One collection of data people obsess over in the weeks before a big election is the early vote – a relatively new feature of American elections. You can’t know who people voted for but you can know who is voting by different metrics and you can draw inferences from that. Take all this early vote data with a huge, huge grain of salt. On the one hand it’s just way to easy to interpret the numbers through the prism of one’s hopes. But there’s a bigger and more immediate reason for skepticism today. To understand what early numbers mean you need to have some theory of who is going to vote early and who late. And basically, I think, we don’t. Or not a good one.
As we know, early voting and mail voting has become hugely politicized in recent years. Democrats disproportionately voted by mail during the pandemic. Donald Trump made in-person voting an ideological signifier. Then Republicans, if not always Trump, decided this was super dumb and counterproductive and tried to get their voters to vote early and by mail again in 2022 and especially in 2024. The problem is that these shifts are overlaid by the differences between midterm and presidential years. Is the baseline for 2026 the 2024 election? Well, that’s a presidential year so it can’t really be. But 2022 shouldn’t be either because that’s probably before a lot of the shift back to less partisanized vote-time patterns. We just don’t have a good point of comparison, even for the inherently iffy attempt to infer election results from who is voting when. The answer is just to ignore this stuff. There’s lots of interesting information in the early vote but not about who is going to win.
