SPSS Lesson 4

Creating Data Summaries

For Your Research Report
The user-friendliness of SPSS makes it easy to compute a million statistics and/or
generate a gazillion charts and tables. Further, the windows compatibility of SPSS means
that any charts, tables and/or figures you generate can easily be cut and pasted directly
from SPSS into Microsoft Word and other word processing application. But will either
your professor or TA be impressed if you do so and then dump them en masse into your
research report? Not.
As is the case with every other step of the process, you need to know the proper way to
represent your data, and understand that the responsibility is yours to present information
that is meaningful, on-topic, non-redundant, and to do so in as succinct and clear a
manner as possible. Given that objective, you will find there are some ways in which
SPSS is very helpful in allowing you to just run procedures and cut and paste the
elements directly into your report. Unfortunately, there are at least as many other types of
output that contain the information you want along with a bunch of other clutter that you
dont want, so you end up having to extract the information you need for a table (or
whatever) that you must construct yourself.
In this lesson well go through the various statistical procedures we worked with in
Lesson 3 and focus on the way the information you gain through those procedures might
appear in your research report.

FREQUENCIES
The FREQUENCIES procedure in SPSS creates frequency distributions and will
compute descriptive summary statistics (e.g., mean, median, mode) for each of your
variables. There are basically two different kinds of information that will be useful for
you via this procedure.
First is the situation where you have a bunch of variables you want to summarize. For
example, you might have given 7 different Likert-type (agree-disagree) items in your
survey, as I did in the two versions of the Crim 320 QuickSurvey data base, and you
may want to show summary information for all 7 in your report. In the SPSS printout,
just one of the seven items produced the following information:

Statistics
Traditional values undermined
N
Valid
Missing
Mean
Median
Mode
Std. Deviation

46
47
3.07
3.00
4
1.18

Traditional values undermined

Valid

Disagree Strongly
Disagree Somewhat
Neutral Don't Know
Agree Somewhat
Agree Strongly
Total
9

Missing
Total

Frequency
7
6
13
17
3
46
47
93

Percent
7.5
6.5
14.0
18.3
3.2
49.5
50.5
100.0

Valid Percent
15.2
13.0
28.3
37.0
6.5
100.0

Cumulative
Percent
15.2
28.3
56.5
93.5
100.0

Traditional values undermined

20

Frequency

10

0
Disagree Strongly

Neutral Don't Know

Disagree Somewhat

Traditional values undermined

Agree Strongly

Agree Somewhat

Wow. A whole page of information for just one item. And just think we have 7 items!
Will we be impressed if you give us 7 pages of charts and tables including all of that
information? Not at all; your task is to take it all and figure out a way to put it on one
page, preferably in one table. Here is one way to do so:

Table 1
Overall Distributions and Summary Statistics for
Internationalist and Protectionist Survey Items

Item

Internationalist Items
Canada's military should
continue primarily in a
"peacekeeping" rather
than "military" role.
Canada should do
everything it can to foster
its image as an open
country that values human
rights.
Canadians, in general, are
a tolerant people who
value cultural diversity.
Protectionist Items
It seems that many
traditional Canadian
values are being
undermined these days.
The unemployment level
in Canada is far too high..
There is far too much
many spent these days on
social welfare services.
Criterion Item
Current immigration
levels are too high; fewer
immigrants should be let
in the country each year.
x

Distribution of Responsesx
1
2
3
4
DSo N/DK ASo
DSt
N
N
N
N
(%)
(%)
(%)
(%)

5
ASt
N
(%)

Mean

47

4.11

1
(2.1)

3
(6.4)

6
(12.8)

17
(36.2)

20
(42.6)

47

4.60

0
(0.0)

2
(4.3)

1
(2.1)

11
(23.4)

33
(70.2)

48

3.73

0
(0.0)

7
(14.6)

5
(10.4)

30
(62.5)

6
(12.5)

46

3.07

7
(15.2)

6
(13.0)

13
(28.3)

17
37.0)

3
(6.5)

45

3.80

0
(0.0)

9
(20.0)

3
(6.7)

21
46.7)

12
(26.7)

45

2.87

6
(13.3)

13
(28.9)

11
(24.4)

11
(24.4)

4
(8.9)

92

2.78

18
(19.6)

23
(25.0)

24
(26.1)

15
(16.3)

12
(13.0)

Dst = Disagree Strongly; Dso = Disagree somewhat; N/DK = Neutral or Dont Know; Aso = Agree
Somewhat; Ast = Agree Strongly

