Since the end of World War II the structure of Congressional voting has changed dramatically. With few exceptions, roll call voting throughout American history has been simply structured. Only two dimensions are required to account for the great bulk of roll call voting. The primary dimension is the basic issue of the role of the government in the economy, in modern terms liberal-moderate-conservative. The second dimension picked up regional differences with the United States — first slavery, then bimetalism, and after 1937, Civil Rights for African-Americans. With the passage of the 1964 Civil Rights Act, the 1965 Voting Rights Act, and the 1968 Open Housing Act, this second dimension slowly declined in importance and is now almost totally absent. Race related issues – affirmative action, welfare, Medicaid, subsidized housing, etc. – are now largely questions of redistribution. Voting on race related issues now takes place along the liberal-conservative dimension and the old split in the Democratic Party between North and South has largely disappeared. Voting in Congress is now almost purely one-dimensional – a single dimension accounts for over 94 percent of roll call voting choices in the 111th House and Senate – and the two parties are increasingly polarized.
The first graph below uses Optimal Classification to analyze the 32 Houses and Senates since the end of World War II in one and two dimensions. The upper red line shows the fit (correct classification) in two dimensions and the lower blue line shows the fit in one dimension. After bottoming out in the 1970s the fits in one and two dimensions have been steadily rising as more and more issues are drawn into the left-right alignment.
Click image to enlarge
The two plots below shows the difference in correct classification between the first and second dimensions in the left panels and the difference in the aggregate proportional reduction in error (APRE) in the right panels. The APRE statistic controls for the roll call margin. In both the House and Senate as issues related to Civil Rights aligned along the first dimension the second dimension begins to decline in importance. The second dimension still picks up some voting on Abortion, gun control, and other lifestyle issues but these also are increasingly voted on along liberal-conservative lines.
Click images to enlarge
In the graph below we show APRE for two dimensions over time for the House and Senate by two groups of roll call margins. The top two lines show the APRE for close votes and the bottom two lines show the APRE for lopsided votes. Close votes tend to include most of the party-line and party-related organizational votes and these will be a good fit to the spatial (geometric) model. The trends in the APRE for close votes track the upward climb in the overall correct classifications for both chambers since the 1970s.
The APRE for lopsided votes tends to track the APRE lines for close votes but are uniformly lower. The lopsided votes are those with margins of 75-25, 85-15, and so on. When the minority on the vote is only 15 percent of the membership this means that if the cutting point or line is placed at the edge of the the legislator configuration, then by definition the model is classifying 85 percent of the members but is not explaining anything. In this case the PRE for that roll call would be zero. However, if the roll call is one where all of one party plus a large group of the opposite party adjacent to it in the configuration all vote YEA and the 15 percent splinter of the second party votes NAY then an interior cutting point (cutting line) can be placed that accounts for most of the voting. Hence, the APRE measure is almost certainly going to be lower for lopsided votes. However, again, since the 1970s the APRE measure for lopsided votes flattens out and then starts rising again in the 1990s.
The bottom line here is that even though both Chambers have become much more one-dimension and hyper-partisan, the geometric model is doing well because there are many roll calls where one party plus a faction of the opposite party are voting together against the remainder of the of the opposite party. This is true of both parties and why the geometric model works so well even though party-line voting continues to increase.