www.wsj.com/politics/pol...

But this doesn't really tell the whole story. As Trump's approval rating has dropped, the magnitude of Democratic over-performances has grown. In the specials since April 1, the median Democratic over-performance is 16.1%.

More recommendations on this slide:

This slide deck from @yiqingxu.bsky.social and co-authors on issues with TWFE in Political Science based on their large re-analysis study that's conditionally accepted in APSR is really great. Key takeaways below. yiqingxu.org/papers/engli...

Even if preferences are two-dimensional, we show that we can address B&L’s concern by allowing for a second group of non-Downsian respondents. When we do this, the estimated share of Downsian respondents decreases, but our other substantive conclusions are virtually unchanged. (5/7)

We also offer new analyses showing that a one-dimensional model summarizes public opinion well. A one-dimensional model correctly predicts more than 75 percent of CES respondents' policy views, and the addition of a second dimension only minimally improves these predictions. (4/7)

First, B&L’s visualizations of our data do not undermine our method or results. This graph shows that even if survey responses were perfectly explained by our mixture model and even if our classifications were perfect, we would expect to see the visual patterns that they find. (3/7)

This paper from B&L in APSR First View responds to our study (Fowler et al. 2023). We very much appreciate the dialogue. However, we believe our results are robust to the exercise B&L have undertaken and have produced a series of analyses to support that conclusion. tinyurl.com/5a7x26e7 (1/7)

Here's a neat plot from an early stage project with Devin Caughey and Seth Hill. The correlation between states' policy liberalism and an index of their societal outcomes was .8 in 2021.

In the paper, we provide an illustrative application focusing on descriptive representation in local governments, and the continuing under-representation of women and racial minorities at nearly all levels of government.

Lots of discussion on how economy will affect 2024 election. Evidence indicates that election year economy matters a lot (esp. real wage growth), but the economy this year prob. doesn't matter much. So a lot depends on what happens next year! See this figure from my paper w/ @jdbk.bsky.social.