Introduction to SPSS

Written by: Emory Morrison

Center for Social Science Computation & Research

145 Savery Hall
University of Washington
Seattle WA 98195 U.S.A.
(206)543-8110

October 99
http://augustus.csscr.washington.edu/pdf/spss.pdf
 CSSCR SPSS 10/19/99 EM Page 1 of 14

INTRODUCTION TO SPSS
SPSS (Statistical Package for the Social Sciences) is a versatile software package which primarily
assists users in performing complex statistical analyses of quantitative data sets. The software
allows users to create, modify, and analyze data, as well as to produce graphics to display findings
in reports or presentations. SPSS is comparable to other general statistical analysis software such
as SAS, Stata, or S-Plus. Relative to these other software packages, SPSS is easier to learn and
more simple to use. However, it is more restrictive than the other statistics packages; advanced
users have a tougher time tailoring SPSS to meet their specialized analytical needs.
This document should help new users get started in SPSS. It is structured in question and answer
format, and addresses in a logical sequence the questions that a new user might have. Once you are
on your way, (or if you already have a basic knowledge of SPSS), you can use the set of manuals that
SPSS provides, the help features within SPSS, or run the SPSS on-line tutorial for more compre-
hensive guidance.
Please note, at the time of the writing of this document, CSSCR is using version 9.0 of SPSS soft-
ware. All information in this document will pertain to SPSS version 9.0.

How do I start using SPSS at CSSCR?

SPSS (version 9.0) is on all the PCs in CSSCR. The process of launching the software is the same as
for any other frequently used package in the lab. At the initial window (once Windows has been
launched) find the SPSS icon. Double click with your mouse, and voila.
This screen is the data editor window in SPSS. It is one of a seven basic windows that SPSS pro-
vides (other windows include: the output viewer, syntax editor, chart editor, pivot table edi-
tor, text output editor, and script editor). The data editor window is the most fundamental.
With the exception of the more advanced syntactical procedures, you must enter data in the data
editor window before SPSS can perform any operations.

Figure 1

CSSCR SPSS 10/19/99 EM Page 1 of 15 http://augustus.csscr.washington.edu/pdf/spss.pdf

 CSSCR SPSS 10/19/99 EM Page 2 of 14

What Are Data?

SPSS is very restrictive in what it recognizes as data. For SPSS, data are members drawn from a
similar set (cases), that are each described by the same set of features (variables). Cases are listed
in rows, and variables are listed along columns. The block where a row and column intersect is
known as a cell. Each cell contains the value on the variable (the column) for the case (the row).
Thus, data must take the form of a spreadsheet. An illustration of data in the SPSS data editor
window is presented below.

Figure 2

In Figure 2, the cases are countries of the world (thirteen of which appear in the figure) and the
variables are characteristics of each of the countries (i.e. population, population density, percent of
population residing in urban regions, etc.). Since SPSS only recognizes a variable title eight charac-
ters long, many of the variables have shortened names. For example, the column that contains
information on population is titled populatn.
The value for the case Australia for the variable ÒpopulatnÓ is 17800 (which, since the units are in
thousands, indicates that Australia has a population of 17.8 million people). The value for the case
Afghanistan for the variable religion is Muslim. Values can either be numeric or in character (also
known as string) format.
However, if you want to use a variable in an analysis, you will need to code it in numeric format
(even if a character format makes more sense to you). So, if you want to analyze the relationship
between population size and nationsÕ principle religion, both variables will need to be coded in
numeric format. Each religion should have a unique number, so that Muslim could be 1, Christian-
ity could be 2, Buddhist could be 3, Hindu could be 4, etc.

What do I Do Once I Open SPSS?

Input the data that you want to analyze into the SPSS data editor window. SPSS can get data
through a variety of methods. First, you can manually type the data directly into the data editor
window. If you want to enter data this way, define your variables first. Start by moving your mouse
to the gray area at the top of the left most column. Your cursor should form a white arrow. Double-
click. Your screen should look like Figure 3.
 CSSCR SPSS 10/19/99 EM Page 3 of 14

