Below we use Optimal Classification (OC) in R to plot two recent votes in the House. The first is its July 24 vote on an amendment by Rep. Justin Amash (R-MI) to defund the NSA’s controversial phone surveillance program, which failed by a 205-217 margin. The second is its July 17 vote to delay implementation of the Affordable Care Act’s (ACA) individual mandate to purchase health insurance, following a decision by the Obama administration that it will delay the employer mandate for one year. This vote passed by a 251-174 margin.
The House’s vote on the Amash amendment is interesting because both parties split fairly evenly on the issue: Democrats voted 111-83 in favor, while Republicans opposed it 94-134. As the plot below illustrates, there is not a great deal of ideological structure to the vote. To the extent there is, it seems to follow somewhat of a “two ends against the middle” pattern with the most liberal Democrats and most conservative Republicans supporting the measure while the relative moderates in both parties were more likely to oppose it. The first-dimension divide is more pronounced among Democrats: the mean first dimension coordinate of Yea Democrats was -0.75, while for Nay Democrats it was -0.62 (p < .01). The mean of Yea Republicans is 0.64 and the mean of Nay Republicans is 0.57 (p < .-01). OC actually calculates the optimal cutting line as dividing legislators along the second dimension, which which may be picking up an "establishment vs. outsider" divide (although the Proportional Reduction in Error [PRE] produced by this cutting line is a fairly low 0.42).
On the vote to delay the individual mandate, the first dimension is much more powerful in classifying the 22 House Democrats who supported the measure. These 22 Democrats constitute most of the remaining moderate-to-conservative Blue Dogs remaining in the House.