seminar 1 (2011) an introduction to data entry, data analysis, and graphing using spss
TRANSCRIPT
Seminar 1 (2011)
An introduction to data entry, data analysis, and graphing
using SPSS
What is SPSS?
Statistical Package for the Social Sciences
A commonly used computer package in business, government, research and academic organizations.
It is especially used in the social and behavioural sciences for processing and analysing data and for
producing graphs.
In this session
Learn how to navigate through the different windows of SPSSLearn how to open and save data files Learn how to calculate simple statistics from variables in a data file Learn how to calculate new variables using the COMPUTE functionLearn how to compare different subsets and groups of data using the SPLIT FILE and SELECT CASES IF functionsLearn how to produce and edit simple graphs in SPSS and incorporate them into a word document
Seminar WorksheetsDuring the seminar you will have worksheets to complete.
When you complete the worksheet enter your answers on the online worksheet which is on the U24103 resources page.
On Friday you will receive a email with your mark and the correct answers.
Why are Statistics important?
They help put a number in its context* *if used correctly
4,065,000 to 10, 000 which you do by dividing by 406.5.
Then divide the 584 pregnancies by 406.5 which gives 1.43.
To get the percentage you divide 1.43 by 100 which gives 0.014%
Putting a number in its contextJanuary 8th, 2011 by Ben Goldacre in bad science
The many Media outlets reported the story that 584 woman with the contraceptive implant had unplanned pregnancies.
MHRA estimate that 1.355 million implants have been sold.
Each implant lasts 3 years, this gives a total exposure time of 4.06 million women-years at risk.
584 unplanned pregnancies in this exposed population means there were 1.4 unwanted pregnancies reported for every 10,000 women with implants per year.
Or, you can say that the failure rate is 0.014% per year.
This is rather good:– Implants are still the most reliable form of
contraception
Back to SPSS
The headline:
Is a lot less scary when the number 584 it is put in context:
The failure rate for the contraceptive implant is 0.014% per 10,000 per year Or without numbers:
Implants are still the most reliable form of Contraception
Switch between “Data view” &
“Variable view”
Where to find help
File menu.
Here is where
you load and
save files.
Current ‘active’ data-entry cell
SPSS version 19 interface
A sample of 50 people asked to answer a set
of questions for a survey of health
behaviour.
How to open an SPSS data file
File menu: Open: Data
U:\data U24103 Seminar 1 SurveyData_Seminar1
Select your file in the “Open Data” window
The SPSS ‘Data View’ window
Each row represents a different participant
Each column represents a
different variable
The SPSS ‘Variable View’
Each row represents a
different variable (column in the data view)
Each column represents a different
property of that variable
This is where you tell SPSS what kinds of variables you have
“Measure” property: what is the ‘level of
measurement’ in the variable
Nominal, Ordinal, or Scale (i.e. interval or ratio)
“Type” property: what is contained in the variable:
Numeric = Numbers
String = Text
Name is where you give your variable/column a short name. Remember
you cannot use a space so you need to use a “_”
Label is where you can give your variable a
longer name
If you are missing a some data you should not just leave the cell blank. You should decide on a
number and tell SPSS what it is in this box. People commonly use “666” or “999”
RolesWhat the variables role is Input = independent variable (IV).Target = dependent variable (DV).Both= ether IV or DVNone. The variable has no role assignment.Partition. used to partition the data into separate samples.
“Values” You can assign labels for each value of a variable. E.g. Male = 1,
Female=0.
Analyse menu
DESCRIPTIVES function
(and all other functions relating to ANALYSIS of DATA)
Graph menu
Bar Chart function
(and other graph-related functions)
Transform menu
COMPUTE function
(and all other functions relating to modifying values and producing new variables from your data)
Data menu
SPLIT FILE function
(and other functions relating to selecting/ ordering data based on criteria)
Orienting through SPSS menu options
Click here to see any names you have given to numbers in the Values
The SPSS OUTPUT window
Click on these to navigate to the output
from previous analyses
Use: “–” to hide things &
“+” to show them again Output from an analysis
Print out of what you told SPSS to do
Where the data file is saved
Do not close this window keep the same window
open for the whole session
Generating simple descriptive statistics in SPSS
SPSS can generate a multitude of statistics. We will not be using all of them in this course.
Analyse menu
DESCRIPTIVES
(and all other functions relating to ANALYSIS of DATA)
Today we are using Descriptive Statistics to look at the variables in your data, measures of central tendency, measures of dispersion etc
“Frequencies” function
“Descriptives” function
“Frequencies” and “Descriptives” have a lot of overlapping functions (e.g. both can give means, standard deviations).
Frequencies has a greater range of options (e.g. it can also compute medians, modes).