Figure 3

The gray box that appears in the middle of the spreadsheet is the Define Variables dialogue box.
You should decide what to name your variable (eight characters or less) and type it into the white
rectangle next to the prompt Variable Name.
Imagine that you have the data from Figure 3 on paper and you would like to type it into the com-
puter. Type ÒCountryÓ into the rectangle provided for a variable name. Since this variable is the
case identification, and you want to define your cases using characters (rather than numbers),
youÕll need tell SPSS. Double click-on the Type button.
This will activate a new window which allows you to define the way the data editor interprets the
data that you enter. Click the button to the left of the word Òstring.Ó Next to where the prompt says
ÒCharacters,Ó type the number of spaces you will need for your longest entry (perhaps 20 will suf-
fice). Click Continue. This will return you to the Define Variable dialogue box.
Note that once you have told SPSS that you will be using string data for this variable, it automati-
cally switches your Measurement setting to Nominal. Now, click on the OK button. You will be able
to enter the data for the first column; strike the return key between entries to automatically put
each new entry in a new row.
When you have entered the data for that column, double-click on the gray area above the second
column. Type ÒPopulatnÓ in the slot provided for Variable Name. Since the values for this variable
are numeric, be sure that under ÒTypeÓ the numeric button is marked. Next click the button marked
Labels.
The top of this new dialogue box asks you for a Variable Label. In the box next to this prompt, type
Òpopulation of the country in thousands.Ó When you have a data set with a lot of variables and
information, the variable labels will help you keep track of what each variable represents. Click
Continue. Since population is a scalar variable, make sure that under the Measurement prompt,
scale is marked. Click OK. Type in column 2 and continue in the same fashion until Column 5.
Once you get to the fifth column, the one titled Religion, you will have an interesting situation. The
data clearly is in character format, but if you want to use it for analysis, it must be in numeric for-
mat. The solution to this problem is to assign each string value a numeric code, and then indicate
how the coding works in the Define Variable window.
 CSSCR SPSS 10/19/99 EM Page 4 of 14

So, after you have double-clicked on the gray heading of the fifth column and have typed ÒReligionÓ
next to the Variable Name, click on Labels. Next to the prompt Variable Label, type Principle reli-
gion of the country. Next to the prompt Value type the number 1. Hit the tab bar. You will now be
positioned to tell SPSS what the number 1 indicates for this column. Type ÒMuslimÓ and click on
the Add button. Repeat the process until you have entered the numeric code for each possible entry.
When you are done, click Continue. Next, click on the Type button to ensure that the variable is
numeric. Finally, since the measurement level is nominalist the Nominal button is activated. Click
OK. You will now be able to type in the numeric codes for the religion variable.
Using .sav files
The manual method of entering data directly into SPSS is both cumbersome and time-consuming,
especially when you want to analyze a large amount of data. A second method for introducing data
to SPSS is to find a file that has already been prepared in SPSS format, and open it. Such files have
a .sav suffix affixed to them. For example, the file in Figure 4 is titled ÒWorld95.sav.Ó It is stored on
each the PCs in CSSCR on the C drive, within the directory C:\program files\SPSS. In order to
open it, click on the File menu. A list of options will appear under the menu bar; the second option is
Open. Click on it.

Figure 4

You will see the dialogue box below.

Figure 5

The top prompt, labeled ÒLook in:Ó provides a tree directory of your computer; find the drive and
folders in which your file is stored. Next to the prompt File name type in the name. Next to Files of
type check that the SPSS(*.sav) option is up. Click on the Open button. You will now be ready for
analysis.
Aside from the above two procedures, other methods are possible to input data. One may take data
that has been stored using other software packages (like excel or access). One may get data that has
been stored in ASCII format. One can enter data though syntax. This document will not detail how
to get data through these methods.
 CSSCR SPSS 10/19/99 EM Page 5 of 14

Okay, IÕve Got My Data, What Will SPSS Do With It?

Once you have your data entered, you can use SPSS to manipulate and analyze your data set. SPSS
will allow you to add new variables to your data set, to pare down cases, to merge files together, etc.
SPSS will also allow you to generate summary statistics for your variables, to run statistical tests
like correlations on your variables, it will also allow you to run many types of regression analyses,
as well as to run variants of the standard ANOVA models, and it will let you run all sorts of
advanced statistical models and tests. SPSS will also let you create graphical representations of
your data set (such as pie charts, bar graphs, scatter plots, etc.)
The new user will use items off the menu bar to direct SPSS to perform tasks. More experienced
users may use syntax to direct SPSS. Since this document is directed to the new user, it will only
address items that are on the menu bar. The menu bar contains ten headings: File, Edit, View,
Data, Transform, Analyze, Graph, Utilities, Windows, and Help. Clicking on any of the ten head-
ings will provide the user with a number of options. Most of the options under the File, Edit, Win-
dows, and Help menus are not unique to SPSS; therefore, this document will address options under
the other headings.
Click on View at the top of the menu bar. Figure 6 shows the View options.

