In the seven battleground states Donald Trump and Kamala Harris are polling within a percentage point or two. Photos / Doug Mills and Jamie Kelter Davis, The New York Times
In the seven battleground states Donald Trump and Kamala Harris are polling within a percentage point or two. Photos / Doug Mills and Jamie Kelter Davis, The New York Times
There’s growing fears among Democrats that the race is slipping away from Kamala Harris.
Seven battleground states are all polling within a percentage point or two.
Nate Silver, the founder and former editor of FiveThirtyEight and the author of On the Edge: The Art of Risking Everything, writes the newsletter Silver Bulletin.
OPINION
A 50-50 forecast really does mean 50-50. And you should be open to the possibility that those forecasts are wrong.
In an election where the seven battleground states are all polling within a percentage point or two, 50-50 is the only responsible forecast. Since the debate between Kamala Harris and Donald Trump, that is more or less exactly where my model has had it.
Yet when I deliver this unsatisfying news, I inevitably get a question: “C’mon, Nate, what’s your gut say?”
So OK, I’ll tell you. My gut says Donald Trump. And my guess is that it is true for many anxious Democrats.
But I don’t think you should put any value whatsoever on anyone’s gut – including mine. Instead, you should resign yourself to the fact that a 50-50 forecast really does mean 50-50. And you should be open to the possibility that those forecasts are wrong, and that could be the case equally in the direction of Trump or Harris.
It’s not that I’m inherently against intuition. In poker, for example, it plays a large role. Most of the expert players I have spoken with over the years will say it gives you a little something extra. You’re never certain, but your intuition might tilt the odds to 60-40 in your favour by picking up patterns of when a competitor is bluffing.
But poker players base that little something on thousands of hands of experience. There are presidential elections only every four years. When asked who will win, most people say Trump because of recency bias – he won in 2016, when he wasn’t expected to, and then almost won in 2020 despite being well behind in the polls. But we might not remember 2012, when Barack Obama not only won but beat his polls. It’s extremely hard to predict the direction of polling errors.
Why Trump could beat his polls
The people whose gut tells them Trump will win frequently invoke the notion of “shy Trump voters”. The theory, adopted from the term “shy Tories” for the tendency of British polls to underestimate Conservatives, is that people do not want to admit to voting for conservative parties because of the social stigma attached to them.
But there’s not much evidence for the shy-voter theory – nor has there been any persistent tendency in elections worldwide for right-wing parties to outperform their polls. (Case in point: Marine Le Pen’s National Rally party underachieved its polls in this summer’s French legislative elections.) There’s even a certain snobbery to the theory. Many people are proud to admit their support for Trump, and if anything, there’s less stigma to voting for him than ever.
Instead, the likely problem is what pollsters call nonresponse bias. It’s not that Trump voters are lying to pollsters; it’s that in 2016 and 2020, pollsters weren’t reaching enough of them.
Trump during a campaign rally in Uniondale, New York in September. Photo / Doug Mills, The New York Times
Nonresponse bias can be a hard problem to solve. Response rates to even the best telephone polls are in the single digits – in some sense, the people who choose to respond to polls are unusual. Trump supporters often have lower civic engagement and social trust, so they can be less inclined to complete a survey from a news organisation. Pollsters are attempting to correct for this problem with increasingly aggressive data-massaging techniques, like weighing by educational attainment (college-educated voters are more likely to respond to surveys) or even by how people say they voted in the past. There’s no guarantee any of this will work.
If Trump does beat his polling, there will have been at least one clear sign of it: Democrats no longer have a consistent edge in party identification – about as many people now identify as Republicans.
There’s also the fact that Harris is running to become the first female president and the second Black one. The so-called Bradley effect – named after former Los Angeles Mayor Tom Bradley, who underperformed his polls in the 1982 California governor’s race, for the supposed tendency of voters to say they’re undecided rather than admit they won’t vote for a Black candidate – wasn’t a problem for Barack Obama in 2008 or 2012. Still, the only other time a woman was her party’s nominee, undecided voters tilted heavily against her. So perhaps Harris should have some concerns about a “Hillary effect”.
A surprise in polling that underestimates Harris isn’t necessarily less likely than one for Trump. On average, polls miss by three or four points. If Harris does that, she will win by the largest margin in both the popular vote and the Electoral College since Obama in 2008.
How might that happen? It could be because of something like what happened in Britain in 2017, related to the “shy Tories” theory. Expected to be a Tory sweep, the election instead resulted in Conservatives losing their majority. There was a lot of disagreement among pollsters, and some did nail the outcome. But others made the mistake of not trusting their data, making ad hoc adjustments after years of being worried about “shy Tories”.
Polls are increasingly like mini-models, with pollsters facing many decision points about how to translate non-representative raw data into an accurate representation of the electorate. If pollsters are terrified of missing low on Trump again, they may consciously or unconsciously make assumptions that favour him.
Harris makes a campaign appearance in Redford Township, Michigan, earlier this month. Photo / Valaurian Waller, The New York Times
For instance, the new techniques that pollsters are applying could be overkill. One problem with using one of those – “weighting on recalled vote,” or trying to account for how voters report their pick in the last election – is that people often misremember or misstate whom they voted for and are more likely to say they voted for the winner (in 2020, Biden).
That could plausibly bias the polls against Harris because people who say they voted for Biden but actually voted for Trump will get flagged as new Trump voters when they aren’t. There’s also a credible case that 2020 polling errors were partly because of Covid restrictions: Democrats were more likely to stay at home and therefore had more time on their hands to answer phone calls. If pollsters are correcting for what was a once-in-a-century occurrence, they may be overdoing it this time.
Last, there is Democrats’ persistently strong performance over the past two years – since the Supreme Court overturned Roe v. Wade – in special elections, ballot referendums and the 2022 midterms. Democrats shouldn’t hang their hopes on this one: high-quality surveys like the New York Times/Siena College polls can replicate these results by showing Democrats polling strongly among the most motivated voters who show up in these low-turnout elections – but Trump making up for it by winning most of the marginal voters. So Democrats may be rooting for lower turnout. If those marginal voters don’t show up, Harris could overperform; if they do, Trump could.
Or maybe pollsters are herding toward a false consensus
Here’s another counterintuitive finding: it’s surprisingly likely that the election won’t be a photo finish.
With polling averages so close, even a small systematic polling error like the one the industry experienced in 2016 or 2020 could produce a comfortable Electoral College victory for Harris or Trump. According to my model, there’s about a 60% chance that one candidate will sweep at least six of seven battleground states.
Polling firms are pilloried on social media whenever they publish a result deemed an “outlier” – so most of them don’t, instead herding toward a consensus and matching what polling averages (and people’s instincts) show. The Times/Siena polls are one of the few regular exceptions, and they depict a much different electorate than others, with Trump making significant gains with Black and Hispanic voters but lagging in the blue-wall states of Michigan, Wisconsin and Pennsylvania.
Don’t be surprised if a relatively decisive win for one of the candidates is in the cards – or if there are bigger shifts from 2020 than most people’s guts might tell them.
