introduction to questionnaire design
Post on 23-Jan-2016
78 Views
Preview:
DESCRIPTION
TRANSCRIPT
Training Course on Basic Statistics for Research
August 24-28, 2009
STATISTICAL RESEARCH AND TRAINING CENTERJ and S Building, 104 Kalayaan Avenue, Diliman, Quezon City
Introduction toQuestionnaire Design
2
Statistical Research and Training Center Training Course on Basic Statistics for ResearchAugust 24 - 28, 2009
Questionnaire
refers to any list of questions, which can be self-administered by the respondent or read to the respondent by an interviewer
3
Statistical Research and Training Center Training Course on Basic Statistics for ResearchAugust 24 - 28, 2009
Process of Constructing the Questionnaire
1. Decide on the type of questionnaire that will be constructed.
2. Write down the questions.
3. Combine all questions and decide on which order to present them.
4. Write the introductory statement or cover letter.
5. Write the instructions to interviewers or respondents.
6. Design the Form.
7. If the study is a nationwide survey, translate the questionnaire to other major languages.
8. Pretest the questionnaire.
4
Statistical Research and Training Center Training Course on Basic Statistics for ResearchAugust 24 - 28, 2009
Strategies in Writing the Questions
Use the research objectives as the basis for your questions.
You may borrow questions that were already asked in past studies.
Anticipate the possible problem that you might encounter in extracting information from the respondents.
Consider the resources available. Consider the medium used in acquiring the responses Consider the level of measurement required by the
analysis that you have in mind.
5
Statistical Research and Training Center Training Course on Basic Statistics for ResearchAugust 24 - 28, 2009
Types of Questionnaire
Structured wording and order of questions are uniform for all respondents
Unstructured wording and order of questions can vary for different subjects; usually used for qualitative studies like FGDs, case studies, etc.
6
Statistical Research and Training Center Training Course on Basic Statistics for ResearchAugust 24 - 28, 2009
Types of Questions in Survey Research
Open-Ended Questions
Closed-Ended Questions
7
Statistical Research and Training Center Training Course on Basic Statistics for ResearchAugust 24 - 28, 2009
Open – ended :response categories are not specified; the respondents are free to answer as they please
Example:We would like to get your opinion regarding the effectiveness of the project in your barangay.
a. What do you think are the strengths of the project? Why did you say so?
b. How do you think can the project still be improved to meet its objective?
Types of Questions
8
Statistical Research and Training Center Training Course on Basic Statistics for ResearchAugust 24 - 28, 2009
Closed - ended respondent selects one or more of the specific categories provided by the researcher or the question is asked as an open question and the interviewer allocates the answer to the appropriate code category.
Example:
Do you practice family planning?
__ Yes __ No
If YES, what methods did you use?
__ Condom __Pills
__ Ligation __IUD
__ Injectables __Natural
__ Others, specify ____________________
Types of Questions
9
Statistical Research and Training Center Training Course on Basic Statistics for ResearchAugust 24 - 28, 2009
Advantages of Closed-Ended Questions
The answers are standard, and can be compared from person to person.
The answers are much easier to code and analyze, and often can be coded directly from the questionnaire.
A respondent who is unsure about the meaning of the question can often tell from the answer categories what is expected.
Irrelevant responses are avoided.
10
Statistical Research and Training Center Training Course on Basic Statistics for ResearchAugust 24 - 28, 2009
It is very easy for a respondent who does not know the answer or has no opinion to try to guess the appropriate answer or even to answer randomly.
The respondent may feel frustrated because the appropriate category for his/her answer either is not provided at all or is not provided in sufficient detail .
Variations in answers among the different respondents may be eliminated artificially by forced-choice responses.
Disadvantages of Closed-Ended Questions
11
Statistical Research and Training Center Training Course on Basic Statistics for ResearchAugust 24 - 28, 2009
They can be used when all of the possible answer categories are not known, or when the investigator wishes to see what the respondent views as appropriate answer categories.
They allow the respondent to answer adequately, in all the detail he/she likes, and to qualify and clarify his/her answer.
They can be used when there are too many potential answer categories to list on the questionnaire.
They allow more opportunity for creativity and self-expression by the respondent.