Figure 6

The only option worth noting here is the bottom one, Value Labels. Click on it to either activate it
(check it), or deactivate it (uncheck it). When it is checked, the spread sheet will display in charac-
ter form the values for variables. Thus, while it is checked, the column Religion displays the label
Muslim. When it is unchecked, the spread sheet will display only the numeric value of the variable
(so under religion the first entry would be 1).
Click on the Data pull-down menu in the menu bar. Figure 7 shows the commands under the Data
menu.

Figure 7

These options allow you to do things such as move within your data set (Go to Case), edit your
data set (Insert Variable, Insert Case, Define Variable), expand your data set by joining two
data sets together (Merge Files), contract your data set by selecting a subset of cases (Select
Cases), Group your cases into categories for separate analyses (Split File), as well as perform
other functions. For details on how these functions work, consult an SPSS manual.
 CSSCR SPSS 10/19/99 EM Page 6 of 14
Now click on the Transform pull-down menu. Here are the commands under the Transform menu.

Figure 8

These options help you to create new variables, and to assign values to the new variables based on
specific procedures. Two options that we will look more closely at are Compute and Recode, both
of which allow you to use existing variables to create new ones. First, we shall examine the Com-
pute function. Imagine that we wish to create a new variable which records the life expectancy dif-
ferences between males and females in each country in the data set. The life expectancy for males
already exists, lifeexpm and the life expectancy for females already exists, lifeexpf. In order to cre-
ate our new variable (complete with values for all the cases) we need to click on the compute button.
A new dialogue box will open up, which is displayed in Figure 9.

Figure 9

You need to pick a name for the new variable, and type it into the box labeled Target Variable. Next
under the box Numeric Expression you need to type in a formula which defines how the new vari-
able will be calculated. In the current case, the new variable, (the difference in life expectancy
between men and women), will be named Òlifediff.Ó It will be ÒlifeexpmÓ minus Òlifeexpf.Ó Once this
information is typed into the appropriate boxes, click the OK button. SPSS will have created the
new variable and will have assigned values to it for all cases which had valid scores for both life-
expm and lifeexpf.

Figure 10

Notice in Figure 10 how SPSS formed a new column, lifediff. Note also how the values in
lifediff are, in fact, the difference between the values in the columns lifeexpm and lifeexpf.
 CSSCR SPSS 10/19/99 EM Page 7 of 14

Recoding
In addition to calculating new variables, people frequently use SPSS to transform variables by cat-
egorizing them using the Recode option. From the Transform pull-down menu, choose Recode;
click the Into Different Variable option. A Recode dialogue box will appear.

Figure 11

In the large white box in the middle, type in the name of the existing variable that you would like to
recode. For example, imagine that you would like to recode the variable which reports the rate of
aids cases that a country reports (per 100,000 population), aids_rt. This variable ranges from 0 to
327. You would like to identify four categories of countries based on the extent of the aids problem
within the country. Some countries (perhaps where the aids rate is less than 1 per 100,000) you
would like to label as controlled problem; other countries (perhaps where the aids rate is more than
1, but less than 10 per 100,000) you would like to label as problematic; still other countries (perhaps
where the aids rate is between 10 and 100 per 100,000) you would like to label as in crisis; finally
you would like to identify the countries (perhaps those with an aids rate greater than 100 per
100,000) as Òsevere crisis.Ó
In order to perform this task, type Òaids_rtÓ in the large white box. Type the name of your new vari-
able in the Output Variable Name box. Once you type in the name of a new variable (and perhaps a
label for it too), click on the Change button. Next click on Old and New Values... to tell SPSS how to
change the values from one variable to another. After this step, youÕll see a new dialogue box (Fig-
ure 12).
You will need to spend several minutes working with this new dialogue
CSSCR box.
SPSSFirst,
10/19/99 EM Page 8 of 14

