can only accommodate those social factors.
Method
This survey was administered by the Centre of Leadership and Governance
(CLG), University of the West Indies, Mona, Kingston, in May 2007. The sample was
randomly selected from the fourteen parishes of Jamaica, using the descriptive research
design. The sample frame is representative of the population based on gender and
ethnicity. A total of 1,438 respondents aged 18 years and older were interviewed for this
study, with a sampling error of approximately ± 3%, at the 95% confidence level (i.e. CI).
The results that are presented here are based solely on Jamaicans’ opinion of their
political orientation. Descriptive statistics were used to analyze the data.
For each survey, the sample was selected using a multistage sampling approach of
the fourteen parishes of Jamaica. Each parish was called a cluster, and each cluster was
further divided into urban and rural zones, male and female, and upper, middle and lower
549
social classes. The final sample was then randomly selected from the clusters. The first
survey saw a sample of 1,338 respondents, with an average age of 34 years and 11
months ± 13 yrs and 7 months. On the second survey, 1,438 respondents aged 18 years
and older were interviewed, with a sampling error of approximately ± 3%, at the 95%
confidence level. The results presented here are based solely on Jamaicans’ opinion of
their political orientation.
Operational Definitions
It is necessary here to provide some clarity on the terms that are being used in this
study. We are attempting to make some predictions on voting behaviour, which is the
level of voters’ participation in a democratic society. In other words, voting behavior here
refers to “which party you intend to either vote for or have voted for,” and the frequency
of support or lack of. Survey participants were asked if they were (a) definitely voting
for the PNP, (b) definitely voting for the JLP, (c) probably voting for the JLP, or (d)
probably voting for the PNP. Voter enumeration is another important term that we are
dealing with in this study. Enumeration here is defined as the self-report of people who
indicated that they are registered to vote in an election. In the survey it was denoted as a
binary value (0=No, 1=Yes).
This paper also attempts to look at Jamaica’s political culture in terms of social
constructions, such as gender, and social class. We recognize gender as a social construct
and set of learned characteristics that identify the socio-cultural prescribed roles that men
and women are expected to play. In the survey it is also represented as a binary value
(0=female, 1=male). Social class here is defined subjectively. Respondents were asked to
550
indicate using their self-assessment as to which social class they consider themselves to
be in (1) working class, (2) middle class, (3) upper-middle class or (4) upper class.
Educational level is an integral part of defining social class, even subjectively. By
educational level we are referring to the total number of years of schooling, (including
apprenticeship and/or the completion of particular typology of school) that an individual
completes within the formal educational system (1=primary and/or preparatory and
below; 1=secondary or high; 3= vocational; 4=undergraduate and graduate education, and
5=post-university qualification).
Lastly, age is defined as the length of time that one has existed; a time in life that
is based on the number of years lived; duration of life. Age is represented as a non-binary
measure (1=young, 1=middle age- 26 to 59 years and 3=elderly). The United Nations has
defined the aged as people of 60 years and older (WHO 2007). Oftentimes, ageing (i.e.
the elderly) means the period in which an individual stops working or he/she begins to
receive payment from the state. Many countries are, however, using 60 years and over as
the definition of the elderly including Professor Eldemire (1995) but for this paper, we
will use the chronological age of 60 years and beyond.
Results
Sociodemographic factors
Some background information on May 2007 survey is helpful here. According to
the Statistical Institute of Jamaica (2001) 91.61% of Jamaica is African (Black), while
0.89% are East Indian, and those of Chinese, and European descent comprise 0.20% and
0.18% of Jamaica’s population respectively. (6.21% of Jamaicans were classified as
551
“other.”) Some 81.3% (n=1168) of the sampled respondents considered themselves to be
Africans (or Blacks), 3.8% (n=54) Indians, 0.5% (n=Asians – Chinese), 0.5% (n=7)
Syrians (or Lebanese), 0.2% (n=3) Europeans (or Caucasians or Britain or French), 0.1%
(n=1) North American Caucasians and 13.2% (n=190) reported mixed.