Advantages of Open-Ended Questions
12
Statistical Research and Training Center Training Course on Basic Statistics for ResearchAugust 24 - 28, 2009
They may lead to collection of worthless and irrelevant information.
Coding is often very difficult and subjective, leading to low inter-coder reliability.
Open-ended questions require superior writing skills, better ability to express one’s feelings verbally, and generally a higher educational level than do closed-ended questions.
Open-ended questions require much more of the respondent’s time and effort, and may engender a high refusal rate.
Disadvantages of Open-Ended Questions
13
Statistical Research and Training Center Training Course on Basic Statistics for ResearchAugust 24 - 28, 2009
Examples of Pitfalls in Question Construction
Incorrect ordering of questionsQ7. How would you rate the scent of lotion X?Q8. What do you think of lotion X?
Improvement: ________________________________________ ________________________________________
Double-barreled questions
“Does your department have a special recruitment policy for ethnic minorities and women?”
Improvement: ____________________________________ _____________________________________
14
Statistical Research and Training Center Training Course on Basic Statistics for ResearchAugust 24 - 28, 2009
Examples of Pitfalls in Question Construction
Words with double meaning
“Does your household participate in the activities of the barangay?”
The National Statistics Office defines the household as an aggregate of persons generally but not necessarily bound by ties of kinship, who live together under the same roof and eat together or share common the household food.
15
Statistical Research and Training Center Training Course on Basic Statistics for ResearchAugust 24 - 28, 2009
Examples of Pitfalls in Question Construction
Difficult words or technical terms
“What proportion of your evening viewing time do you spend watching news programs?”
Improvement: ___________________________________________ ___________________________________________
16
Statistical Research and Training Center Training Course on Basic Statistics for ResearchAugust 24 - 28, 2009
Presuming questions
“Do you think the Clean Air Act is being implemented properly by local authorities?”
Improvement: _____________________________________ ____________________________________
Examples of Pitfalls in Question Construction
17
Statistical Research and Training Center Training Course on Basic Statistics for ResearchAugust 24 - 28, 2009
Sensitive or Threatening Questions
“Do you masturbate? Yes _____ No _____ If yes, how often? Once a week, once a month, everyday?”
Improvement: ___________________________________________ ________________________________________
Examples of Pitfalls in Question Construction
18
Statistical Research and Training Center Training Course on Basic Statistics for ResearchAugust 24 - 28, 2009
Unrealistic questions
“What brand of perfume do you think you will be using three years from now?”
Improvement: ___________________________________________
__________________________________________
Examples of Pitfalls in Question Construction
19
Statistical Research and Training Center Training Course on Basic Statistics for ResearchAugust 24 - 28, 2009
Incomplete/non-exhaustive listing
“Did you learn about the brand from TV, radio, newspaper, or friends?”
Improvement: ____________________________________________
____________________________________________
Examples of Pitfalls in Question Construction
20
Statistical Research and Training Center Training Course on Basic Statistics for ResearchAugust 24 - 28, 2009
Biased Wording
“Do you think that decent, low-cost funerals are sensible?
Improvement: ___________________________________________ _________________________________________
Examples of Pitfalls in Question Construction
21
Statistical Research and Training Center Training Course on Basic Statistics for ResearchAugust 24 - 28, 2009
Leading questions
“The majority of physicians in the Philippines feel that smoking is harmful; do you agree?”
Improvement:__________________________________ ______________________________________
Examples of Pitfalls in Question Construction
22
Statistical Research and Training Center Training Course on Basic Statistics for ResearchAugust 24 - 28, 2009
Examples of Pitfalls in Question Construction
Double Negatives especially in agree/disagree items
“Please tell me whether you agree or disagree with the following statement about teachers in public schools: Teachers should not be required to supervise students in the halls, the lunchroom and school parking lots.”
Improvement: ____________________________________________
____________________________________________
23
Statistical Research and Training Center Training Course on Basic Statistics for ResearchAugust 24 - 28, 2009
Dangling Alternatives
“Would you say that it is very often, frequently, seldom, or never that you and your boss disagree about advertising campaigns?”
Improvement: _______________________________________________
_______________________________________
Examples of Pitfalls in Question Construction
24
Statistical Research and Training Center Training Course on Basic Statistics for ResearchAugust 24 - 28, 2009
Examples of Pitfalls in Question Construction
Insufficiently Specific or Unidimensional
“Are you satisfied with your canteen?”