Figure 12

you need to direct SPSS to recode the first set of values. Click the Lowest through
Range button. Type the number 1 in the box. Next, move to the New Value box; type a 1. Click on
the Add button.
You will see Lowest thru 1 --> 1 appear in the box under Old --> New. Next, click the button next to
the uppermost Range. Type 1 into the left rectangle and 10 into the right rectangle. Go back to New
Value and type 2; click on the Add button.
Move back left again, click on the same Range button. In the left rectangle type 10; type 100 in the
right one. Type a 3 into the New Value box. Click on the Add button.
Now, click on the through highest Range button. Type a 100 into the box. Under New Value type a 4.
Click the Add button.
Click the Continue button; you will be returned to the first dialogue box. Click the OK button. You
will have created the new variable that you wanted.
Now let us look at what types of analyses that we can do with SPSS. Figure 13 shows the options
under the Analyze menu. Most of these analyses require that you have some knowledge of statistics
so that you can interpret your results. However, you can find some easy-to-interpret information
under the option Descriptive Statistics.
Imagine that you need to know about the distribution of values for the variable ÒfertilityÓ (which
reports the average number of births in a particular country). From the Analyze menu, go to
Descriptive Statistics/Descriptives.

Figure 13
 CSSCR SPSS 10/19/99 EM Page 9 of 14

Figure 14

Drag the variable that you want information on from the left box (which lists all the variables) to
the right box. In order to do this task: scroll down the list of variables on the left side; find the one
you would like to analyze; double-click on it; voila -- it is done! Click OK. SPSS will show you the
information in an output file. In this case the output will look like Figure 15.

Figure 15

Click on these
to move
between the
output win-
dow and the
data editor
window.

The window in which SPSS posts the results is separate from the data editor window. The output
file window is also discrete from the data file. Therefore, if you want to print your output, work from
your output window -- not your data editor window. Likewise, if you want to save your output, be
sure that you are working from the correct window. SPSS assigns output files the suffix Ò.spoÓ and
will only recognize files with this suffix as output files. Toggle between the open Program buttons
on the task bar.

The output window displays the answers to the statistical questions you asked SPSS. In the case of
Figure 15, SPSS had been asked to provide descriptive statistics for the distribution of scores for
the fertility variable. The table in the main section of the output reports that 107 countries have
valid (non-missing) scores for fertility; in the country with the rate is 1.3 children; in the country
with the highest fertility the rate is 8.2 children. All countries measured average a 3.56 fertility
rate with a standard deviation of 1.9.
To see the full distribution of scores for this variable, go to Analyze/Descriptive Statistics/Frequen-
cies. The output that SPSS will generate is displayed in Figure 16.
 CSSCR SPSS 10/19/99 EM Page 10 of 14

Toggle
between
these to
see the Figure 16
different
output
files.

The table in
Figure 16 displays five columns of information. The first column displays in ascending order the list
of values that appear in the data set for fertility. Since this table has too many values to be dis-
played on one screen, you must scroll down the window to see more of the table. The second column
shows how many cases in the data set resulted in the value to the column to the left. For example,
one nation had an average of 1.3 kids per adult woman; two nations had on average 1.4 kids per
woman. The third column lets you know what percent of the data set had the value in the first col-
umn. The fourth column calculates the same number but only considers the cases that do not have
missing data. Finally, the fifth column reports the percentage of cases that have at least the value
in the first column.
Note also, in the output window that is Figure 16, the left side has information that is distinct from
the table. This information is the navigator window. With this window, you can see what findings
are displayed in your output. The output file in Figure 16 contains two tables, one for ÒDescriptivesÓ
and one for ÒFrequencies.Ó To see the output for the descriptives that were run earlier, double-click
on the icon next to the word ÒDescriptivesÓ in the navigator window.
Now, letÕs do some real analysis. LetÕs see how SPSS manages a simple ordinary least squares (OLS)
regression. WeÕre going to look at regressing fertility on womenÕs literacy rates, on the log of the
gross domestic product, and on the death rate of the nation. To perform this operation, go to Ana-
lyze/Regression/Linear. You will see the dialog box illustrated in Figure 17.
 CSSCR SPSS 10/19/99 EM Page 11 of 14