Approximately 33% (n=468) of the respondents were youth, 62.3% (n=891) were
middle age and 5.0% were elderly. Some 28.7% (202) of the males are youth, 65.9%
(n=463) are middle age while 5.4% (n=38) are 60 years and older. Concerning the
female population, 36.6% (n=266) are youth, 58.9% (n=428) are middle age and 4.5%
(n=33) are senior citizens. 74.4% (n=1009) of those who supplied data on their ages
indicated that the current government favours the rich more than the poor. Of those who
reported that the government is fostering the interest of the rich, 33.3% (n=336) were
youth, 62.3% (n=629) were middle age and 4.4% (n=44) were elderly. Disaggregating
the data reveal that 50.4% (n=506) of those who indicated that the current policies favour
the affluent are males compared to 49.6% (n=498) of the females. Most (58.8%, n=293)
of the female respondents who reported that that the present policies of the government
favour the rich are middle age, with 37.6% (n=187) who are youth compared to 3.6%
(n=18) who are elderly. More middle- aged men (65.8%, n=333) than middle- aged
women (58.8%, n=293) believe that the current administration’s policies favour the rich.
A major difference between the genders and age cohort was found as substantially more
youth females (37.6%, n=187) than youth males perceived that government’s policies are
anti-poor.
Voting Patterns
552
Several important shifts can be seen to have taken place in voter attitudes over the
past ten months, if one compares the August 2006 and the May 2007 CLG survey results.
When asked who they would “vote for in the next general elections”, the current (May
2007) survey indicates that PNP still retains a 3 percent lead (36.2% PNP to 33.2% JLP)
among eligible voters. However, a substantial narrowing has occurred since August 2006,
when the comparable figures were 53% PNP and 23.1% JLP; this represents a 10% net
increase for JLP, and a 17% decrease for PNP. There has also been a shift in ‘overall
party support’ during that same period. Again, PNP remains slightly ahead, but has lost
ground in the intervening months. When asked what party they “always vote for” or
“usually vote for”, 43% of the respondents to the May 2007 survey say they “usually” or
“always” vote for PNP, whereas 36.3% “usually” or “always” vote for JLP. As of the
August 2006 survey, the comparable figures were 57.2% PNP supporters and 25.2% JLP
supporters -- an 11% increase for JLP and 14% drop for PNP over a ten-month period
(see for example, Bourne 2007).
A shift in terms of political orientation seems to be taking place as 5.3% of
‘Definite’ supporters of the PNP reported that they would definitely be voting for the JLP
compared to 4.7% of the ‘Definite’ JLP who indicated that they would definitely be
marking an X for the PNP. Further, 1.5% of ‘Definite’ PNP indicated a possibility of
voting for the JLP compared to 2.8% of ‘die-hearted’ JLP supporters who mentioned that
they probably might be marking that ‘X’ for the PNP. Furthermore, 3.4% of those who
have a political leniency toward the JLP reported that they will definitely be voting for
the PNP with 4.3% mentioned ‘probably’. However, among those with the PNP
orientation, 18.9% of those who voted PNP in the last general elections reported that they
553
will be voting for the JLP, with another 16.5% who said that they might be marking that
X for the JLP.
Those whose political culture is not party based, but whose perspective is shaped
possibly on issues, 21.3% indicated that they might vote for the PNP compared to 15.7%
for the JLP. Of this same group of voters, 25% reported a definitely preference for the
PNP with the JLP receiving the same percentage. The dissatisfaction with the political
system is higher for those with a PNP orientation as against with a JLP belief: 9% of
‘Definite’ PNP voters reported that they will not be vote in the upcoming elections
compared to 5.7% for JLP. Political culture is not static and so, of those who expressed a
leniency toward a party, the dissatisfaction is higher, again, for the PNP as 15% reported
that they will definitely not be voting in the upcoming general elections compared to 10%
for the JLP.
The study found a positive statistical relationship between future voting behaviour
of those who are enumerated and past voting behaviour. The findings reveal that 75.5%
of those who are ‘sympathizers’ of the JLP support will retain this position in the
upcoming elections compared to 68.2% for the PNP. Continuing, of ‘Definite’ voters,
11.3% of the JLP supporters reported that they ‘probably’ will vote for their party
compared to 15.9% of the PNP supporters.