Descriptives
Left hand box contains all the
numerical variables in your
data set
Put the variables that you want to compute statistics from by selecting them and clicking this arrow to move them into the right hand side box.
“Options…”: This gives some options for the type of statistics you wish to show.
Click on “Range” tick-box so you also
get this statistic in your output
The OUTPUT window for Descriptives
Each row designates one of your variables.
The columns show the calculated values for each of the statistical measures you ask for
Number of participants
Q1: Find out the means and
ranges of the heights and
weights of the participants
using the ‘Descriptives’
command.
Frequencies
If checked ‘Frequencies’ outputs a list of occurrences of a particular value (useful for categorical and ordinal variables)
To select multiple items in a row click on the first item you want then hold down the “shift” key and click on the last one you want
The “Statistics…” option gives output options for ‘Frequences’
Options for ‘median’
and ‘mode’
Options for ‘Standard deviation’ and ‘standard error’
Frequencies table.
e.g. for the ‘Cigarettes’ variable it tells you how many (and what percentage) of your data file are Smokers.
Each column designates a separate variable. The different calculated values for that variable are shown on the individual rows.
Q2: Use ‘Frequencies’ to find out
a) How many males took part in the survey?
b) To calculate what percentage of the
sample are ‘Skilled labourers’
c)To find the median weight of the sample
How to produce new variables from values in a data-file
SPSS allows us to calculate new variables based on combining the variables we already have in our file.
For example, Body Mass Index (BMI) is a measure which is mathematically derived from a persons height and weight.
Formula for BMI
BMI is an (indirect) indicator of the proportion of body fat a person has and is thus a useful health measure.
The Compute functionTransform menu
COMPUTE function
(and all other functions relating to modifying values and producing new variables from
your data)
Enter the Target variable name here.
This is the name that SPSS will give the new calculated variable.
Note that spaces and certain characters aren’t allowed in variable names (e.g. symbols such as ‘&’). The “Label” option box below it allows you to specify things like the level of measurement etc. of the variable.
Type a valid mathematical formula to calculate the variable here.
If you wish to use existing variables then these can be moved in using the arrow from the horizontal variable box on the left.
We want to calculate BMI
To do this we need peoples weight in metric Kilograms rather than imperial pounds (lb). The formula to do this is: kg = lb ÷ 2.2
First type a name for the new variable here:
e.g. Weight_in_kg
(**don’t use an existing variable name otherwise it will be overwritten**).
To calculate the Weight in pounds we need the data variable for the weight in Kg.
The conversion formula then requires us to divide this variable’s values by 2.2.
Weight_In_lb / 2.2
(N.B. for computers the * is a multiplication sign and the / is the division sign).
What we are telling SPSS to do the sum:
Weight_In_Kg = Weight_In_Lb / 2.2
Press and SPSS will now create and compute this new variable based on the formula you have given it.
In the data view window you should see that a new variable called Weight in kilograms has been calculated for each of the 50 participants in the sample.
In the output window SPSS has printed out the sum you put into the compute window. Q3: Now you know how to use
‘Compute’ try to create a variable for
BMI. Remember the formula is:
Here are the BMI values for the first five participants (P00001 to P00005).
How to analyse data groups in SPSSOften we are interested in looking at or comparing values of different groups within our data. For instance we might want to compare the average height of males and females.
SPSS has several ways to allow us to do this.
Data menu
SPLIT FILE,
SELECT CASES IF… function
(and all other functions relating to selecting or ordering data based on given criteria)
Or you can use the SPLIT FILE button
Change to ‘Compare groups’
If we want to compare Males and Females then move the ‘Gender’ variable to this box
Split file function
Note that in the bottom-right corner of the SPSS data window it informs you that the file is now split by gender.
This remains the case until you turn this off again in the ‘Split File’ function.
Select :
Now when you run any analysis again (e.g. Descriptives). You get separate values for Males and Females in the table.
Q4: Use ‘Split File’ to generate the
mean heights of males and females in
our sample.
**When you have done this make sure
you turn off Split file again**.
Sometimes we wish to include only a subset of the cases (participants) in our analysis. The ‘Select Cases If’ function allows us to include only a subset of cases (and ignore others) based on a criterion that we give it.
Select Cases If.. function
Select
“If condition is satisfied” Then press the IF button so we can enter our criterion.
Criterion window. What we need to do here is enter a Boolean condition (i.e. a mathematical statement which is either True or False)
Gender = 0
What this is telling SPSS is to select only those cases (participants) which have the Gender value of “0” (and therefore ignore all those that have the value 1).
