Predicting elections is tricky business, and any prediction is always a little bit wrong. In some cases, they can be very wrong. Comparing my prediction from last week to the results we have all been staring at for the past few days, it seems that mine is one of those that went very wrong. Although some races have not been called at the the time of writing, this is what the picture looks like at the moment: instead of Harris sweeping every swing state, Trump is on track to win them all, flipping the states he lost in 2020 and likely adding Nevada to his map; Republicans won the Senate, as I predicted, but they are likely to acquire an even bigger majority than I expected by flipping seats in Ohio and Pennsylvania; and Republicans seem set to hold the House of Representatives, instead of it narrowly flipping to Democrats. Trump has also ended a 20-year drought for the Republican Party by winning the popular vote, avoiding the stain of illegitimacy that hung over his presidency after the 2016 election.
In this article, I will look at some of the reasons my prediction was wrong as well as some interesting data points and trends we are starting to see in the results.
What Went Wrong
In my article, I noted that numerous predictive indicators besides polling showed that Harris was favored over Trump. These included fundraising, shifts in suburban/educated voters, the “Dobbs Effect,” and the Washington Primary, among others.
While it appears Harris’s strong fundraising numbers allowed her to keep the results in many swing states close, it was not enough to counter a general nationwide swing against her. This is not unprecedented; while a financial advantage is certainly a benefit to any candidate, many have won with a strong fundraising deficit.
On shifts among suburban and educated voters, it seems that we have a classic case of assuming that a trend is guaranteed to continue simply because it has been present for a long time. Although it is not yet entirely clear, it appears that while fast-growing suburbs in places such as the Atlanta metropolitan area did indeed swing left slightly, this pattern was not replicated nationwide; suburbs generally shifted at a similar rate to the rest of the country. Further, educated voters did not continue to trend left, bucking a pattern that seemed unstoppable in 2016 and 2020.
As for the Dobbs effect, it appears that, unlike in 2022, the economy clearly trumped (pun not intended) abortion in priority among most voters. Besides the fact that the overturning of Roe v. Wade is further back in voters’ minds due to the simple passage of time, it is possible that referenda on abortion, which were on the ballot in 10 states this year, provided “release valves” for pro-choice voters to stick with their principles while also voting for Trump. In Florida, for instance, although Trump won the state by 13 points, voters supported a referendum to protect abortion rights by a 14-point margin (it still failed, however, because Florida recently began requiring that referenda receive at least 60% of the vote to pass).
Finally, the Washington Primary did not provide its usual insight into election outcomes. Along with other well-known predictors such as the gold-standard Selzer poll of Iowa (which, despite getting the results in the state almost exactly right for nearly two decades, was roughly 16 points off from the final result this year), Nevada elections guru Jon Ralston, and creator of the “Thirteen Keys to the White House” Allan Lichtman (whom TJI recently interviewed), the Washington Primary’s status as an electoral crystal ball came to an inglorious end in 2024. While this is much easier to state with the benefit of retrospect, perhaps it was unwise to assume that a single state’s primary election results could really be extrapolated to the nation at large. Indeed, in what seems like an act of cosmic retribution for my hubris, it appears that Washington will be the only state in the nation to trend further Democratic than 2020.
In the end, simply going with the polling may have been the best best—or was it? On the one hand, polling was not horribly inaccurate this year, as it accurately captured that the swing states would be close and was only around 2-3 points off of the final popular vote total—which will be slightly narrower than current results might show, as the blue states on the West Coast are very slow to count their votes. On the other hand, issues with herding and a deluge of low-quality, Republican-biased pollsters still remain. Indeed, one could say that the ultimate goal of “herding” was accomplished: pollsters hedged their bets on a close election, tweaked results to eliminate outliers from that assumption, and were subsequently rewarded for it, as the current political system generally produces close results. Of course, this does not resolve the fact that the number of tied or near-tied polls produced this election was statistically improbable. These issues could spell doom for pollsters in a future cycle where they are not so lucky as to find themselves in a typically close election.
A Look at the Data
Since the results have not been fully tallied, it is a bit difficult to make definite conclusions about them. From the counties that have fully counted their votes, however, we can extrapolate some trends.
Looking at the (currently incomplete) swing map compiled by the BBC, some trends are immediately noticeable. For one, while there was certainly a swing to the right in most of the country, the largest swings came from highly populated “safe” states such as Florida, Texas, New York, New Jersey, and California. In swing states, by contrast, there was a much more modest swing right, and some suburban areas swung slightly to the left. The areas that swung most to the right appear to be the New York City metropolitan area, South Florida, and areas of Texas along the US-Mexico border. All of these areas have large Hispanic populations, likely indicating a large rightward shift among that group.
Analysis by The New York Times confirms this. It found that Hispanic- and Native American-majority counties experienced double-digit swings to the right; Black-majority counties, on the other hand, swung only slightly to the right, and less than the nation as a whole. Interestingly, there does not appear to be much of a correlation between the education level of a county or its level of urbanization and how much it swung toward Trump. Trump gained in suburban, urban, and rural counties alike; the difference in the swing between these three categories was, at most, under 2%. The difference in swing between less-educated and more-educated counties was less than a percentage point, indicating that there was likely not an educational realignment either.
These results are, of course, provisional. Analysts will likely have to wait weeks to make definite conclusions about the election.
The opinions expressed within this piece represent the views of the author alone and do not necessarily reflect the views of The Jefferson Independent.
Leave a Reply