Social Class
There appear to be important ‘class-related’ differences in Jamaicans’ election
preferences, yet they are paradoxical -- tending to have different effects depending on
whether one is looking at voting, party, or candidate preferences. Approximately 67% of
554
the respondents to the May 2007 survey perceived themselves to be in the “working
class” (i.e. the lower class), 27% in the “middle class”, 4% within the “upper-middle”
class, and 2% “upper class.” Although the survey shows PNP with a slight advantage in
the vote across all of the social classes, that advantage tends to be weakest and most
vulnerable among the lower class (36.7% PNP, 34.7% JLP), who make up approximately
two-thirds of voting age adults. The PNP’s advantage is somewhat stronger among
middle class voters (35.6% PNP, 31.2% JLP), and is strongest among the ‘upper-middle’
and ‘upper’ class voters (44.3% PNP, 31.1% JLP). With respect to ‘party identification’
(“which do you consider yourself to be?”), PNP has a slight advantage among the lower
(43.2% PNP, 39.6% JLP) and middle (38.6% PNP, 35.6% JLP) classes. However, in the
“upper-middle and upper class” category, JLP has the edge in party identification. (40.3%
PNP, 43.5% JLP)
Within the lower class, marginally more people believe that Simpson-Miller
(38.6%) “Would do a better job of running the country” compared to Golding (36.2%).
However more people within the middle class reported that Golding (37.4%) would do a
better job of running the country than Simpson-Miller (31.9%). Upper-middle and upper
class respondents, on the other hand, give Mrs. Simpson-Miller the nod over Mr. Golding
(40.3%, 33.8% respectively).
Clearly, there is a class dimension to the voting preferences. Most of the sampled
population had completed secondary school (including traditional and non-traditional
high schools) (31.9%, n=459).53 Approximately 23 % (n=333) of the respondents had at
least an undergraduate level training, with 13.4% being current students. Only 4.7% of
the sampled population (n=1,438) had mostly primary or preparatory level education.
53 This includes traditional and non-traditional high schools.
555
Political Socialization
Have you ever stopped to think about WHY you have the political beliefs and values you do? Where did they come from? Are they simply your own ideas or have others influenced you in your thinking? Political scientists call the process by which individuals acquire their political beliefs and attitudes "political socialization." What people think and how they come to think it is of critical importance to the stability and health of popular government. The beliefs and values of the people are the basis for a society's political culture and that culture defines the parameters of political life and governmental action (Mott, 2006).
Unlike other species whose behaviour is instinctively driven, human beings rely
on social experiences to learn the nuances of their culture in order to survive (Macionis
and Plummer, 1998). “Social experience is also the foundation of personality, a person’s
fairly consistent patterns of thinking, feeling and acting” (Macionis and Plummer, 1998),
which is explained by Mott that political socialization helps to explain one’s attitude to
people, institution and governance. In cases where there is non-existence of social
experiences, as the case of a few individuals, personality does not emerge at all (Macionis
and Plummer, 1998). An example here is the wolf boy (Baron, Bryne and Branscombe
2006). They noted that a boy who was raised by wolves, when he was brought from that
situation into the space of human existence in which he was required to wear clothing and
other social events died in less than two years from frustration. This happening goes to
show the degree to which individuals are ‘culturalized’ by society, and that what makes
us humans is simply not mere physical existence but the consent of society of that which
is accepted as the definition of humans.
556
Macionis and Plummer argued that Charles Darwin supports the view that human
nature leads us to create and learn cultural traits. “The family is the most important
agent of socialization because it represents the centre of children’s lives” (Macionis and
Plummer, 1998). Charles A Beard (in Tomlinson, 1964) believed that mothers should be
appropriately called “constant, carriers of common culture”; this emphasizes the very
principal tunnel to which mother guide their young, and they are equally conduits of the
transfer of values, norms, ideology and perspective on the world for their children.
Infants are almost totally dependent on others (family) for their survivability, and this
explain the pivotal role of parents and-or other family member. The socialization process
begins with the family, and more so those individuals to which the child will rely for
survival. This happening emphasizes the how the child is fashioned into a human, and
not merely because of birth. The child learns to speak, the language, actions, mode of
communication, value system, norms and the meaning of things through adoption,
repetition, and observation of the social actions of people within the environment. The
process of becoming a human is simply only performed by the family but other socio-
political agents.
