"When legislators vote against their party... their constituents are very unlikely to guess that. They become worse than chance at guessing how their member voted on a roll call vote. When their member voted with their party, then people are pretty good at guessing how their member voted."

This research speaks to the debate about whether moderates do better in elections. In our view the effects are small because voters have a very limited ability to discriminate within-party. But it does not follow that parties are best served by taking more extreme stances.

There is much more in the paper, including discussion of the main exception to our story (the Affordable Care Act), examination of heuristics other than party, and an analysis of how knowledge of positions affects vote choice and approval.

That said, in an original data collection we show that a majority of people can guess the party of their representative, and on most issues solid majorities can guess the positions of the two parties. Accountability is possible, but it is mainly partisan accountability.

Most of the counterexamples end up being consistent with the importance of party. For instance, people are more likely to get Republicans right when they vote for the Violence Against Women Act, because people mistakenly believe that most Republicans voted for it.

We show that for most individual votes, people are much more likely than chance to guess their representatives’ positions when they vote with their party, and less likely than chance to guess their positions when they vote against their party.

We reanalyze AK’s Table 3, showing that perceptions of roll call votes are related to reality. It turns out this is only true when representatives vote with their party. When they don’t, the correlation is reversed.

AK claim that “perceived issue agreement on key legislation does translate into electoral support, and the effects are substantial and operate independent of party.” We agree with all but the last clause. Party is essential.

We pick on work by Ansolabehere, Jones and @shirokuriwaki. This is high quality political science that has inspired us personally, but in our view makes claims that are not correct. Ansolabehere and Kuriwaki (2022), henceforth AK, is possibly the most influential paper on this.

New paper with @johnmmurph.bsky.social just conditionally accepted at @apsrjournal.bsky.social: Can people guess how majorities of each party in Congress voted on bills? Mostly yes. Can they guess how their own representatives voted on bills, above what is accounted for by party? Mostly no. 🧵