2. Types of questions
1 Dichotomous (Yes/No, Male/Female)
2 Multiple choice (select only one from several options)
3 Likert scale (e.g. Strongly disagree … Strongly agree)
4 Rating (numerical scale, e.g. from 1 to 10)
5 Multiple response (checklist)
6 Ranking
7 Open-ended (qualitative – not to be analyzed by SPSS)
3. Hot Tip
Branching should be avoided as far as possible:
Example:
If ‘Yes’, go to question 14;
If ‘No’, go to question 16.
4. Questionnaire
Once your questionnaire is designed, perform a pilot
testing with the initial version by distributing it among
15 – 20 people. There are no restrictions as to how you
may choose these people (they can be your friends,
members of your family). This exercise will help in
Detecting any misprints, grammatical mistakes
Broadening your perspective with respect to the
range of expected answers to your proposed
multiple-choice questions
Amend or redesign your questionnaire (if necessary)
before it is finally approved.
5. Set up the Questionnaire in SPSS
There are two views in SPSS:
Variable view
Data view
In ‘Variable view’, you declare all your variables
(questions) whilst in ‘Data view’, you simply enter the
data from your collected questionnaires.
For example, ‘Gender’ is a variable with values ‘Male’
and ‘Female’.
6. Variable View
The first variable to be declared is ‘Questionnaire ID’.
This is an extremely important declaration, especially if
there are data entry errors.
In SPSS, we often have recourse to data sorting for
analysis purposes. Should we save and close a file after
sorting, the file will next be opened in this last state
(sorted data). In such a case, if an entry error is located,
there would be no other option but to physically verify
each collected questionnaire to be able to correct that
entry!
8. Variable View
Variable name
The name of a variable can be generic or chosen at
your own convenience.
For example, if the first question of your questionnaire is
Gender (choose between ‘Male’ and ‘Female’),
you may opt for the generic name ‘gender’ itself or ‘q1’.
However, if you choose ‘q1’, make sure that you label
this variable properly!
9. Variable View
Label
The label for a variable has to be written in the same
way as you wish it to appear (as a heading) in an output
table or chart.
If the label is omitted for a variable that has been named
‘q1’, then ‘q1’ will appear as the heading for its
corresponding output table or chart. That will obviously
not be understood by another reader.
10. Variable View
Values
These are to be assigned to variables which are
categorical in nature (variables that have options, e.g.
dichotomous, Likert scale, multiple-choice). For
example, ‘Male’ and ‘Female’ would be assigned values
1 and 2 respectively for the variable ‘gender’.
Numerical variables like rating scales, age last birthday,
etc are not assigned any values because these are
entered directly as chosen by the respondent in the
questionnaire.
11. Variable View
Values
Multiple response questions are assigned values in a
different manner for analysis purposes.
For example, to the question ‘How did you come to
know about this product?’, a respondent may choose
more than one answer to the options:
□ Radio □ TV □ Newspapers/Magazines
□ Billboard □ Internet □ Other (please specify)
12. Variable View
Values
In such a case, we have to break down this question
into 6 variables, namely Radio, TV,
Newspaper/Magazines, Billboard, Internet and Other.
These variables will be declared as dichotomous with
values 0 (if chosen) and 1 (if not chosen).
Now make sure that you don’t include too many multiple
response questions in your questionnaire because you
might end up with more than 100 variables in SPSS!!!
13. Data View
This is meant for data entry – each row represent the
entries for a collected questionnaire.
The declared variables are now column headings under
which the corresponding chosen response (by each
respondent) has to be entered as a number, whether
the variable is numerical or categorical.
Once all the collected data have been entered and
verified to be correct, we may proceed to descriptive or
inferential analysis.
14. Descriptive Statistics
For each variable, there should be comments
accompanying its chart and frequency table.
Pie charts are usually very explicit – remember they
only display percentages! Bar charts display
frequencies – comments must be made in terms of
skewness, that is, where its peak lies (middle, to the left
or to the right).
Middle: symmetrical distribution
To the left: positively skewed distribution
To the right: negatively skewed distribution
15. Inferential Statistics
This part of the analysis is directly related with the
objectives of the research.
Here, we perform all kinds of testing (mostly, the testing
of research hypotheses in order to achieve the research
objectives).
It is important that we include both categorical and
numerical variables in our questionnaire so as to be
able to use the various tests as prescribed by Curtin
(see the Marketing Research 200 unit outline)
16. Inferential Statistics
Tests to be used:
1 Chi-Squared test of Independence
2 Independent Samples T-test
3 ANOVA (ANova Of VAriance)
4 Correlation
5 Factor Analysis
6 Multiple Regression Analysis
17. Hot Tip
Irrespective of whether you used or did not use
a specific test in your research project, you may
be examined on all of them!
18. Inferential Statistics
Tests for categorical variables
1 Chi-Squared test of Independence
2 Correlation
3 Factor Analysis
4 Multiple Regression Analysis
19. Hot Tip
For multiple regression analysis, the
dependent variable has to be numerical.