Improvement: ___________________________________________
__________________________________________
25
Statistical Research and Training Center Training Course on Basic Statistics for ResearchAugust 24 - 28, 2009
Hypothetical Questions
“If you are the president of the Philippines, would you push through with the Charter Change?”
Improvement: _______________________________________________
_______________________________________
Examples of Pitfalls in Question Construction
26
Statistical Research and Training Center Training Course on Basic Statistics for ResearchAugust 24 - 28, 2009
Vague questions
“How often do you wash your hair?”
Very often
Often Not too often Never
Imrovement: ______________________________________ __________________________________________
Examples of Pitfalls in Question Construction
27
Statistical Research and Training Center Training Course on Basic Statistics for ResearchAugust 24 - 28, 2009
Example to Avoid Unrealistic Questions
Getting information on number of hours spent shopping
(i) Did you go shopping in the past week? (IF YES) How many hours did you spend shopping in the past week?
(ii) On a typical week, how many hours did you spend shopping? Or, on the average, how many hours do you spend shopping in a week?
(iii) Do you go shopping? (IF YES) On your last shopping trip, how many hours did you spend shopping?
28
Statistical Research and Training Center Training Course on Basic Statistics for ResearchAugust 24 - 28, 2009
Use of Simple Words
WORD SIMPLER WORDacquaint inform, tell
assist helpconsider thinkinitiate begin, startmajor important, chief, main,
principalmaterialize come about, happen, occurpurchase buysufficient enoughterminate end
29
Statistical Research and Training Center Training Course on Basic Statistics for ResearchAugust 24 - 28, 2009
Exercise:
Evaluate the following statements:1. According to the Chief Justice of the Supreme Court, the court
should be free of bribery and corruption.2. Do you think that the Congress, as an institution, is definitely free
from external controls, influence or pressure?3. Does the government agency have criminal and civil cases?4. What is the nature of the complaint against the Chief of the
Division?Have you ever filed an administrative complaint against a Division Chief?
5. Do you accept money for a favorable decision? If yes, how often?
6. Approximately how many administrative cases did you have in your agency from 2005 to 2008?
30
Statistical Research and Training Center Training Course on Basic Statistics for ResearchAugust 24 - 28, 2009
Filter and Contingency Questions
A contingency question is a question whose relevance to the respondent is determined by the response to an earlier filter question.
There are some questions that are not relevant to all of the respondents in the study, particular if the population under study is very heterogeneous.
Instead on using separate questionnaires for the different subgroups, only one questionnaire containing
filter and contingency questions is used.
31
Statistical Research and Training Center Training Course on Basic Statistics for ResearchAugust 24 - 28, 2009
Possible Format of Filter and Contingency Questions
Instructions telling which to answer next in the event of the category being chosen are placed beside each response category.
In the past two months, did you have any illness that kept you in bed, indoors, or away from your usual activities?
Yes …………(ASK A)……………1
No…………..(SKIP TO Q.25)……2
Or
Yes…………………….1 ASK A
No ……………………. 2 SKIP to Q.25
32
Statistical Research and Training Center Training Course on Basic Statistics for ResearchAugust 24 - 28, 2009
Possible Format of Filter and Contingency Questions
Used boxed contingency questions with no arrows leading to them.
Have you ever had a miscarriage?
( ) Yes
( ) No
If YES: How old were you when you had your first miscarriage? ____________
33
Statistical Research and Training Center Training Course on Basic Statistics for ResearchAugust 24 - 28, 2009
Response Category Format
Response categories should be exhaustive. Response categories should be mutually exclusive. Response categories should be made easy to data collectors,
respondents, or coders to select appropriate options. Response categories should include the “Other” option with
sufficient space for writing answers. Response categories should anticipate analytical needs and
enable the collection of data that are suitable for those analyses.
It should be decided whether single or multiple answers are to be allowed.
34
Statistical Research and Training Center Training Course on Basic Statistics for ResearchAugust 24 - 28, 2009
Response Category Format
Number of Categories There is no optimal number of categories. 5 to 6 categories are commonly used
Odd or Even Number of Categories If odd, usually the middle category is the neutral position. If even, there will be no neutral category. Thus, the
respondent is forced to take a position.
