Are the Polls Frightening You? Read This!

Welcome to How Good Is This, Really?—renamed from the original How Bad Is This, Really?—a recurring feature in which we take the temperature of the presidential election and what seems likely to happen in November.

Ahhhhhhhhhh!

Ahhh! Ahhhhhh!

Those, of course, are the seasonally appropriate screams of Democrats who have observed, with Halloween and Election Day both approaching, that the polls are a-tightenin’. Or are they? There are a lot of signals out there in polling world, and for any of them that point a hopeful way for either candidate, there’s another that points … toward doom. Let’s review!

Harris’ margins have gotten smaller … but it might not mean a whole lot because of a lack of high-quality data.

Kamala Harris’ lead in FiveThirtyEight’s national polling average has fallen by about a point this month, to just under 2 percent—which might not be enough to win the Electoral College, given that it disproportionately favors the kind of small, rural states where Republicans currently dominate. Harris’ averages in swing states are dipping, too. She only leads the New York Times’ averages in four of the seven swing states (Michigan, Nevada, Pennsylvania, and Wisconsin) by less than a point, which is preposterously close.

On the other hand, a lot of the polls that have been released in the past few weeks were conducted by smaller and historically less accurate pollsters. Greg Sargent of the New Republic points out that the Washington Post’s poll tracker for this cycle, which only uses the work of pollsters with the most established track records, still has Harris’ national and state leads at about the same place as they were a few weeks ago. Which is a slightly more comfortable position for her, but still means that margins are close enough that the race could easily go either way.

Some of the early voting returns are tilted toward Republicans … but it might not mean anything because of COVID and changes to GOP strategy.

When someone submits their ballot before Election Day—by mail, or via early in-person voting, which is getting underway in earnest across the country—that information gets logged by the state in which they voted. And while we can’t see who early voters voted for, we can see where they voted and who they are in terms of age, gender, and party registration.

In Nevada, journalist Jon Ralston is known for making accurate predictions about whether the state is swinging right or left in a given cycle based on this data about who’s voting early. In 2024, he reports, early ballots are much more heavily Republican than they’ve been in years past. And if Democrats are losing Nevada, where polling has generally shown a 50-50 dead heat, they’re probably losing the election overall.

However, it’s entirely possible that Republicans have simply become more likely to embrace early voting than they were in past elections, particularly 2020’s. During that cycle, Democrats tended to be more concerned than Republicans about COVID transmission in public places; Republican leaders like Donald Trump, meanwhile, claimed mail-in voting was a vector for fraud. But the GOP, having decided that making it easier to vote is actually in its interest these days thanks to Trump’s appeal to Americans who are otherwise politically disengaged, has started to encourage early voting in states like Nevada. While COVID is still around, most Democrats don’t consider it a major enough threat to avoid public places, which means some of them might be going back to in-person voting. The point is: Be skeptical of anyone who seems certain that early numbers for either party suggest a wave in its direction given how much turnout-timing patterns are probably in flux.

Harris is still doing better on the economics question … but maybe not in the right places.

Earlier this month, we noted that Harris had pulled even with Trump on poll questions about which candidate was better prepared to handle the economy and deal with the “cost of living.” We thought this might lead to an overall polling advantage, given that one of Harris’ initial weaknesses as a candidate had been voters blaming the Biden administration for inflation. As discussed above, that bump hasn’t materialized (yet?). But a new AP/NORC poll does find that Harris has sustained her improvement on the economy: The poll pegs her with slight advantages over Trump on the question of who will better handle “jobs and unemployment” and “the cost of housing” as well as a 12-point advantage on the issue of “taxes on the middle class.” (Presumably, the middle class is mostly against middle-class taxes.)

On the other hand, the New York Times this week dug into a recent poll that it conducted which found Harris narrowly leading Trump nationally on the issue of which candidate voters felt would most help them “personally,” which is another sort of “kitchen-table” question. The Times noticed that Harris trailed Trump on the question in the five swing states it surveyed, by 6 points. So maybe she hasn’t caught up in the places where voters are being most inundated with advertising and news about the race.

Maybe the polls are still failing to find Trump supporters … or maybe they’re failing to find Democrats this time?

In the Times on Tuesday, Senior Polling Man Nate Cohn laid out a worst-case scenario for Democrats: that Donald Trump’s singular personality will once again draw out voters who don’t otherwise show up for elections and rarely respond to polls, which is what seems to have happened in 2016 and might have happened in 2020. As Cohn writes, pollsters have tried to adjust to avoid getting surprised by this again—but there’s no way of telling whether they’ve succeeded until the actual votes get counted.

On the other hand, Cohn says, it’s possible that polls underestimated Trump in 2020 for a different reason, like that Democrats were overrepresented in 2020 surveys because they were more likely to be at home, picking up random phone calls, during COVID. As some data enthusiasts who are a part of “polling Twitter” have been pointing out, moreover, Democrats seem to have become less likely than Republicans to answer pollsters during the current cycle for reasons that no one is sure about.

This could mean that Dems are now the ones now being underestimated, a theory that is lent support by the results of August’s Washington state “jungle” primaries. In Washington, Democratic and Republican primary candidates appear on the same ballots and the top two vote-getters move on to the general election. This structure means supporters of both parties are incentivized to turn out at high rates to make sure their party doesn’t get shut out of the November runoff. As such, the Washington primary has historically been useful for predicting what the national House of Representatives vote will look like once you account for the difference between Washington, which is a heavily blue state, and the country at large.

This year, Washington’s results suggested that Democrats will win the national congressional vote by around 3 points—more than the 1-point win polls currently predict, and akin to how the party performed in 2020 when it won both the House and the presidency. Democrats similarly did well in the special election this year that was held to replace Republican Rep. George Santos in New York.

In sum, cases could be made either way for a lot of things here, huh? How interesting and exciting, for all of us.

We conclude this column with a rating on the Shovel Meter, a measure of exactly how sedated you might want to be, on a scale of one to five shovel blows to the head, if you’re concerned about Trump’s reelection.

This week: Four shovels, but with two hitting you on either side of the head, for balance.

Ahhhh! Ghost shovels!