Figure 17

The left side of this dialogue box is a list of the variables contained in the data set. We must select
the variables we want to include in the analysis from this list. First, we pick the ÒDependentÓ vari-
able, which in this case is Òfertility.Ó Find it in the list of variables and click on it. Now click on the
little arrow beside the white rectangle under the heading Dependent. Observe how the variable
ÒfertilityÓ moves into the Dependent list. Next click on the variable Òlit_fema;Ó click on the arrow
next to the Independent(s) list. Notice that youÕve just moved this variable into the Independent(s)
list. Repeat this routine with the variables Òlog_gdpÓ and Òdeath_rt.Ó SPSS can forthwith run the
regression. Click on OK. The results are displayed in Figure 18.
For OLS regression, SPSS generates four tables of results -- as the navigator portion of the output
window shows. In fact, SPSS generates too much information to display in one screen. Figure 18
displays three of the four tables of information. You should have a basic knowledge of statistics to
interpret the information displayed in these tables.
The first table in Figure 18 (the second table that SPSS generates as part of this output) shows that
the ÒR-squaredÓ statistic indicates that the three independent variables explain 72 percent of the
variance in fertility among nations. The final table suggests that two of the three variables in this
model are significant predictors of the average fertility rate (at the 95 percent confidence level).
By comparing the ÒBetaÓ statistics to each other (also known as the standardized regression coeffi-
cient), we can tell that the most powerful predictor of fertility among the three is womenÕs illiteracy.
According to the ÒBÓ statistic associated with Òfemales who read,Ó for every one percent drop in
womenÕs literacy, we can anticipate a .04 increase in the fertility rate. This OLS regression analysis
is only one of many that SPSS can perform. For descriptions of how to perform or interpret findings
from other analyses, consult the SPSS manuals.

Figure 18
 CSSCR SPSS 10/19/99 EM Page 12 of 14
Graphics
SPSS also lets you create (and edit) graphics that allow you to display findings in ways that are
visually more accessible than tabular format. The ÒGraphsÓ heading on the menu bar contains the
options for creating these images. The options under Graphs are displayed in Figure 19.

Figure 19

We can display the distribution of the data for the variable Òfertility,Ó in the same manner as our
frequency analysis. Instead of generating a frequency table we will make a histogram (which is eas-
ier to look at). Go to Graphs/Histogram. The Histogram dialogue box should look like Figure 20.

Figure 20

Move the variable ÒfertilityÓ from the list of variables on the left side to the white rectangle under
the heading Variable. Click OK. Figure 21 displays the histogram.

Figure 21
 CSSCR SPSS 10/19/99 EM Page 13 of 14
Notice that the histogram is displayed in the output window. You can edit this histogram if you
wish. Double-click on it, and you will open up a chart editor window with your histogram inside.
From the chart editor window, you can do things like add text to the figure, change the scale of the
axis, add lines or other graphic features, or even ÒfudgeÓ the findings by changing bar heights. For
information about how to use the chart editor window, consult the SPSS manuals.
You can also create graphics which use more than one variable. For example, you can make a scat-
ter plot to show the relationship between female literacy and fertility (similar to the OLS analysis
which we performed earlier). To do this, go to Graphs/Scatter/Simple/Define. You will see a dialogue
box like Figure 22.

Figure 22

Place the ÒfertilityÓ variable in the box under the heading ÒY-axisÓ; place the Òlit_femaÓ variable in
the box under the heading ÒX-axis.Ó Click OK. Figure 23 displays the results.
 CSSCR SPSS 10/19/99 EM Page 14 of 14

Figure 23

Figure 23 shows a nice linear relationship between the two variables: the more women read, the
less children they have. As with the histogram, double-click on the graph to open up a chart editor
window to manipulate it. You can easily add the best fit line, or display the regression equation.

Conclusion
If you have a basic knowledge of statistics, SPSS is among the easiest statistical analysis packages
to learn and use. Learn it by experimenting with its features, using the on-line tutorial that is
built into the Help menu, or by consulting the manuals that come with the package. I have three
pieces of advice to keep in mind when youÕre using SPSS in CSSCRÕs labs.
1) If you are manipulating data files, save your original under one name, and save your updates
under a new name in the ÒC:\tempÓ directory of your computer. Keep your original file intact
in case you make mistakes in the changes you have made to your data set. You will also want
to save frequently, because if you give SPSS a very demanding job that requires a lot of mem-
ory, you are in danger of crashing your machine and losing your work.
2) When you save or print your file, be very aware of which window you have open. For example,
if you want to print your findings, you must have the output window open. If you have your
data editor window open and hit the print button you will not print your output, but rather
your data file, which could potentially be hundreds (or even thousands) of pages of nothing
that you were interested in seeing on paper, tying up the printers, killing trees, costing you
money.
3) If you have questions about how to perform a specific task, ask the CSSCR consultants. They
should be able to help you.
Good luck with SPSS!

emory morrisson, Sept 99