Almost exactly a year ago, I expressed skepticism that the Republican Party’s wins in Bucks County in the off-cycle 2021 election “amount[ed] to the GOP roaring back to life in the Pennsylvania suburbs.” In that post, I grumbled that: “No one ever seriously discusses turnout” when making predictions, and I remarked that “[w]e would have to see next year’s midterm turnout drop to 2014 levels for the 2021 election to be a useful benchmark of GOP support in the state.”
So here we are, one year later. Democrats in Pennsylvania captured both state-wide offices, kept their hold on three competitive House districts that many feared would flip, and seemingly secured the very slimmest majority in the state’s lower house for the first time in over a decade. They held on to suburban voters, and those voters turned out at high rates—as one might have expected based on the fact that high SES, college-educated individuals are consistent voters.
Democrats in Pennsylvania did much better than expected, as they did in Colorado and elsewhere. But this was not true not everywhere. The President’s party did worse than expected in Florida, but also in New York and Connecticut, and both Beto O’Rourke and Stacey Abrams lost their elections by wider margins than in their previous bids.
It is well-known that states vary a great deal when it comes to turnout. Turnout in a handful of states always outpaces the rest of the country. Minnesota, Maine, Colorado, Oregon, and Wisconsin generally leading the pack, including in midterms.
So, what might we learn if we compared turnout in states where the Democrats exceeded expectations to turnout where their losses were worse than anticipated? The simplest and most direct comparison would be Pennsylvania versus Florida—two consistently competitive states; both of which generally have middling turnout numbers.
Preliminary reports suggest that turnout in Pennsylvania was extremely high. U.S. Elections Project estimates 55% of eligible voters turned out in Pennsylvania in 2022. By contrast, it estimates that turnout in Florida was 49.3% of VEP. In other words, Pennsylvanians turned out at a higher rate than in the 2018 midterm election (which boasted the highest overall turnout in 100 years), while Floridians turned out at rates only slightly above the 2014 midterm (which set the record for the lowest voter turnout since 1942). For those who are curious, both states are capable of high and low turnout: Turnout in Florida in 2018 was 54.3%, and turnout in Pennsylvania in 2014 was 36.5%.
It is too early to have a good picture of how representative the electorate that turned out on November 8 was along axes of class, race, and education. But we do have some early information about representativeness with respect to age. A critical mass of young voters turned out in Pennsylvania, and some argue they were decisive to Fetterman’s victory. Florida’s young voters turned out at comparable levels, but as in Ohio, young voters in Florida are more divided in their partisan preferences than those in Pennsylvania.
The voter turnout story is not simple, however. In Colorado, where Democrats exceeded expectations, voter turnout is estimated to be slightly down from 2018, but hardly bad: Turnout in Colorado in 2018 was 61.4%. This year 57. 9% of voters in Colorado are estimated to have turned out. Turnout was also high in Michigan where it is estimated to have reached 59% this cycle (compared to 57.7% in 2018 and 43.2% in 2014).
In New York, where Democrats did worse than expected voter turnout out was abysmal (42.6%), but nowhere near as low as it was in 2014 (29.0%) and only slightly down from 2018 (45.7%). And, at least, based on the preliminary data, it is hard to see either Beto O’Rourke or Stacey Abrams’ losses as a product of significantly depressed voter turnout.
The conventional wisdom is, of course, that low voter turnout doesn’t matter because the electorate that turns out largely shares the same preferences as those who stay home. I am on record as skeptical about this conventional wisdom.
So, would more systematic attention to variable voter turnout complicated the conventional wisdom that the President’s party always gets a shellacking in the first midterm after their election? It is hard to tell, but probably, to varying degrees, in different places.
But the real point is: Wouldn’t it be great if we could hold turnout constant in our efforts to analyze the political implications of elections? Wouldn’t it be great if more states were like Minnesota, Maine, Colorado, and Wisconsin and fewer were like New York, Texas and Tennessee (which had the lowest turnout this cycle)?