FiveThirtyEight forecasts the Republican Party is 85% likely to win the House of Representatives. That’s about a 5-in-6 chance. Inversely, Democrats have a 1-in-6 chance of winning. That’s about the likelihood of rolling a certain number on a typical die. Meanwhile, FiveThirtyEight also forecasts Republicans have become the favorites in the Senate, and are in fact more likely than not to sweep both chambers.
But if they and others are wrong, it wouldn’t be the first time prognosticators screwed up. Just ask presidents Thomas Dewey and Hillary Clinton.
About a week before the 2016 election, I wrote a piece called “How Trump Wins,” and I walked the reader through how Trump, despite being down in the polls, could still win. His victory taught us that it’s possible for everything to break in a surprising direction. We should therefore ask: If Democrats end up winning the House, what the heck happened?
The answer centers around a plausible scenario: the polls are wrong. Specifically, they aren’t measuring the right electorate.
One reason that might be the case is that the Dobbs v. Jackson ruling convinced a lot of people who never voted before to vote for Democrats, and pollsters aren’t factoring in enough of these people.
There’s a bit of evidence to support this theory. First, recall that back when pollsters stuck with “Registered Voters” — before their end-of-summer switch to “Likely Voters” — Democrats did much better on the generic ballot. If you go deep into the generic ballot polling from Real Clear Politics, you’ll see a time where Democrats led just as often as they trailed. As recently as late August, they led some of these Registered Voter polls by 5, 6, even 8 points. Once the switch to Likely Voter models occurred, however, Republicans led more often than not.
But what qualifies a respondent as a “Likely Voter”? Pollsters certainly ask the question of whether the respondent is likely to vote, but a lot more people (perhaps 90%) say they intend to vote, but not all of them actually do, and pollsters know this. They therefore ask various follow-up questions to ascertain other bits of information, among which is the respondent’s rate of voting in past elections. If they have consistently voted in the past, they are sorted into the “Likely Voter” category now. Those who haven’t voted in the past are less likely to be categorized as “Likely Voters” in the polling results.
So maybe Dobbs brings to voting precincts the formerly apathetic pro-choice-minded folks who thought Roe had settled the issue but whose poor voting record categorized them as an unlikely voter. We can see this issue being particularly resonant with those of a fertile age. (I can’t tell if that’s the worst way to put it or the absolute best way.) Young people, who are typically less reliable voters and might more often be categorized as “unlikely” to vote as a result, might make it a mission to turn out to the polls — and make sure their significant other is too. Recall that in 2018, youth turnout (age 18-29) climbed at a higher rate than any other age group, a big reason why Democrats had their greatest midterm performance in the House of Representatives since Watergate.
Keep in mind that a surge in voter turnout is typically good for Democrats for the same reason that Republicans are typically the stronger midterm party. Republicans skew older and whiter, two demographics that show up to vote at a high rate, even in non-presidential years, which gives them disproportionate power when other demographics stay home. But when more people show up to the polls, that means minorities and young people, who skew Democratic, are casting ballots at high rates as well. That’s what happens in presidential years, which explains why Democrats have won seven of the last eight popular votes for the presidency. It also happened in 2018, which was the highest midterm turnout in a hundred years. It’s once again looking to be an enormous midterm electorate, the size of which is on track to actually surpass that of 2018. This could mean a surprising Democratic turnout.
Another sign of high political engagement is fundraising, which is yet another indicator that Democrats are, despite what the polls suggest, still very much invested in the election. Democrats in key Senate races have far out-earned their Republican opponents, including in an important metric: number of small donations, which is even more representative of the voting public than general fundraising. Democrats’ lead in fundraising and small donations suggests they’re just as engaged as Republicans, if not more so.
And then there’s this familiar accusation: the polls are rigged. I’ve been monitoring a belief that Republican pollsters (Trafalgar, Rasmussen, American Greatness/Insider Advantage, Cygnal, Remington Research) have been flooding the polling aggregators (like Real Clear Politics) with partisan surveys, inflating polling averages for Republican candidates in swing states. If you scroll through the RCP polling in Georgia, Pennsylvania, and Nevada, we can certainly see discrepancies between those pollsters and the others.
