Today marks the 50th anniversary of the enactment of the Civil Rights Act. The Civil Rights Act passed the House in a 290-130 vote and the Senate in a 73-27 vote. Below we use DW-NOMINATE scores to plot both votes. First dimension DW-NOMINATE scores represent legislators’ ideological positions along the liberal-conservative spectrum and, during this period (as well as throughout the mid-twentieth century), the second dimension captures regional differences among members of Congress. Regional divides were most apparent on votes involving race and civil rights issues.
Accordingly, the divide on both of these votes separating predicted “Yeas” from predicted “Nays” runs mostly along the second (regional) dimension. Most Southern Democrats (with high second dimension scores) voted against the bill in both chambers and were joined by a handful of mostly conservative Republicans (like Senator Barry Goldwater [R-AZ]). But both votes far more reflected regional differences than ideological ones.
In the half-century since passage of the Civil Rights Act, the second regional dimension has all but disappeared in congressional voting. Racial issues have become questions of redistribution (a fundamental liberal-conservative disagreement), and social/cultural issues like gun control and abortion have also folded into the liberal-conservative dimension. To illustrate this, in the final image we show how well one and two-dimensional models classify congressional voting on the issues of abortion, gay rights, gun control, and immigration. The measure of classification is the APRE (Aggregate Proportional Reduction in Error), and measures how much improvement the model offers in classification of legislators’ votes. The maximum APRE is 1 (complete improvement), while a value of 0 indicates no improvement over the null model. As can be seen, the fit of all four issues to a one-dimensional ideological (liberal-conservative) model has steadily climbed to nearly 1 over recent decades. Both racial and social/cultural issues are now largely encompassed by the primary liberal-conservative dimension.