Introduction to Research Methods in Political Science: |

**CROSSTABS**

From the menu bar, click on “Analyze,” then on “Descriptive Statistics,” and on “Crosstabs….” In the left window, select the variable(s) you wish to place in the rows, and click on the top right arrow. Then select the variable(s) you wish to place in the columns, and click on the second right arrow. If you have one or more control variables, select them and click on the third right arrow. For second and subsequent layers of controls, click on “Next” before selecting the variable(s). Note: CROSSTABS will produce tables for each combination of variables specified. For example, if you select two row and two columns variables, SPSS will generate four tables. If you wish to accept all defaults (which will provide the number of cases in each table cell), simply click on “OK.”

If you want to “Display clustered bar charts,” and/or “Suppress tables,” check the appropriate boxes before clicking on “OK.” (I strongly recommend, however, that you not use Crosstabs to produce clustered bar charts, but instead use clustered bar charts, which provide you with much more control over the output.)

There are a number of additional options you can choose before clicking on “OK.”

To obtain measures of statistical significance and/or association, click on “Statistics” and check the appropriate boxes, then click on “Continue.” Note the following:

- If one or both of your variables is nominal, use chi-square as the measure of statistical significance. If both your independent and dependent variables are ordinal, use the “Approx. Sig.” (i.e., approximate significance) that comes with the measure of association you are using. (This measure is actually a t-test.)
- If you ask for “chi-square,” you will be given several chi-square tests. The most common (and the one described in the statistical significance topic) is “Pearson’s chi-square.” (If you have a table consisting of two rows and two columns, use the “Continuity Correction,” also known as “Yates’ correction.” This is especially important for small samples; for large samples, the correction will not make much difference.)
- If the probability of a test for statistical significance is given as “.000” this does not really mean that there is a zero probability of the relationship occurring by chance. Rather, it means that the probability is less than .0005.
- Some measures of association (Lambda, Goodman and Kruskal tau, the Uncertainty Coefficient, and Somers’ d) are “asymmetric” measures. Since SPSS does not know which of your variables is dependent, it will calculate the statistics both ways (and, except for Goodman and Kruskal tau, will also provide a “symmetric” average of the two). It is up to you to know which of your variables is the dependent variable, and to choose the appropriate measure accordingly.

To obtain additional cell information, and/or suppress “observed cell counts,” click on “Cells,” check or uncheck the appropriate boxes, then click on “Continue.” Note: be sure to percentage in the direction of the independent variables. If you have placed the independent variables in the columns, check “Column” percentages.

By default, table rows are displayed in ascending order. To display rows in descending order, click on “Format,” select “Descending,” then click on “Continue.”

Note: CROSSTABS provides the option of producing clustered bar charts. Avoid it, since it allows bars to represent only raw counts, not percentages. Instead, use the more flexible BAR CHART tool.

Last updated
April 28, 2013 .

© 2003---2013 John L. Korey. Licensed under a Creative Commons Attribution-Noncommercial-Share Alike 3.0 Unported License.