In other words - select only if the participant is ‘Female’ otherwise ignore
The is equal to sign ‘=‘ is a commonly used relation in Select Cases IF statements
Others common signs for this function are:
greater than ‘>’
less than ‘<‘
not equal to ‘<>’ (note Gender <> 1 would have the same effect here as Gender = 0)
P.S. How to remember ‘less than’ & ‘more than’?
So the p value is less
than .05 or p<.05So the p value is greater
than .05 or p>.05
Pacman’s evil statistics-loving twin brother always eats the largest number
P .05 .05P
“Filter On”
You can see that the male cases are temporarily crossed out.
A new filter variable (called filter_$) is inserted
If you now run any analysis with the filter ON the analysis will only be performed on the selected cases (others will be
ignored in the calculation). **Remember to always turn off the filter after you finish with it in your
analyses**
Go back to the Select Cases IF menu and click on “Select all cases”.
Q5: Using Select Cases IF and
the ‘Descriptives’ function,
calculate the mean weight for
people who drink less than 15
units of alcohol a week.
You now have been shown the basics of data handling in SPSS.
Now might be a good time to save a personal copy of the data file onto your personal folder (H:) or pen-drive.
Type your file name here
Q6: Answer the following questions using what you have learned (**Remember to take off Filter/Split File after use**)a) What is the mode average of sleep that participants in our sample have? b) What percentage of our sample are Students? c) Do males or of females have a larger standard deviation for BMI? d) What is the median hours of sleep that someone in a manual labour job
reports they get? e) How many people in our sample are aged 35 or over? f) Who has a higher mean BMI in our sample, Smokers or non-smokers?
Q7: Some more difficult questions (note that for these questions AND, OR, NOT can be used as well in Boolean conditions)g) How many people in our sample are both smokers and drink 15 or more
units of alcohol per week? h) What is the mode average of units alcohol drank by someone who is over
the age of 45 and is in either in a manual-labour, skilled labour, or administrative/clerical/sales job?
We are going to stop for 20 min so you can work through Q6 & Q7 and
have a break
Q6: Answer the following questions using what you have learned
a) What is the mode average of sleep that participants in our sample have?
......8......
b) What percentage of our sample are Students?
...14..........
c) Do males or of females have a larger standard deviation for BMI?
...Females….
d) What is the median hours of sleep that someone in a manual labour job reports they get?
....8.5.....
e) How many people in our sample are aged 35 or over?
.....24........
f) Who has a higher mean BMI in our sample, Smokers or non-smokers?
.. non-smokers.....
Q7: Some more difficult questions)
a) How many people in our sample are both smokers and drink 15 or more units of alcohol per week?
..6...
b) What is the mode average of units alcohol drank by someone who is over the age of 45 and is in either in a manual-labour, skilled labour, or administrative/clerical/sales job?
... up to 14 units per week....
Using the graph functions
Graph menu
All graph-related
functions
SPSS can plot graphs from any of the data in your file.
HistogramsAre used to look at the distribution of data.
Here is an example for age. This variable is clearly not very normally distributed.
Q8: Create histograms for height,
weight, BMI from the data file. Do
these variables show a normal
distribution?
Useful for comparing different participant groups on some measure.Requires two variables
Bar chart
One usually ordinal or scalar for the Y axis.
One categorical (for the x-axis)
X-axis variable
(what groups do you want the columns to represent)
Y-axis variable
(what values do you want the height of the individual columns to show)
Select whether the height of the columns represents the mean, median or mode average (or some other measure) of the group.
An example SPSS bar chart showing a difference in height between the genders.Q9: Now produce a Bar graph
showing the median BMI for the
different occupation groups.
Which of the different occupation
groups has the lowest median BMI?
Scatterplot chartUseful for plotting the relationship between two interval (or ratio) level
variables
Need to give two variables:– One for the x-axis and One for the y-axis
You can have different markers for different groups
Each point is an individual participant’s score on the two values.
Double click on the graph to open chart editor
Q10: Now produce an SPSS graph for height (Y
axis) verses weight (X axis) where gender is
distinguished with different markers.
Edit it so it is easy to understand when printed in
black and white
Are the lines of best fit for males and females
roughly parallel?
Right click on the finished chart.
Copy the chart.
Open Word.
Paste into word as a picture.
Self-study exercises for Seminar 1
That’s all for today. Its worth spending a bit of time on your own using SPSS to really familiarise yourself with its functions.
Try some exercises from the online book on the psychology resources page in the Statistics folder :
Secure Resources SPSS Version 17
A Beginner's Guide to SPSS for Windows: Entering and Analysing Questionnaire data
Using SPSS…– Open the data file “spssraw.sav”
this can be found at: u:\data\SOCSCI\spssraw.savHave a look particularly at section 1. (pg. 1-6)Have a look at section 6. and section 7 (pg. 22-31) Have a look at section 10 (pg. 112-123).