Brian Gaines is a professor of political science at the University of Illinois Urbana-Champaign and the Honorable W. Russell Arrington Professor in State Politics at the U of I System’s Institute of Government and Public Affairs. Gaines, who studies elections and public opinion, spoke with News Bureau business and law editor Phil Ciciora about the state of the 2024 presidential election.
How do you interpret the results of the presidential election? Was it a “change” election meant to punish the incumbent party or a fundamental realignment of the electoral map?
I’m very wary of “realignment” claims because they suggest stability going forward. President-elect Donald Trump seems to have made notable gains with some surprising demographics, including nonwhite males. But I’m not confident that those swings reflect enduring changes of party loyalty rather than short-term candidate effects.
How would you characterize the efficacy of pre-election polling versus the actual results from the 2016 and 2020 presidential elections and the recent midterm elections?
Polling is getting harder, in part because there are more and more people unwilling to do surveys of any kind. The oceans have been too heavily fished.
Moreover, polls increasingly mix distinct biases, due to different contact and sampling technologies, from robocalls to online panels to text messages. The increasing use of early voting, whether in-person or remote, also means that late polls report a mixture of intentions about future behavior and reports of past behavior. That fact could be helpful, but it’s another case of the data changing and old assumptions requiring revision.
Do pollsters need to reevaluate how they operate after whiffing so badly in the last few election cycles?
There is a debate underway about whether 2024 is another case of a bad whiff or is, instead, a success.
It’s not new that most of the toss-up cases broke the same way — that is often true. If you take seriously the probabilities that poll aggregators produced for the seven presidential swing states, which were allegedly in the 40-60 range for probability of a Trump win, the probability of a sweep by Trump should have been very low. The probability of either seven or zero heads when tossing seven fair coins is less than 2%.
But the election results are not independent, unlike the coins. Much of the problem is the presentation of calculations. If you really care only who wins, you don’t complain later that some state results were off.
There is nothing surprising in either candidate winning an election that polls suggested was a tie. Donald Trump exceeded poll predictions yet again, but not uniformly so. One can always find individual polls that look horribly wrong. A Des Moines Register poll from late October had Vice President Harris up 47-44, plus or minus 3 points in Iowa, which turned out to be about 56-43 in favor of Trump. YouGov’s final predictions for Trump’s vote shares in Mississippi, Texas and Florida were 53, 51 and 51. The actual results were 62, 56 and 56. But if you compare the joint predictions of Trump and Harris shares taking into account YouGov’s reported margins of error, most of the swing-state results are in the region covered by the “confidence oval,” the set of outcomes they were 95% confident would contain the true values.
Big errors in safe states are, arguably, not too bothersome, even if they show that forecasting is never easy.
The “red wave” never materialized in the 2022 midterm elections. Did it finally arrive with the 2024 presidential election?
I think that the dynamics were somewhat different. 2022 could have revealed more discontent with the president’s party, as did 2018, 2014 and 2010. But Democrats intervened in primaries successfully, saddling the Republican party with some weak candidates and then won the “ground game.” Many concluded that they had proven superior at get-out-the-vote efforts, noted the spending advantage Harris was building, and concluded that 2024 was in the bag for the Democrats.
It turns out that winning over marginal voters is not just about money and organization. A candidate who turns off swing voters cannot be saved by a billion dollars, more offices and better software.
How do you explain the split-ticket discrepancies between the presidency and the Senate that we’re seeing in some states such as Michigan and Wisconsin?
I have argued for a while that polarization, though real, is exaggerated.
Split-ticket voting never vanished. Most claims about split-ticket voting disappearing are based on only president-Senate or president-House comparisons. Consider a few more races, like governor or down-ballot local office, and lots of Americans turn out to be content picking blue-red mixtures. This point is underappreciated in part because so few counties release ballot-image data, which show all of the votes for each voter, rather than only aggregates.
What usually happens when one political party controls the presidency, the U.S. Senate and the U.S. House of Representatives?
Notwithstanding 2022, a good prediction to start off with is that 2026 might be a good year for Democrats. Midterm revolts are not guaranteed, but they have been common.
In the short term, the GOP will have a window to push through a few priorities. One mistake is to try to do too much. Another is to lurch too far in the direction of the party’s core supporters, as Biden arguably did.
Moreover, it looks like the Republican majority in the House will be very small, and Trump has already plucked several members out for his cabinet. Rebels in the Republican caucus have proven able and willing to make life miserable for their own party’s leadership. Control of a legislative chamber is never absolute.
Also, I think Trump should expect significant foot-dragging and passive resistance within the bureaucracy. Having seen a good deal of that last time, he should prioritize ramping up appointments at all levels quickly. In foreign policy, new laws are less important than presidential actions, so I think we might see the most dramatic changes in that realm.