Emory political scientist Alan Abramowitz has done two studies of how voting ideology affected the electoral fortunes of Republican and Democratic senatorial incumbents over the 2000-2008 period. The study on Republicans is here, and the study on Democrats is here. Over this time period, 57 of 61 Democratic incumbents won their re-election campaigns, while just 52 of 69 Republican incumbents won theirs.
Using DW-NOMINATE ideology scores, which are based on congressional roll call votes, Abramowitz finds that for Republican U.S. Senate incumbents every additional point of conservatism correlates with a three-point decline in electoral performance relative to the Republican presidential candidate. However, for Democratic senatorial incumbents, there’s no such effect.
Does this mean that insurgent Tea Party candidates that want the party to “move right” will actually cause more Republicans to go down to defeat in November, while Democrats can afford to indulge their liberal wing? I doubt we can draw those inferences. It seems to me that there are two caveats about Abramowitz’s results.
First, Tea Party-ism focuses on fiscal issues, one area where most Republicans did not vote “conservatively” during the Bush years. DW-NOMINATE scores are not an absolute measure of ideology against some fixed scale (like the Nolan Chart), but a description of how often a senator tended to vote with other senators of his or her own party. It’s really a measure of partisan polarization. If you voted with leadership 100% of the time (including in favor of Medicare Part D, for instance), you would end up looking 100% conservative. By this standard, some Tea Party Republicans might have looked moderate during this period by this measure. It’s not surprising that Republican incumbents were punished by voters for sticking with the party line on issues such as Iraq, where the “party line” eventually became deeply unpopular.
Second, if politicians are strategic, there should be endogeneity in these voting-ideology models that biases the coefficients on ideology toward zero. If you expect to have a close race, you will modify your voting record in a moderate direction. Abramowitz does include state presidential vote share as a control variable, but the best thing to do would be to find some instruments for ideology – factors that cause candidates to become more partisan than their state is willing to support. A more sophisticated analysis might indeed find that, like Republicans, Democrats are hurt by party-line voting.
HT: Ed Kilgore.