Democrats’ fear is not only that this strategy helps drive the narrative of an inevitable Republican win, which might discourage Democratic turnout and/or convince independents to climb aboard the band wagon, but that such a practice will produce something much worse: Republicans will use these skewed polling averages as evidence of election fraud once Democrats win. Regardless of the motivation, these Republican-affiliated pollsters have indeed been publishing their polls at a much higher rate than other polling outlets, including in key Senate battlegrounds, thus shaping the polling averages off of which PPFA and so many others work.
But polls don’t win elections. Voters do. And so far this year, voters have shown it’s actually the Democratic Party that’s overperforming. There is no poll more relevant than actual election results, and Democrats have come away pleased in the five post-Dobbs special House elections to fill vacancies, beating Biden’s 2020 performance and the partisan lean of each district. There was also the Kansas abortion referendum surprisingly won by the pro-choice side.
Unlike skewed polls, these are actual ballots, and nothing about those ballots suggests Democratic regression. If anything, they suggest Democrats will gain ground.
Working in tandem with the idea that polls are off and voters will surprise us are reports of early voting trends. To be clear, this data, which tells us the partisan breakdown of who has turned in ballots (but not who they voted for), is not always predictive. Each side likes to divine the results of the election from these early ballots and are wrong just as often as they are right.
But that hasn’t stopped some Democrats from getting high on some hopeium by heavily using these early voting statistics in their analyses. Indeed, some of this early voting data should raise some eyebrows. You can find early voting data at any number of websites, like this one from NBC news which shows that over 33 million people have already voted, and 45% of them have been registered Democrats compared to just 36% by registered Republicans. Registered Democrats also outpace registered Republicans in the battlegrounds. In Pennsylvania, for instance, early Democratic voters (and again, it’s just registered Democrats; we don’t know who they actually voted for) comprise 69% of early ballots compared to just 21% for Republicans. Registered Democrats who have turned in ballots also outpace their Republican counterparts in Arizona, Georgia, Nevada, and New Hampshire, which together, with Pennsylvania, comprise the swingiest states in this election.
Of course, that statistic alone isn’t indicative of much. We know it’s become part of political culture for Democrats to trust in early voting and Republicans to believe “2,000 Mules” was a legitimate documentary. The crucial context, therefore, is to examine how these trends compare to past voting cycles, particularly 2018 (when Democrats’ blue wave took the House) and 2020 (when President Trump signaled to Republicans to not trust early voting).
That’s where the helpful TargetEarly website comes in. It not only keeps track of early voting across the country, separating by affiliation, but it also compares this data with that of 2018 and 2020 at the same point in the election cycle. Here’s what it has as of last night, comparing it to five days before the past two elections:
At this point in 2018, it was actually registered Republican voters who had a narrow lead among early voters. That was in an election where Democrats ultimately won the House popular vote by over 8 points, flipping 40 seats to win a big majority. But now, it’s Democrats who make up the much larger share of early voters, and that’s with more overall votes cast than at this point in 2018.
In the early voting two years later, we see that registered Democrats far outpaced the members of the party whose leader was telling them to vote in person because he planned on delegitimizing mail-in voting, and the Democratic Party ultimately won the House, Senate, and presidency. This year, registered Democratic voters are beating that rate.
The lead is lower if we just consider the battleground states (AZ, CO, FL, GA, NC, NH, NV, OH, PA, UT, WI) identified by TargetEarly, but this year’s registered Democrats in those states still outpace their predecessors from the last two elections, and again with larger overall numbers compared to the last midterm election:
Last weekend, the website’s CEO took a closer look at Pennsylvania and noted that not only are early-voting registered Democrats this year outpacing Democrats from two years ago, but the gender gap is more tilted in Democrats’ favor as well, which certainly fits the narrative that once-politically apathetic women have been activated by Dobbs. (Indeed, female registration nationally has climbed at a higher rate than male registration in states where reproductive versus fetal rights are top-of-mind.) There were also already about 210,000 Pennsylvania voters who have already cast their ballot in this election but had not voted in 2018, and among that group registered Democrats had a 47-point over registered Republicans.
Considering all of the above — uncategorized Likely Voters, Democratic overperformance in special elections, the Democratic lead in fundraising, Republican-pollsters skewing polling averages, and early voting trends — does it really look like a red wave to you? At the very least it appears like a competitive election.
And maybe, just maybe… Democrats might win after all?
See you on Sunday for my final deep dive into the Senate.
Today’s featured image is courtesy of Steven Braeger, CC0, via Wikimedia Commons. No donkeys were harmed in the making of this post.