Outline of how the CROSSTABS command works.

Cross tabulation

Let's begin by understanding a cross-tabulation of variables, a basic analytical procedure in social research generated by the Crosstabs program. Basically Crosstabs produces tables of data that classify the values of a variable in rows by the values of another variable in columns. (Rows are read horizontally from left to right; columns are read vertically, from top to bottom.)

A common name for a table that classifies one variable against another is a cross tabulation. The entries in the simplest cross-tabulation are merely counts of the cases that fall in each cell of the table (a cell is the "box"created by the intersection of a row with a column).

Crosstabulation can best be explained through an example from the Crosstabs program itself:
 
DISCUSS  v03 Do you DISCUSS POLITICS often?  by  NATION  v01 NATION
                    NATION                                     
            Count  | 
           Col Pct |United   Great    West     France   East 
                   |States   Britain  Germany           Germany    Row
                   |     1  |     2  |     3  |     4  |     5  | Total
DISCUSS    --------+--------+--------+--------+--------+--------+ 
                1  |   271  |   216  |   510  |   117  |   671  |  1785
  Frequently       |  14.8  |  14.5  |  24.6  |  11.8  |  50.8  |  23.2
                   +--------+--------+--------+--------+--------+
                2  |  1054  |   766  |  1229  |   532  |   571  |  4152
  Occasionally     |  57.4  |  51.6  |  59.3  |  53.6  |  43.2  |  53.9
                   +--------+--------+--------+--------+--------+ 
                3  |   511  |   503  |   333  |   343  |    80  |  1769
  Never            |  27.8  |  33.9  |  16.1  |  34.6  |   6.1  |  23.0
                   +--------+--------+--------+--------+--------+ 
            Column    1835     1485     2073      992     1322     7706 
             Total    23.8     19.3     26.9     12.9     17.2    100.0
This table cross-tabulates two variables: DISCUSS (how often the respondent discusses politics) by NATION (the respondent's country).
 
The two rows at the bottom of the table report the number of respondents in each NATION (1,835 for the U.S.) and the percentage (23.8%) of the total number of respondents (7,706) who gave valid responses.
  • The column of numbers on the right hand side reports the same information for DISCUSS.
  • The entries in each cell report the number of cases and the % that number is of the column total.
  • Thus, 271 persons in the U.S. discussed politics "frequently," and that is 14.8% of the total U.S. respondents (1,835).

Armed with this knowledge of how to read the Crosstabs output, you should be able to explore some substantive analyses.

The CROSSTABS command in SPSS

Under the syntax rules of the SPSS program, the CROSSTABS command crosstabulates a dependent variable (Y) by an independent variable (X).

A dependent variable is the variable to be analyzed; it depends on a causal variable, the independent variable.

Thus the command

crosstab discuss by nation/cells=count col.

can be interpreted by this template

crosstabs [dependent var] BY [independent var]/ cells=count col.

[the expression after the slash (/) says: for each cell give the number of cases and the % of the column total]

So the SPSS output at the top above was produced with this command above.

Analyzing Variables Controlling for Nations

In addition to uncovering differences across nations for the variables listed above, you my want to determine whether relationships between variables holds within nations. CROSSTABS allows you to do this by specifying a second BY command with NATION listed after it. Consider this example, which cross-tabulates DISCUSS BY L.RSCALE BY NATION.

CROSSTABS DISCUSS BY L.RSCALE BY NATION/CELLS=COUNT COL.

You'll obtain five separate tables--each one reporting the cross-tabulation for a different country. Here's the first table, for the U.S. (Five different tables will be reported, one for each nation.)

 

         
         
DISCUSS  v03 Do you DISCUSS POLITICS often?
                    by  L.RSCALE  v 14 LEFT/RIGHT SCALE
                        Controlling for..NATION  v01 NATION  
                                          Value = 1  United States
                    L.RSCALE                       Page 1 of 1
            Count  |
           Col Pct |Left (1, Center/l Center/r Right (8
                   |2,3)     eft (4,5 ight (6, ,9,10)     Row
                   |     1  |     2  |     3  |     4  | Total
DISCUSS    --------+--------+--------+--------+--------+ 
                1  |    32  |   108  |    62  |    50  |   253
  Frequently       |  22.9  |  15.2  |  13.5  |  18.4  |  16.0
                   +--------+--------+--------+--------+
                2  |    72  |   417  |   297  |   144  |   930
  Occasionally     |  51.3  |  58.7  |  64.6  |  53.0  |  58.8
                   +--------+--------+--------+--------+
                3  |    36  |   185  |   101  |    78  |   400
  Never            |  25.8  |  26.1  |  21.9  |  28.6  |  25.3
                   +--------+--------+--------+--------+
            Column     141      711      460      271     1583
             Total     8.9     44.9     29.0     17.1    100.0

         
         

      		
   

   

      
         
Here, you see that Americans on the ideological extremes
         tend to discuss more frequently than those in the center.
         What about for other countries? You can study the other four
         tables to see whether the same relationship holds in the
         other countries.