35
Statistical Research and Training Center Training Course on Basic Statistics for ResearchAugust 24 - 28, 2009
Some Types of Category Formats
Verbal rating scale This requires a respondent to indicate their position by
selecting among verbally identified categories. Examples:
Strongly agree, Agree, Neither Agree nor Disagree, Disagree, Strongly Disagree
Very important, Important, Somewhat Important, Not at all Important
Excellent, Good, All right, Poor, Bad
36
Statistical Research and Training Center Training Course on Basic Statistics for ResearchAugust 24 - 28, 2009
Some Types of Category Formats
Graphic Rating Scale This requires a respondent to indicate his position on a
continuum which ranges from one extreme of the attitude in question to the other extreme.
Since the respondent is not using a predetermined set of response categories, the researcher cannot determine how a particular respondent has subdivided the continuum.
Example: “To what extent do you feel employees should have a voice in determining performance evaluation content?”
Employees should make Performance evaluation matters
the decisions about 0 1 2 3 4 5 6 7 8 9 10 should be left entirely
performance to the Administration
evaluation matters
37
Statistical Research and Training Center Training Course on Basic Statistics for ResearchAugust 24 - 28, 2009
Some Types of Category Formats
Rank-Order Scale This involves having the respondent rank various objects with respect to
the attitude in question. This is actually a forced scale. It is possible that respondent actually dislikes all of the objects they were asked
to rank according to a favorable characteristic. This attitude is not captured by the scale.
Example: The following are some of the problems faced by the local government. Please rank them in terms of importance, from 1 (most important) to 5 (least important).
_____ Pollution
_____ Garbage
_____ Traffic
_____ Drug Addiction
_____ Crime
38
Statistical Research and Training Center Training Course on Basic Statistics for ResearchAugust 24 - 28, 2009
Some Types of Category Formats
Checklist Question provides more than two alternatives from which the respondent can
choose as many responses that apply to them. Example: What kind of ball sports do you like to play? basketball volleyball football tennis baseball water polo
others, pls. specify _____________
39
Statistical Research and Training Center Training Course on Basic Statistics for ResearchAugust 24 - 28, 2009
The order of questions should be in logical sequence so that the trend of thought of the respondent is not broken.
The first part of the questionnaire are on Identification Information consisting of the geographic identification, interviewer’s record and the information about the sample.
Sensitive questions are placed towards the end of the interview schedule or questionnaire.
Question Order
40
Statistical Research and Training Center Training Course on Basic Statistics for ResearchAugust 24 - 28, 2009
Approaches to Question Sequence
Funnel Approach Uses a general-to-specific question flow.
Work Approach Requires that difficult to answer questions are placed deep
inside the questionnaire.
Sections Approach Arranges the questionnaire into sections or topic category
divisions that are dictated by the objectives of the study
41
Statistical Research and Training Center Training Course on Basic Statistics for ResearchAugust 24 - 28, 2009
Approaches to Question Sequence
Quintamensional Design This 5-question approach was developed by George Gallup to
cover the most essential features of an opinion:
Open-ended question on awareness/knowledge of the issue
Open-ended question on the attitudes toward the issue
Closed-ended question on attitude towards specific issues
Open-ended question on reason behind attitude
Closed-ended question on the intensity of attitudes
42
Statistical Research and Training Center Training Course on Basic Statistics for ResearchAugust 24 - 28, 2009
Questionnaire Lay-out & Design
Flow of questions should be respondent-friendly .
Vary questions by length and type.
Place scale items according to response required.
Separate reliability-check question pairs.
43
Statistical Research and Training Center Training Course on Basic Statistics for ResearchAugust 24 - 28, 2009
Physical Design of a Questionnaire
(1) The appearance of the form should be simple and well spaced;
(2) The size and type of font used should be readable;
(3) The font size and type used for questions and directions should be different.
