Congressional Democrats who get reelected on rainy days become more conservative

It’s raining today across most of the East Coast, and research indicates that this could impact not just today’s election but the subsequent behavior of members of congress — even incumbents who win reelection..

It’s true that, logically speaking, whether it rained on the first Tuesday of the previous November should not influence a member of Congress’s thinking about a pending piece of legislation.

But a paper from John Henderson at Yale and John Brooks from UC Berkeley shows not only that it does, but that it does so in a specific direction — House members amass a more conservative voting record if it rained on their district’s most recent Election Day.

This effect, the authors find, is concentrated “mainly in competitive districts with the greatest risk of defeat,” and Democratic members of Congress are much more likely to experience rain effects than are Republicans.

Studying a big sample of US House races from 1956 to 2008, they find that this is likely driven by the impact of rain on voter turnout.

Specifically, Democrats are pickier about whether they vote than are Republicans (which is why the GOP gets a turnout edge in midterms), and thus rainy days appear to depress Democratic vote margins by 1.4 to 2 percentage points.

In other words, a Democrat who would win with 53 percent of the vote on a sunny day might only win by 51 percent on a rainy day.

It’s easy enough to see how an unexpectedly close reelection fight driven by unexpectedly low Democratic turnout could push an incumbent Democrat to reposition himself somewhat more conservatively in the future. But in theory, savvy political advisers ought to be able to tell you that the unexpectedly low turnout was driven by unexpectedly bad weather and there’s no need to have this reaction.

And yet, this doesn’t happen.

Either political professionals don’t realize that rain depresses Democratic turnout (which seems strange to me, since I heard political pros tell me this many times over the years before I saw research proving it) or else they somehow don’t connect the dots correctly. Henderson and Brooks refer to this phenomenon as “elite information uncertainty,” which I think in layman’s terms I would call not thinking clearly.

The fact that random fluctuations in the weather have a meaningful impact on congressional voting behavior and policy outcomes is, itself, a little bit weird. But the important point to grasp is that it just happens to be the case that accurate and comprehensive weather reports are available that date back to the 1950s. That makes the influence of past rain on future legislation a tractable subject for quantitative social science research, which is how we got this paper.

But there are presumably other kinds of random events that also influence turnout. And if members of Congress aren’t capable of correcting for the weather when thinking back to their last election, they probably aren’t correcting for any of the rest of them.