Our political upbringing is simply political socialization (Munroe, 2002).
Munroe suggests that the ways and means through which our views about politics and our
values in relation to politics are formed is part of our political socialization. Munroe
states that, “It is also our upbringing that made us believe that politics is corrupt, dirty
and prone to violence.” The astute professor of governance, Trevor Munroe, shows that,
there are ranges of channels through which our political personalities are formed and
these are known as primary and secondary agents of political socialization. This is in
557
keeping with other scholars that argue that socialization albeit political or otherwise
shapes the belief system, the attribute, the customs, the culture and the norms of a group
of people. It is undoubtedly clear from Munroe’s, Macionis and Plummer’s and
Haralambos and Holborn’s positions that, individuals are directly and indirectly
influenced by the family, the school, the church, the mass media, political institutions and
the peer group, as they all share the same focal view on socialization. That is, the
political and sociological scientists have converged on a point of principle, that
socialization albeit it may be political or sociological is one of the same.
The family imparts its political beliefs on the children by way of its biases,
acceptance and approval of a particular political ideology (Munroe, 2002). He believes
that, the indirect approach is one that the attitudes being formed are only indirectly
related to politics, and are not directly political. For example, in the school or workplace
there is some form of authority. The relationship form of authority develops an attitude
to authority. This means that the attitude formed towards authority spills over to
government. Both Political Scientists’ and Sociologists’ propositions of socialization are
similar except that the Political Scientists look at socialization from a political aspect
(political ideology as a result of socialization). Sociologists, on the other hand, examine
the process of socialization and its impact on society, on the individual general, and not
from a micro unit of the political system as that is only an aspect in the ‘culturalization’
process of the individual. Hence, are we proposing that human behaviour and
conceptions are learned?
Formal education that is branch within the socialization units provide the
individual with a particular premise upon which the rationale his/her decisions.
558
Education is no different from the family in the socialization process. It is able to make
available certain set of tools in how events are view; matters are conceptualized and
interpreted along with the reasoned conclusion on matters. The lack of this product
means that the individual must rely on the other agents of socialization such as the
family, the church, the mass media, and political institution for a platform upon which to
interpret the world. Education is associated with social class. This, therefore, means that
particular classes with have more of it (middle-class) than others (working or lower class)
and even the upper class. The irony that holds here is that the upper class has the
resources and wealth and so they are able to purchase the middle class skills to execute
their objectives. Therefore, the issue of political socialization is carried out through
education and social classes.
It follows that amongst the working class, the political preference is one that
favours the PNP (Table 1). In the ‘Definite’ supporters, the PNP has a lead of 2.0% over
the JLP and an even smaller advantage in the probably category (0.8%). In the lower-
middle middle class, the ‘Definite’ supported favour the JLP by 1.4% over the PNP and
the reverse is the case in the probably group (i.e. 2.1%). This means that the PNP has an
advantage of 0.7% in the lower-middle middle class. The JLP’s ‘Definite’ supporters in
the upper middle class are 4.2% more than that of the PNP’s. However, the PNP trails the
JLP in the probably category by 20.8%. In the upper class, the JLP has an advantage
over the PNP in the probably category (i.e. by 7.7%), compared to 69.2% preference of
the PNP in the ‘Definite’ supporters.
559
Table 1 Likely Voter for the 2007 General Elections by Subjective Social Class
560
Subjective Social Class
Working
class Middle class
Upper-
middle class Upper class
Probably PNP
71
12.7%
28
14.7%
3
12.5%
1
7.7%
Definitely PNP
162
28.9%
50
26.2%
5
20.8%
9
69.2%
Probably JLP
67
11.9%
24
12.6%
8
33.3%
2
15.4%
Definitely JLP
151
26.9%
52
27.2%
6
25.0%
0
0.0%
Would not vote
110
19.6%
37
19.4%
2
8.3%
1
7.7%
Total 561 191 24 13
Gender
Stone’s work did not give an accurate depiction of the female participation in
political life either by using representative involvement in positions of authority or by the
use of mass meetings, dialogue and other such events. The number of women who are
actively involvement in the mass meetings, and canvassing outstrip that of the men (see
for example Figueroa 2004). Contrary to Professor Stone’s belief, women are the
mobilizing engines of the political parties, and their male counterparts are face of the
assiduous work that was spent to fashion the event to be seen by the publics. In
Figureroa’s work (2004), he argued that women play a dominant role in political
participation than their male counterparts. Among the findings of Powell, Bourne and
Waller (2007, 79), 13% (n=169) of the sampled population (n=1,338) reported that they
agreed with the statement “Generally speaking, men make better political leaders than
women…” compared to 85% (n=1,142). If Jamaicans believe that men are not
genetically better leaders than women are, this begs the questions ‘What explains the
contemporary situation of one female prime minister in the nation’s annals; and why the
disproportionate gender imbalance in parliament’?