44
Statistical Research and Training Center Training Course on Basic Statistics for ResearchAugust 24 - 28, 2009
Basic Contents of the Cover Letter/Introduction
Identification of the persons or organization conducting the study
Explanation of the importance of the study and reason/s why it is being conducted
Reason why the respondent should be the one to answer the questionnaire
Assurance of no right or wrong answers Guarantee the confidentiality of answers
45
Statistical Research and Training Center Training Course on Basic Statistics for ResearchAugust 24 - 28, 2009
Example of an Introduction
Good morning/afternoon/evening. I am Josie Almeda from the UP School of Statistics Research Foundation. This study seeks to know how Filipinos feel about various issues and problems affecting their lives. This information is valuable both for planning and for scientific research. Your household is very important to this survey because it represents hundreds of others which are not in our sample. Everything you tell us will be strictly confidential. Your name will be in no way connected to the findings of this important study.
46
Statistical Research and Training Center Training Course on Basic Statistics for ResearchAugust 24 - 28, 2009
The Pretest
Objective: To determine if there are errors in the questionnaires or areas that need improvement such as ordering of questions, wording, and the natural flow from one section to another.
Respondents: No rules on who the respondents should be in the pretest and on the size of the sample.
47
Statistical Research and Training Center Training Course on Basic Statistics for ResearchAugust 24 - 28, 2009
Types of Pretest
Participating Pretest Respondents are told that this is a practice run, and are asked
to explain their reactions and answers.
This allows for a very detailed probe on certain questions such as:What did the whole question mean to you? What did the (term) make you think of?
Undeclared Pretest The respondent is not told that this is a questionnaire under
construction, and the interviewer plays it straight.
48
Statistical Research and Training Center Training Course on Basic Statistics for ResearchAugust 24 - 28, 2009
Evaluating the Pretest
Before the pretest, tell interviewers to write marginal comments on the questionnaire.
Conduct an oral debriefing after the pretest so that the interviewers can report on their experience.
Some of the issues that can be discussed are:Were there questions that made the respondent uncomfortableWhich questions have to be repeatedDid the respondent misinterpret any questionWhich questions were the most difficult or awkward to readDid any of the sections bore the respondentWere there questions that respondent found stupidWere there any section in which the respondent would have liked the opportunity to say more
49
Statistical Research and Training Center Training Course on Basic Statistics for ResearchAugust 24 - 28, 2009
Evaluating the Pretest
Review answers to open-ended questions
Tabulate the number of no responses, unable to answer, and don’t knows for each question.
Investigate questions that yield too many such answers.
Watch out for response sets.
Look at the variability of responses.
50
Statistical Research and Training Center Training Course on Basic Statistics for ResearchAugust 24 - 28, 2009
WORKSHOP
Write a short questionnaire to measure employees’ job satisfaction on their current position. Incorporate topics on the employees’ preferences. Put questions regarding the personal characteristics of employees like age, sex, department, position, and others. Maximum of 15 questions only. Keep in mind the guidelines on questionnaire construction.
Include a question that allows for multiple response.
51
Statistical Research and Training Center Training Course on Basic Statistics for ResearchAugust 24 - 28, 2009
Data Coding
Coding is a process of grouping the responses to a question into categories and assigning numbers, characters, and/or other symbols, called codes, to these categories.
52
Statistical Research and Training Center Training Course on Basic Statistics for ResearchAugust 24 - 28, 2009
REASONS FOR CODING
To simplify the recording of information.
To preserve the confidentiality of the responses either for personal, industrial, or security reasons.
To make it easier to enter the data into the computer, minimize storage space, and speed up processing.
To facilitate the management of files (e.g., sorting, merging, transforming, listing, etc.) using the coded values of one or more key variables.
To perform data analysis using computer software like MS Excel, STATISTICA, MICROSTAT, etc., which require numeric inputs.
53
Statistical Research and Training Center Training Course on Basic Statistics for ResearchAugust 24 - 28, 2009
Basic Principles in Coding
Consistent codes for identical answers should be used throughout the instrument.
Consistent conventions should be used to record missing data, and when appropriate, to indicate the reasons data are missing.
Assign codes that minimize transformations during data analysis
Select standard or commonly used classification whenever possible
54
Statistical Research and Training Center Training Course on Basic Statistics for ResearchAugust 24 - 28, 2009
TYPES OF CODES
Listing codes
Bracket codes
Scale codes
Series codes
System codes
55
Statistical Research and Training Center Training Course on Basic Statistics for ResearchAugust 24 - 28, 2009
LISTING CODES
Listing codes simply correspond to a list of choices or possible responses.