Note that what started off as a tonne of information was boiled down to less than one
page, largely by deciding what was most important to me and guided by a desire to place
all the information in one place so that the reader could easily look at the distributions,
compare means, etc. The resulting table was given a number, a name that reflected what
was being summarized inside it, and expressed the information far more accessibly than
page after page of printout would have done.
Of course not everything has to go in a table. There are times where you may want to
show the distribution of a single key variable, for example, which you might do in a
table, or could do in a chart or figure. Charts and figures, I should note, are quite
amenable to straight cut and pasting if you take the time to set them up well. This starts
with doing a good job of defining your variables (e.g., with informative variable and
value labels), but needs to be supplemented by some planning at the point where you
want to generate the chart.
For example, lets say that you want to show people the distribution of scores on the item
that dealt with criminology students career preferences. To do it as a bar graph, you
would follow the following procedures:
1. Open your SPSS data file and click graphs on the menu at the top of the page.
2. A dropdown menu appears; click bar on that drop down menu in order to create
a bar graph.
3. A bar chart dialogue box appears. Decide at left what sort of graph you have
for a single variable like this we would simply check simple. At right, click
define in order to tell SPSS what you want a graph of.
4. A define simple bar dialogue box appears. Fill in the information the box
requires:
a. At left, click whatever variable you want to graph, and then click the
arrow that will take it to the category axis box.
b. At the top, click what you want the bars to represent I want it to show
the percentage of students falling in each group, so I click % of cases.
5. Note that there are also places in that dialogue box for you to do a title for your
graph. I do so by clicking the title button. On Line 1 I place a Figure number
Figure 1 because graphs are known as figures. On Line 2, I place a title:
Career aspirations of Crim 320 Students to make clear what is being
represented. Click continue to get back to the Define simple bar
dialogue box.
6. Click OK
Upon following these procedures, I get the nicely labelled graph below, which can be cut
and paste directly into your word processing programme (MSWord in my case) as
follows:
1. Place the cursor on top of the graph and right-click so that a dialogue box opens;
2. Click on copy objects
3. Go to your document and place the cursor in the location you want the graph
inserted and either right-click on that spot and then click paste or;

4. Simply click paste.

If it looks a little too small or too big, click on the graph, at which point you will see the
outline of the figure, including some black squares at the corners. To make the graph
bigger (or smaller), just drag on a corner and watch it grow to whatever size you want.

Figure 1
Career Aspirations of Crim 320 Students
30

20

Percent

10

ot
N
su
re
o
ol
in

t
en

gy
lo

rim
C

o
in

t
ye

in

rim
C

y
rit

er
ch

e
ol

r
ea
es

r
Pa
or

op
el
ev
D

ng
hi
ot
N

er
th
O

y
lic
Po

ye
w
La

ic

u
ec
/S

n
io
at

em

g
in
lic
Po

ob
Pr

ad
Ac

Current career preference

Note that you would follow essentially the same procedures to do a pie chart, except that
you pick pie instead of bar when following the procedures outlined above. When you
open the pie dialogue box, you will see that it looks pretty much identical to the bar
dialogue box, including giving you the possibility of doing a title.
One word to the wise: Pie charts can be pretty spiffy as long as you do not have too many
categories and/or have too many small slivers.
The career aspirations pie chart is pushing it a bit in terms of the number of categories,
but still looks pretty good:

Figure 2
Career Aspirations of Crim 320 Students
Academic Researcher
Not sure yet

Probation or Parole

Policing / Security
Nothing in Criminolo

Other Criminology
Policy Development

Lawyer

CROSS-TABULATIONS AND CHI-SQUARE

Cross-tabulations allow you to look at how two variables interact while the chi-square
statistic is the measure of whether the association between the two variables represented
is statistically significant (p<.05), or within the realm of what youd expect on the basis
of chance variation alone (p>.05, or simply p=ns).
The procedures for running a chi-square were discussed in lesson 3, and what you should
have seen at that time is that the cross-tabulation procedure allows you considerable
flexibility in how you set up the table in the first place. For example, you can determine
which is the row and which is the column variable, you can have the categories in
descending or descending order, and you can control which of many elements appears
within each cell. Because of this, you can get a cross-tabs table pretty close to what you
would want it to look like in your report. The one unfortunate aspect is that SPSS does
not allow you to place a title on it like it does for the charts/graphs it does a default
version that is a bit too cryptic, so that you would surely have to redo it if you were
submitting it for a journal, but its close enough that I would allow you to include it, with
a couple of small adaptations, in your research report.