While women play an importance in the political culture of Jamaica, it can be
argued they have opted to give the face of their contributions to the men because of the
patriarchal underpinnings of the society. Many women have been socialized with this
male dominated culture, and have come to operate within its infrastructure. In analyzing
the Electoral Office of Jamaica’s data (EOJ), Figueroa found sex differences in role
participation. From Mark Figueroa’s work (2004), women constitute 80% of indoor
agents, 80% of poll clerks, and the list goes on. He pointed out the following that, “In the
561
grass-root structures of the parties, the women predominate” and that, “Women are the
main ones to attend the local party meetings” but he reiterates the point of male
dominance, when he said that, “Yet the base-level organizations still have a tendency to
elect the disproportionate number of male delegates to higher party bodies” (pgs. 138-
139). Therefore, they frequently assume a role ‘second’ to the male in the political arena,
and system that is generally accepted by the wider society. Vassell 2000 (in Figueroa
2004) demonstrates that men continue to dominate leadership positions in Jamaica, in
particular political management. This ranges from the House of Representative to the
Standing Committees of the two main political parties. To further argue this point,
Figueroa (2004) highlighted that none of Jamaica’s Governor Generals or prime ministers
[at the time of writing the article] were females.
“In the second half of the twentieth century, women have moved into many
spaces previously occupied by men” (Figueroa 2004, 146). Does the changing of the
political guard in the PNP from a man to a woman, denote a shift in gender privilege in
the male dominated socio-political arena within Jamaican society? Figueroa provided
some insight on the never-ending cycle of patriarchal society when he said, “Women
have made progress but the old patterns of gender privileging continue to reproduce
themselves” (2004, p 146). Nevertheless, this is the beginning of a transformation in
culture that will take years of reimaging and reimagining of the people’s present
socialization. Because the incumbent Prime Minister is a woman, some have argued that
‘woman time come’ and that gender differences could be a decisive factor in determining
the outcome of the election. If we are to consider the disparity in voter numeration (Table
2), voter participation on general or local government elections, the number of positions
562
in representational politics, and the plethora of males in political leadership positions, this
will automatically skew an appearance of male dominance in the political arena.54 This is
not necessarily the case, as the female execute many roles in the political process.
In the May 2007 survey, 41% of the males identified with PNP and 42% with
JLP, whereas for females 42% identified with PNP and only about 35% with JLP--a
substantial gender difference in party preference. Women also are less satisfied with the
two-party system generally, with 22% opting for “something else”, as compared with
17% among males.
The May survey also indicates about a 3 percent difference in anticipated voting
patterns. Of those who indicated a choice of either PNP or JLP in the coming election,
the males were about evenly split at 50.6% JLP / 49.4% PNP. However, among women,
53.5% said they would vote for PNP and 46.5% for JLP -- a 7-point difference.
Women also appear to be less satisfied with the performance of their existing
MPs. When asked ‘How satisfied are you that the MP from this constituency listens to
the problems of the people?’, 12% of the May 2007 sample said they were ‘satisfied’,
54% said ‘sometimes’ and 35% indicated ‘dissatisfied’. Of those who reported being
‘satisfied’, 51.0% were males and 49.0% were females. However of the ‘dissatisfied’,
46% were males with 54% being females.
54 When the data was disaggregated by gender, in the probably category, males had a marginal preference (0.4%) for the JLP, and for the females the PNP leads by 1.0%.
563
Table 2: “Likely” Voters for the 2007 General Elections by Gender
0
10
20
30
40
50
60
Probably PNP Probably JLP Definitely PNP Definitely JLP
Male
Female
Total
Does age make a difference?