Successive integers, beginning with 0 or 1, are assigned to the items in the list of possible responses.
56
Statistical Research and Training Center Training Course on Basic Statistics for ResearchAugust 24 - 28, 2009
Example:
For questions which allow only one answer, each possible answer can be assigned a code from 1 to n.
Example : Religion
[1] Catholic [3] Iglesia ni Kristo[5] Others, specify
[2] Protestant [4] Fundamentalist
57
Statistical Research and Training Center Training Course on Basic Statistics for ResearchAugust 24 - 28, 2009
BRACKET CODES
Bracket codes represents ranges or class intervals.
These are frequently used for numerical responses like income and age in which the respondent may not be able to provide an exact answer.
58
Statistical Research and Training Center Training Course on Basic Statistics for ResearchAugust 24 - 28, 2009
Example:
Q: What is your monthly income ?(1) less than P 10000.00(2) P 10000.00 to less than P 30000(3) P 30000.00 to less than P 50000(4) P 50000 or more
Provide the codes of the following answers:Ans 1: "My income is P 5000."Ans 2: "My income is between P 15000 and P 35000.”Ans 3. "My income is not more than P 50000".Ans 4. "My income is P 49999.75".
59
Statistical Research and Training Center Training Course on Basic Statistics for ResearchAugust 24 - 28, 2009
SCALE CODES
Scale codes represent grouping of responses which vary in degree or intensity along a continuum.
We represent each segment by a number which serves as the code for any response falling within it.
We often use these for coding responses to attitudinal questions.
60
Statistical Research and Training Center Training Course on Basic Statistics for ResearchAugust 24 - 28, 2009
Example:
1. Two-point scale
Q: "How do you rate the overall performance of this administration?"
1 2
__________________
Low High
Ans: "Inspite of the economic crisis, the performance of
this administration is above my expectation."
Code: ______
61
Statistical Research and Training Center Training Course on Basic Statistics for ResearchAugust 24 - 28, 2009
Example:
2. Three-point scale
(a) This scale has a middle or neutral position
Q: "Do you get along well with your supervisor?"
1 2 3__________________________________
YES YES IN SOME WAYS, NO
NO TO OTHERS
Ans: "Sometimes."
Code: _______
62
Statistical Research and Training Center Training Course on Basic Statistics for ResearchAugust 24 - 28, 2009
Example:
3. Four-point scale(a) This scale has no middle or neutral position.
Q: "How sure are you about the outcome of the election in your province?"1 2 3 4
__________________________________Certain Fairly Fairly Uncertain
certain uncertain
Ans: "I have no idea."Code: _______
63
Statistical Research and Training Center Training Course on Basic Statistics for ResearchAugust 24 - 28, 2009
Example:
4. Five-point scale(a) We use a five-point intensity to code responses with
degree of feeling.Q: "How satisfied are you with your present job?" 1 2 3 4 5
_____________________________________________Very Satisfied Neutral Dissatisfied VerySatisfied Dissatisfied
Ans: "Neither satisfied or dissatisfied."Code: ______
64
Statistical Research and Training Center Training Course on Basic Statistics for ResearchAugust 24 - 28, 2009
Series Codes
Multi-digit codes that are divided into two sets – the first set of leading digits represents a major classification while the second set of trailing digits represents a specific category within the major classification.
This coding scheme can be extended to more than 2 levels.
65
Statistical Research and Training Center Training Course on Basic Statistics for ResearchAugust 24 - 28, 2009
Example of Series Codes
Why do you think that now is a good time to buy a car?
Response:
PRICE and CREDIT Prices are lower now compared to before 11 Uncertain about price in the future – may go up 12 Credit is easy to get 13 Attractive car loan terms 14
66
Statistical Research and Training Center Training Course on Basic Statistics for ResearchAugust 24 - 28, 2009
Continuation of Example of Series Codes
FEATURE of CAR Quality is good/better 21 New models have better features 22 Small cars are available 23
GOOD ECONOMIC PERFORMANCE OF COUNTRY People can afford to buy 31 High employment rate 32
67
Statistical Research and Training Center Training Course on Basic Statistics for ResearchAugust 24 - 28, 2009
SYSTEM CODES
System codes are used primarily in computer data processing to identify and/or classify respondents, households, products, accounts and other data items.
