Analysing the results of a simple questionnaire using SPSS

1
Analysing the results of a simple questionnaire
using SPSS
Dr. Phil RoweReader in Pharmaceutical
Computing Liverpool School of Pharmacy
2
Review this presentation at
http//www.staff.livjm.ac.uk/phaprowe
3
Questionnaire -data collected
Questionnaire number Gender Branch of profession
(Community/Hospital) Q1 Q2 Q3 Q4
Four questions designed to test knowledge in a
certain area - record as correct/incorrect
4
Numbering questionnaires
Number the questionnaires sequentially as
returned. If there is a query about a reply, you
can easily find the relevant questionnaire
5
Package for questionnaire analysis -
SPSS(Statistical Package for the Social Sciences)

• Data recording and analysis package. Similar to
  Minitab
• Data sheet analysis tools
• Oriented to questionnaire analysis
• Why bother?
• More powerful than Minitab
• Useful extra skill for C.V.

6
Types of data used in SPSS

• Numeric - A number - Specify how many decimal
  places
• String - A piece of text - Specify how many
  characters

7
Variables to be used to store the data
Questionnaire number Numeric - no
decimals Gender String - 1 character
(m/f) Branch of profession (Com/Hosp) String - 1
character (c/h) Q1 Numeric - no decimals
(0/1)
0 Incorrect 1
Correct Q2 Numeric - no decimals
(0/1) Q3 Numeric - no decimals
(0/1) Q4 Numeric - no decimals
(0/1) Score Numeric - no decimals (0-4)
8
Summary values
A set of questions with a similar purpose. Often
better to create a single summary value rather
than simply looking at each question. If find an
effect can always explore the individual
questions afterwards.
9
Stage 1Get the data into SPSS
10
Launching SPSS
JMU Applications/Analysis Tools/SPSS v10
11
Initial appearance
A "helpful" initial menu is presented. Pain in
the neck Press Cancel
12
Initial appearance
Each row represents one completed
questionnaireEach column represents one variable
See data stored - (None currently)
See details of each variable
13
Creating a numeric variable with no decimal places
This view allows us to create all the variables
that we need to hold the data. First variable to
create is for the questionnaire number.
(qnr_no) We will use a numeric variable with no
decimal places
Have switched to Variable View
14
Creating a numeric variable with no decimal places
Type in title and press ltReturngt Other details
appear by default. 'Numeric' is OK, but we do
not need the 2 decimal places.
15
Creating a numeric variable with no decimal places
Click in cell
Use down arrow to reduce to zero and press
ltReturngt
16
Creating a numeric variable with no decimal places
First column (variable) now set up to store the
questionnaire numbers.
Have switched back to Data View
17
Creating a string variable with just one character
Enter name as "gender", press ltretgt Appears as
Numeric. Click here to allow a change to be made
18
Creating a string variable with just one character
Change to String with 1 character
19
Creating a string variable with just one character
Two variables now available
Back in data view
20
All variables created
21
Enter the data
22
Calculating a variable (1)
Go to Transform/Compute
23
Calculating a variable (2)
Variable to be calculated
Formula for variable
24
Calculating a variable (3)
Confirm OK
25
Calculating a variable (4)
Scores are calculated
26
Stage 2Producing simple frequencies
27
Frequencies (1)
Go to Analyse/Descriptives/Frequencies
28
Frequencies (2)
Variable to be selected
Click on first variable to be selected Click on
the right-pointing arrow Variable moves over to
right-hand box Click on next variable and
transfer it, etc
Button to select variable
29
Frequencies (3)
All relevant variables transferred across
30
Removing a variable

• If a variable is incorrectly transferred into the
  right-hand box it can be removed
• Click on the variable in the right-hand box to
  highlight it
• The right pointing arrow will become a left
  pointing arrow
• Click on left pointing arrow
• Variable will be transferred back to the
  left-hand box.

31
Appearance of the "Output" window
32
Output from Frequencies procedure (1)
All Missing values are 0. Confirms that that a
value has been entered for all variables for all
20 subjects
33
Output from Frequencies procedure (2)
Numbers of Males and Females
34
Output from Frequencies procedure (3)
Numbers of Community Hospital
35
Output from Frequencies procedure (4)
Numbers who got Q1 wrong (0) or right
(1) Similar tables are produced for Q2, 3 4.
36
Output from Frequencies procedure (5)
Numbers of individuals with overall scores of 0,
1, 2, 3 or 4
37
Switching back to the data sheet
Window - Select Data Editor
38
Stage 3Hypothesis testing

• Is the balance of male and female the same in
  community and hospital?
• Is the level of knowledge the same in community
  and hospital?

39
Is the balance of male and female the same in
community and hospital?(Using Cross-tabulation
plus Chi-square test)

• Use menus Analyze / Descriptive Statistics /
  Crosstabs
• Highlight Gender and transfer it to the Rows
  box
• Highlight Wrk and transfer it to the Columns
  box
• Click the Statistics button
• Select Chi-square

40
Data is OK. No missing values
41
6 females in community etc
P value from chi-square test. Non-significant
Yates corrected
42
Conclusion
There is no evidence of any difference between
community and hospital employment in the
proportions of males and females.
43
Is the level of knowledge the same in community
and hospital?(Using a t-test)

• Use menus Analyze / Compare Means /
  Independent-Samples T Test
• Need to tell SPSS which column contains the data
  to be analysed (Test Variable) and which
  contains the codes for the two groups that are
  being compared ('Grouping Variable)
• Highlight score and transfer to Test
  Variable box
• Highlight wrk and transfer to Grouping
  Variable box
• Continued on next slide ...

44
Is the level of knowledge the same in community
and hospital?(Using a t-test) Continued

• Now need to tell SPSS that you want to compare
  hospital versus community. (There might be more
  than two different codes in the wrk column.)
• Click Define Groups button
• Fill in Group 1 as c
• Fill in Group 2 as h

45
Mean score is higher for Hospital than Community
P lt 0.05 Difference is significant
46
Conclusion
There is significant evidence that hospital
pharmacists have greater knowledge of this
particular area than community pharmacists. (Coul
d now look at each question to see whether the
difference in knowledge extends over all
questions or only applies to 1 or 2 questions.
Do cross-tabulation chi-square test for each
question separately.)
47
Saving the data
File/SaveAs
48
Opening a data file
If you use the 'Open file' symbol there are no
complications. If you use the Menu bar
(File/Open), you will have to select file type as
'Data'.