If we consider Table 3, in regards to ‘Definite’ supporters of the two political
parties, significantly more elderly (16.6%) have indicated a preference for the PNP. The
reason for this probably lies in the fact that the PNP has implemented programs that
significantly reduce health care costs for the elderly. Therefore, campaign issues become
of much more importance to the elderly, who can not always attend political meetings
and the like. The political orientation for the youth was relatively the same in both the
‘Definite’ and the ‘probably’ categorization. In the ‘Definite’ group, the PNP had a 0.9%
lead over the JLP, whereas for the probably grouping, the lead was for the JLP of 1.3%.
This means that the JLP comes out ahead of the PNP in the youth age cohort (by 0.4%).
In the middle age cohort, the PNP has the advantage in both categories. The lead was
0.9% in the ‘Definite’ supporters and 1.7% in the ‘probably’ age cohort. Hence, people’s
choices are dictated to some extent by their ages. With this said, younger voters can be
said to be less interested about social values and are more driven by material resources
564
and personal gratification that politics is of little interest to them except they were
socialized in understand these issues.
With respect to party identification, of the 32% of sampled respondents in the
May 2007 survey who are ‘youth’ (under 25 years), 40.4% of those reported a PNP
orientation, compared to 31.5% who said they leaned toward the JLP. Youth also report
being more disenchanted with the existing two party systems than is the case for their
elders. Some 28% of youth reported that they are ‘something else’ than PNP or JLP,
compared with only 16% who chose this response among the older adults. Among those
who are middle-aged (26-60 years), the difference between those who favour the PNP
and favour the JLP shrinks to only 1% (at 42.2% and 41.4% respectively). The elderly
(over 60), on the other hand, are substantially PNP sympathizers. Approximately 50%
reported a PNP preference compared to 34% for the JLP, which represents a 16%
difference -- a significant preference for the PNP when compared to the other age groups.
In terms of how they intend to vote in the coming election, among ‘youth’ 30.8%
say they will vote for PNP, 26% for JLP, and 34.7% say they will not be voting. The
figures are much closer for middle-aged adults, with 38.7% saying they will vote for PNP
and 36.3% for JLP. Among the elderly, there is a ten-point spread, with 48% for PNP
and 38% for JLP. Levels of nonvoting are highest among youth, with 34.7% saying they
“will not vote”, compared to 19.8% among middle-aged adults, and 10% among the
elderly. These figures are generally in accord with voting studies in many other societies
that have consistently shown that as adults’ age and become more engaged in the social
order; they tend to vote at higher levels.
565
566
Table 3 Likely Voters for the 2007 General Elections by Age Cohort
0
5
10
15
20
25
30
35
40
45
ProbablyPNP
ProbablyJLP
DefinitelyPNP
DefinitelyJLP
Will notvote
Youth
Middle age
Elderly
567
Conclusion
The current survey (May 2007) indicates that Peoples National Party still retains a
small lead among registered voters. More than half of the respondents to the May 2007
survey perceived themselves to be in the “working class” (i.e. the lower class), 27% in
the “middle class”, 4% within the “upper-middle” class, and 2% “upper class”. Although
the survey shows PNP with a slight advantage in the vote across all of the social classes,
that advantage tends to be weakest among the lower class, which makes up
approximately two-thirds of voting age adults. Therefore there remains the question of
what will influence the voting behaviour of this rather substantial voting block. The
PNP’s advantage is somewhat stronger among middle class voters, and is strongest
among the ‘upper-middle’ and ‘upper’ class voters.
We have also evidenced gender dissimilarity in voting behaviour. From the May
2007 survey, 41% of the males identified with PNP and 42% with JLP, whereas for
females 42% identified with PNP and only about 35% with JLP--a substantial gender
difference in party preference. Women also are less satisfied with the two-party system
generally, with 22% opting for “something else”, as compared with 17% among males.