Lets look again at the crosstabs we generated involving gender and involvement in teambased sports. Youve seen the table and chi-square statistic in Lesson 3, heres what I
would make it look like for a research report:

Table 3
Cross-Tabulation of Gender of Respondent and Whether They have
Engaged Recently in a Team-Based Sports Activity 1
Gender of Respondent * Played team-based sport Crosstabulation

Gender of Respondent

Men

Women

Total

Count
% within Gender
of Respondent
Count
% within Gender
of Respondent
Count
% within Gender
of Respondent

Played team-based
sport
No
Yes
19
15

Total
34

55.9%

44.1%

100.0%

49

10

59

83.1%

16.9%

100.0%

68

25

93

73.1%

26.9%

100.0%

Chi-square = 8.10; df=1; p < .05

Note the two things Ive done: (1) assigned a table number and title; and (2) placed the
appropriate statistic the chi-square value, degrees of freedom, and an indication of
whether the statistic is significant or not. Note by the way, that I did not insert the exact
probability level that you can find on the printout (which is .004; see notes for Lesson 3).
Some people have the erroneous idea that the p value is a measure of how significant a
statistic is, but that is not the case. Statistical significance is like a goal in hockey or
soccer. Any shot is either a goal or not, but it is not any more or less of a goal depending
on where in the net it hits. Similarly, you establish a significance level -- .05 normally
and then determine whether the statistic you observe is significant or not, at that level,
period.

THE T-TEST
T-tests are used to look at the difference between two groups, and normally these are just
reported in the text of your report, and not presented in a table, because there is really not
all that much to present. Graphs of group means will sometimes appear in reports (recall,
for example, the graph of mean for the groups in the Donnerstein and Berkowitz sex and
aggression study), but not often, and not likely in your projects.
But lets take a couple of examples and work them through. An easy example is the one
that was done in SPSS Lesson 3, in which we compared the means scores on the

 immigration item for the groups of people who received the two different versions of
the questionnaire. You would reporting something like that in your research papers this
way:
The group who had been randomly assigned to the protectionist group had a
mean rating of 2.83 on the 5-point scale, while the group who received the
internationalist version had a mean rating of 2.74. Although these differences
were in the predicted direction the protectionist group indicated a higher
level of agreement than the internationalist group to an item that stated that
immigration levels were too high a t-test showed that the differences were no
greater than one would expect on the basis of chance variation alone (t = 0.34; df
= 90; p=ns).

On some occasions, however, particularly where you are comparing two groups on
multiple identical or similar dependent variables, it may well be more expedient to place
the information in a table. For example, one of the survey items asked participants to rate
a number of different attributes in terms of how important they considered each to be in
the selection of a mate.

Using the scale at right, indicate how

1
2
essential it is to you that your mate/partner Not essential
have each of the following attributes.
Well-educated:
1
2
Physically attractive:
1
2
Financially secure:
1
2
Good sense of humour:
1
2
Strong sex drive:
1
2
Intelligent:
1
2

6
7
Essential

3
3
3
3
3
3

4
4
4
4
4
4

5
5
5
5
5
5

6
6
6
6
6
6

7
7
7
7
7
7

This survey item was included because of a study reported in a book I had written (called
Research Decisions) that included a study showing that men and women valued different
attributes for a mate, with men emphasizing physical appearance while women
emphasized education and income. Rather than reporting each and every comparison, you
could do a whole series of t-tests (this is not the best statistical test in such a situation, but
we can live with this option for this course), and then summarize the results in a table.
The fact that all the statistics are in the table means that you do not have to repeat them
again in the narrative part of your paper, but can just talk about them on the assumption
that your reader can see the table, and all you need to do is draw their attention to the
highlights. Your write-up might look something like this:
As Table 4 reveals, there were numerous differences between the men and
women in the degree of importance they attached to the different attributes, but

only two of these were statistically significant: the importance attached to

physical attractivess and striong sex drive. In both cases, as predicted, the
ratings by men were significantly higher (at p<.05) than those by females on
those attributes.

Table 4
Comparison of Mens and Womens Ratings Regarding the Desirability
of Six Different Attributes in a Mate
Attribute
Physically Attractive
Strong Sex Drive
Financially Secure
Well-Educated
Sense of Humour
Intelligent
2

Mean Rating2
Men
Women
5.76
4.75
5.94
4.86
4.50
4.88
5.50
5.12
6.24
6.34
6.15
6.03

t
3.74
2.28
1.18
1.13
0.49
0.43

Comparison
df
p
91
<.05
42
<.05
91
ns
91
ns
91
ns
91
ns

The scale ran from 1 (Not essential) to 7 (Essential)

What you should see from this example is that the time you want to use a table (or figure)
is when they can help you summarize a lot of information in one place that, if you had to
report it narratively in the body of your report, would be about as exciting to read as a
laundry list. And once again, note that Table 4 does not come directly from SPSS I had
to put it together on the basis of information extracted from six different analyses that
originally consumed several pages of printout, but was able to boil it down into one small
and easy to read table.

Final Admonitions
There are several themes that go through this lesson regarding the preparation of tables,
charts and graphs:

There are times when SPSS produces great output that can easily be cut and pasted
into your research report (most notably figures/graphs), but many other times where
you have to extract the information from multiple parts of SPSS printouts.
The time to create a table or figure is when using one helps you represent a lot of
information in one economical table or picture. The name of the game is data
summary and reduction.
Every table and figure will have a number (tables and figures are numbered
separately) and a title; the title will make clear what information is in the table.