They are usually combinations of numbers and letters (called alphanumeric).
By using the digit 0 through 9 and the 26 letters of the alphabet, a total of 36 characters, 1,679,616 unique four-character codes can be generated.
68
Statistical Research and Training Center Training Course on Basic Statistics for ResearchAugust 24 - 28, 2009
Example:
The Philippine Standard Occupation Classification (PSOC) uses group classification codes which consist of three levels: major, minor, and unit groups.
The major groups are the highest level, of occupation aggregate and represent broad fields of work.
Each major group has a one-digit code.
Minor groups represent the second level of occupation aggregate and each minor group has a two-digit code.
Finally, unit groups are the third level of occupational groupings with each unit group assigned a three digit code.
69
Statistical Research and Training Center Training Course on Basic Statistics for ResearchAugust 24 - 28, 2009
Example:
List of Philippine Standard Classifications Philippine Standard Geographic Classification
Philippine Standard Industrial Classification
Philippine Standard Commodity Classification
Philippine Standard Occupational Classification
70
Statistical Research and Training Center Training Course on Basic Statistics for ResearchAugust 24 - 28, 2009
Example:
Major Group 0/1: Professional, Technical, and Related
Workers
0-8 Mathematicians, Statisticians, System Analysts and
Related Workers
0-81 Mathematicians and actuaries
0-82 Statisticians
0-83 Systems analyst
0-84 Statistical and Mathematical technicians
71
Statistical Research and Training Center Training Course on Basic Statistics for ResearchAugust 24 - 28, 2009
CODING OF NO RESPONSE
Types of No Response:
A. No Opinion
B. Don't Know
C. Refuse to answer
D. Not applicable
E. Missing (in general)
72
Statistical Research and Training Center Training Course on Basic Statistics for ResearchAugust 24 - 28, 2009
NO RESPONSE
A. NO OPINION NO OPINION should be coded NO or given a code
number directly following the other categories.
Example:
Did you find Mr.______adequate in helping you solve your problems?
________Yes ________No
Answer: No Comment.
We could give Yes a code (1); No a code (0) and a No Opinion/No Comment a code (2).
73
Statistical Research and Training Center Training Course on Basic Statistics for ResearchAugust 24 - 28, 2009
NO RESPONSE
B. DON'T KNOW
occurs when we ask Knowledge questions
assign a 'DON'T KNOW' response a code number
directly following the other categories.
74
Statistical Research and Training Center Training Course on Basic Statistics for ResearchAugust 24 - 28, 2009
NO RESPONSE
Example: How much did you and your household spend on
transportation this past month ?_____(1) less than P200_____(2) P200 to less than P500_____(3) P500 to less than P1000_____(4) P1000 to less than P2000_____(5) more than P2000_____(6) don't know
75
Statistical Research and Training Center Training Course on Basic Statistics for ResearchAugust 24 - 28, 2009
NO RESPONSE
C. REFUSES TO ANSWER We sometimes, unfortunately, encounter not-so
receptive respondents who do not like to readily volunteer their responses.
At times it may be due to an offensive question or altogether sensitive topics.
In cases like these, we assign it a code '9', or '99', or '9999' following the number of columns allotted for the variable, i.e., if no other category has been previously assigned to this.
76
Statistical Research and Training Center Training Course on Basic Statistics for ResearchAugust 24 - 28, 2009
NO RESPONSE
Example:How much was your family income this past month ?
Answer: refuses to answer
Code: _______
77
Statistical Research and Training Center Training Course on Basic Statistics for ResearchAugust 24 - 28, 2009
NO RESPONSE
D.NOT APPLICABLE Some responses to our questions may be dependent
upon answers to previous questions or to some attributes of our respondents. So that if the attribute is not present, the question becomes irrelevant.
In cases like these, we code it as NA; or as '9',or '99' or even '99999', again depending on the number of columns allotted to our variable.
