Click Here to Return Home The Challenge of Democracy
Chapter 11

Crosstabs
-

Return to Table of Contents

Crosstabs
A Computer Program for Analyzing Political Data

The sidebar on page 342 directs you to the Starr Report and other documents related to the House of Representative's deliberations on impeaching President Clinton. Our book was at the printer when the House impeachment vote was taken on 19 December 1998, so the vote is not discussed in the text. However, we were able to include the Members' impeachment votes in our CROSSTABS dataset, and you can analyze their voting patterns.
Most people know that Members cast their votes for and against impeachment largely along party lines. But the amount of partisanship depended on which Article of Impeachment you examine. The House voted on four separate articles:
Article I: Perjury Before the Grand Jury
The president provided perjurious, false and misleading testimony to the grand jury regarding the Paula Jones case and his relationship with Monica Lewinsky. Passed by a vote of 228 to 206.
Article II: Perjury in the Paul Jones Case
The president provided perjurious, false and misleading testimony in the Jones case in his answers to written questions and in his deposition. Failed by a vote of 205 to 229.
Article III: Obstruction of Justice
The president obstructed justice in an effort to delay, impede, cover up and conceal the existence of evidence related to the Jones case. Passed by a vote of 221 to 212.
Article IV: Abuse of Power
The president misused and abused his office by making perjurious, false and misleading statements to Congress. Failed by a vote of 148 to 285.
Only Articles I and III passed, but all four votes were cloaked in partisanship. Here is how Republican and Democratic members (including one Independent) voted on the Articles.


 

The CROSSTABS' "Job Rating" menu in the CONGRESS datset, contains a variable for the number of times each Representative voted for or against impeachment. If you analyze that variable against the Members' party affiliation, you should reproduce the pattern above. In causal language, party affiliation is the cause and vote for impeachment is the effect. Expressed in another way, party is the independent variable and the impeachment voting pattern is the dependent variable.

But you can expand your analysis to include the effect of support for Clinton in Members' congressional districts on their voting. First, run the vote on impeachment against the percentage of the 1996 vote cast for Clinton in the Members' districts. (That variable is under the "District Traits" menu.) Then (in the language of social research) hold constant the effect of party affiliation to assess whether the Members' district vote for Clinton in 1996 affected how they voted for impeachment in 1998. Do this separately for Democrats and Republicans by choosing party as the control variable under the data menu.
The convention for constructing analytical tables in social research is to place dependent variables along the rows of a table, independent variables along the columns, and then compute percentages according to the column totals so that the total percentages in each column sum to 100%. What's different when using a control variable is that you should generate separate tables for each category in the control variable. Given the existence of only one independent (Bernard Sanders of Vermont), you should only generate a crosstabulation for Democrats and another for Republicans.

Place the vote for impeachment in the rows of your crosstab table, put the district vote for Clinton in the columns, choose "% by Cols" in the "Display" menu, and choose party affiliation as the control variable.

Go to Houghton Mifflin's CROSSTABS page to run the program.