It is significant that levels of non-voting are highest among youth, with 34.7%
saying they “will not vote,” compared to 19.8% among middle-aged adults, and only 10%
among the elderly. Stone (1974) found the highest level of age involvement in the
political process occurred for ages between 30 and 49 years (p.54). This study did not
allow us to assess the age cohort in which there is the highest level of involvement in the
political process in present day Jamaica. It is the contention of this paper that this age
cohort holds an important position in determining the outcome of the upcoming election
568
because of the potential for voter enumeration, and therefore the opportunity to exercise
political will in favour of either dominant political party. One area that this study did not
allow us to delve into is the issue of why people are not voting if they are registered to do
so. Further research in this area may allow us to explore other influences concerning
voting behaviour that may be more external than political socialization.
As the populace leader may not be the next prime minister, it appears that the
winner of the election will be dependent on a few conditions. First, will the alleged
uncommitted (or undecided) voters, decide to vote? Secondly, which political leader will
be able to mobilize voters to execute their democratic rights will make the difference?
How will the gender distribution of the votes turn out? Will the Most honourable Mrs.
Portia “Sister P’s” Simpson-Miller gender giver her the advantage or will the opposing
leaders take the advantage because of their actions or lack thereof? Lastly, how will
marginal seat behaviour be on the day in question?
Voting behaviour is not only about political preference, and while people who are
‘undying’ supporters for a party may continue to voting one way (or decides not to vote);
the vast majority of the voting populace are more sympathizers as against being fanatics.
With this said, voting behaviour is never stationary but fluid and dynamic. It is influenced
by a number of social factors. Generally, people vote base on their appreciation of
charismatic leadership, political socialization, their perception of direct benefits,
associates and class affiliation, and gender differences. Increasingly more Jamaicans are
becoming meticulous and are moving away from the stereotypical uncritical and less
responsive to chicanery.
569
570
References
Baron, R. A., Byrne, D., & Branscombe, N. R. (2006). Social Psychology (11th ed.). Boston, MA: Pearson/Allyn and Bacon.
Bourne, Paul A. 2007. Any Shift in Voting behaviour? Evidence from the 2006, 2007 CLG Survey. In Focus. Sunday Gleaner, July 8, 2007. Kingston, Jamaica: Jamaica Gleaner
Branton, Regina P. (2004). Voting in initiative elections: Does the context of racial and ethnic diversity matter? State Politics and Policy Quarterly. 4(3): 294-317.
British Broadcasting Corporation (BBC). (2007). Voting behaviour. http://www.bbc.co.uk/scotland/education/bitesize/higher/modern/uk_gov_politics/elect_vote2_rev.shtml (accessed June 12 2007).
Chevannes, Barry. (2001). Learning to be a man: Culture, socialization and gender identity in five Caribbean communities. Kingston, Jamaica: Univer. of the West Indies Press.
Chressanthis, George A., Kathie S. Gilbert, and Paul W. Grimes. (1991). Ideology, constituent interests, and senatorial voting: The case of abortion. Social Science Quarterly. 72(3), 588-600.
Edie, Carlene J. 1997. Retrospective in commemoration of Carl Stone: Jamaican pioneer of political culture. Journal of Interamerican Studies and World Affairs.
Figueroa, Mark. (2004). Old (Female) glass ceiling and new (male) looking glasses. Challenging gender privileging in the Caribbean. In Gender in the 21 st century: Caribbean Perspective, visions and possibilities, edited Barbara Bailey and Elsa Leo-Rhynie. Kingston, Jamaica: Ian Randle.
Griffin, Leonna D. (2004). US foreign aid and its effects on UN General Assembly voting on important votes. Louisiana State University and Agriculture and Mechnical College, dissertation (thesis), M.A.
Haralambus, M & Holborn, M (2002), Sociology: Themes and Perspective. London: University Tutorial Press
Historylearningsite.ca.uk. (2007). Voting behaviour in America. http://www.historylearningsite.co.uk/voting_behaviour_in_america.htm (accessed 12 June 2007).
Jamieson, A., Shin, H. B, and Day, J. (2002). Voting and Registration in the Election of November 2000: Population Characteristics. U. S. Census Bureau Report.
571
Johnson, Martin, Robert M. Stein, and Robert Winkle. (2003). Language choice, residential stability, and voting among Latino Americans. Social Science Quarterly. 84
Kaufmann, Karen M. (2002). Culture wars, secular realignment, and the gender gap in party identification. Political Behavior. 24(3): 283-307.