78
Statistical Research and Training Center Training Course on Basic Statistics for ResearchAugust 24 - 28, 2009
NO RESPONSE
Example :Do you own a farm ? _____Yes _____NoWhat is the land area of your farm ? _____(hectares)
Code: ________
79
Statistical Research and Training Center Training Course on Basic Statistics for ResearchAugust 24 - 28, 2009
NO RESPONSE
E.MISSING
refers to responses which were not recorded for various reasons, which could not be verified. Hence, we assign to it a special "missing" value symbols.
Following are some characters which are recognized automatically as missing values by three major statistical software packages:
BMDP: blankSAS : blankSPSS : blank
80
Statistical Research and Training Center Training Course on Basic Statistics for ResearchAugust 24 - 28, 2009
Coding Multiple Response
Multiple Dichotomy Method For each possible response, create a dichotomous
variable with a value of 1 to indicate that the response was given and a value of 0 if not.
81
Statistical Research and Training Center Training Course on Basic Statistics for ResearchAugust 24 - 28, 2009
Example of Multiple Dichotomy Method:
Q: What type of TV shows do you watch ? (Please check your answers )[ ] action[ ] drama[ ] game[ ] sport
[ ] none[ ] Pls. specify __________________
A: "I watch action and game shows.”
Code: _____________
82
Statistical Research and Training Center Training Course on Basic Statistics for ResearchAugust 24 - 28, 2009
Example Of Multiple Dichotomy Method
Field Name CodeAction 1 watched 0 did not watchDrama 1 watched
0 did not watchGame 1 watched
0 did not watchSport 1 watched
0 did not watchNone 2 does not watch
83
Statistical Research and Training Center Training Course on Basic Statistics for ResearchAugust 24 - 28, 2009
Some Rules for Designing Code Frames
Create a sufficient range of codes and variables so that the coder need not force data into categories.
Include an “Others” category if you are not assured of the completeness of codes.
Monitor the frequency of the “Others” category. A large frequency indicates a failure in the coding scheme.
Group together related categories of information and use series codes.
84
Statistical Research and Training Center Training Course on Basic Statistics for ResearchAugust 24 - 28, 2009
Quality Control
Training of Encoders Encoders should be trained and supervised
Procedures to be used in coding and entering data should be reviewed in sequence and in detail.
Encoders should be provided with codebooks
85
Statistical Research and Training Center Training Course on Basic Statistics for ResearchAugust 24 - 28, 2009
Quality Control
Range Checking This checks the variable values against predefined
maximum and minimum bounds.
This is often done to catch spurious values or keyboard entry errors.
Cleaning rules can be formulated as logical expressions making various assertions about the data.
86
Statistical Research and Training Center Training Course on Basic Statistics for ResearchAugust 24 - 28, 2009
Quality Control
Consistency Checks Using Contingency Tables The pairs of variables whose answers should be
consistent with each other must be cross-tabulated. The cells representing inconsistent responses must be
empty. After cleaning, the contingency tables are constructed
once more to check that all answers are consistent with each other.
87
Statistical Research and Training Center Training Course on Basic Statistics for ResearchAugust 24 - 28, 2009
Quality Control
Reentering the data file A very expensive method but this is recommended if a
close-to-zero encoding error is desired.
Some software include a “verify” mode in which the data being entered the second time are automatically compared with those entered the first time.
88
Statistical Research and Training Center Training Course on Basic Statistics for ResearchAugust 24 - 28, 2009
WORKSHOP
Construct a code frame for the questionnaire constructed in the last workshop. For each part of the questionnaire, provide the variable name, variable label and code of response.
89
Statistical Research and Training Center Training Course on Basic Statistics for ResearchAugust 24 - 28, 2009
Example
Respondent Identification Variables
Variable Name
Variable Label Code – Response
ID ID Number Encode as is
Brgy Barangay1 – Ibayo2 – Labac
Type Type of Barangay1 – Urban2 – Rural
Resp Age Classification1 – Adult2 – Youth
Intervr Name of Interviewer Encode as is
90
Statistical Research and Training Center Training Course on Basic Statistics for ResearchAugust 24 - 28, 2009
Example
Demographics. Personal Information
Variable Name
Variable LabelCode – Response
A1 A01Name Name
A2 A02SexSex 1 – Lalaki
2 – Babae
A3 A03AgeAge Input number in years
A4 A04CS
Civil Status1 – Single2 – Married3 – Separated4 – Widowed5 – Others
top related