Lipset, S. and S. Rokkan (1967) Party Systems and Voter Alignments-Cross National Perspectives. New York: Free Press.
Mischel, J. 1966. A social learning view of sex difference in behavior. In the Development of Sex difference edited by E. E. Maccoby. Los Angeles: Stanford Univer. Press.
Macionis, J, J. & Plummer, K. (1998). Sociology. New York: Prentice Hall.Mott, Jonathan. (2006). Political Socialization. Retrieved on January 27, 2006 from
http://www.thisnation.com/socialization.html.
Munroe, Trevor. (2002). An Introduction to Politics: Lectures for First Year Students, (3rd ed). Kingston, Jamaica: University of the West Indies Press.
_______. (1999). Renewing democracy into the millennium. The Jamaican experience in perspective. Kingston, Jamaica: University Press of the West Indies.
_______. (1970) Political Change and Constitutional Development in Jamaica. Kingston: University of the West Indies, ISER.
Powell, Lawrence A., Paul Bourne, and Lloyd Waller. (2007). Probing Jamaica’s political culture, volume1: Main Trends in the July-August 2006 Leadership and Governance Survey. Kingston, Jamaica: Centre for Leadership and Governance.
Rae, D. (1967) The Political Consequences of Electoral Laws. New Haven, CT: Yale University Press.
Saalfeld, T. 2004. Party identification and the social bases of voting behaviour in the 2002 Bundestag election. German Politics 13(2):170-200.
Sanbonmatsu, Kira. (2002). Gender stereotypes and vote choice. American Journal of Political Science. 46(1): 20-34.
Stone, Carl. (1992). Patterns and trends in voting in Jamaica, 19550s to 1980s.______. (1981). Public opinion and the 1980s elections in Jamaica. Caribbean Quarterly
27(2): 1-19.______. (1989). Pollstering: “Perspective” interview. Caribbean Perspective 44/45:9______. (1978a) "Class and status voting in Jamaica." Social and Economic Studies 26:
279-293.______. (1978b) "Regional party voting in Jamaica (1959 – 1976). Journal of
Interamerican Studies and world affairs 220:393-420.
572
_____(1977) "Class and the institutionalization of two-party politics in Jamaica." Journal of Commonwealth and Comparative Studies 14: 177-196.
______(1974) Electoral Behaviour and Public Opinion in Jamaica. Kingston: Universityof the West Indies, ISER.
______(1973) Class, Race and Political Behaviour in Urban Jamaica. Kingston: University of the West Indies, ISER.
Thomlinson, R. (1965). Sociological concepts and research. Acquisition, analysis, and interpretation of social information. New York, U.S.: Random House.
Vassell, Linette. 2000. Power, governance and the structure of opportunity for women in
decision making in Jamaica. The construction of gender development indicators
for Jamaica. Kingston Jamaica: PIOJ/UNDP/CIDA Kingston
573
.About the Author
Paul Andrew Bourne is currently a health research scientist in the Department of Community Health and Psychiatry, Faculty of Medical Sciences, the University of the West Indies, Mona Campus, Kingston 7, Jamaica. He also lectures in Research Methods, and Elements of Reasoning, Logics and Critical Thinking at the Jamaica Constabulary Staff College. Bourne teaches Mathematics; Marketing; Marketing Management, and Science, Medicine and Technology at the University of the West Indies Open Campus sites; and lectures Mathematics and Social Research at the Montague Teacher’s College.
He was a political sociologist in the Department of Government, Mona Campus. Bourne has recently co-authored two monographs - (1) Probing Jamaica’s Political Culture: Main Trends in the July-August 2006 Leadership and Governance Survey, Volume 1; and (2) Landscape Assessment of Corruption in Jamaica.
Bourne was employed as a consulting biostatistician to the Caribbean Food and Nutrition Institute an affiliated of PAHO/WHO in Jamaica.
Paul Andrew Bourne’s areas of interest include Statistics, Demography, Political Sociology, Well-being, Elderly, Political Polling and Research Methods.
Department of Community Health and PsychiatryFaculty of Medical SciencesThe University of the West Indies, Mona Campus, Kingston, Jamaica
ISBN 978-976-41-0231-1
574