The Polls Aren’t Working Against Donald Trump

As this election cycle enters its waning moments, there’s been a lot of talk from Donald Trump about a “silent majority” of voters who secretly support him but are too afraid to say it in public or over the phone to pollsters. The term comes from one of Nixon’s 1969 speeches, and it’s been used unsparingly ever since by candidates seeking to shed light on a section of the electorate that they feel isn’t being fully illustrated by polling data. Recently, Trump has latched onto the idea, arguing that many of his supporters will vote their conscience for his ideas in the privacy of the voting booth, propelling him into office.

There’s not a lot of evidence to support that assertion.

Presidential polling has been something of a mixed bag this cycle, but it’s actually been pretty accurate at capturing the general trend of which candidate has been the most competitive. There’s an unfortunate reality in that the data stems almost entirely from the primaries, since few people have actually voted in the general election so far. Admittedly, there are a couple of issues with that – namely that primary voters, which made up only 14 percent of the eligible voting population, don’t necessarily reflect the national electorate overall. It’s an imperfect comparison, and while primary voters alone certainly can’t be substituted for the general population, there are still important lessons that can be learned from the data that they provided. Any argument by Trump that polling itself is inherently inefficient, wrong, or unable to accurately capture trends isn’t supported by much evidence. Polling has mostly moved in line (even if it was by a small margin) with important events like the reveal of Trump’s comments about sexual assault or FBI director Comey’s comments on Clinton’s ongoing email investigation.  

Either way, Trump’s assertion that the polling so far is somehow not painting the real picture should be debunked by its efficiency during the primaries. There were a couple of upsets – like Sanders’ win in Michigan – but overall, in states that had heavy polling and a diversified polling average, the margin of error was typically, at most, within a few percentage points. In fact, during the early Republican primaries, Trump’s support was often overstated – that eventually changed (albeit only slightly) as he picked up steam, but it remains a troubling sign for somebody who’s fundamental path to victory lies in his ability to create a swell of voters unseen by any major polls. So far, it’s just not happening.

In Iowa, the polling average had Trump winning by 4.7 percentage points; the actual result was a Cruz victory by a 3.3 percentage point margin. Similarly, in South Carolina, a 13 point lead for Trump in the polls slimmed down to a 10 point win when the actual votes were cast. In Florida, which was well-polled in advance of the primary, Trump’s lead was just barely larger than what the polling average had suggested – a difference of half a percentage point.

For him to suggest that polling isn’t accurately capturing his base of support is just conjecture at this point, and there’s no evidence from the primaries or anywhere else that he’s being underestimated. Just a few days ago, a specialized study done by Politico found that anonymous polling for Trump supporters had no significant difference from polling in which his voters interacted with interviewers over the phone.

If you look at the demographics of Trump’s support structure, the vast majority of it comes from uneducated white males and a few other subsets – and there likely aren’t enough of them to precipitate the win that he’s arguing will happen. Clinton’s overwhelming majorities among women, minorities, the educated, and the wealthy lend at least partial validation to her numbers in the polling, since together they make up a fairly large part of the voting population. In other words, her numbers make sense – they’re backed up by data. Trump doesn’t have that luxury.

There are certain maps ― like Trump winning in a plethora of battleground states, and even a scenario where the House of Representatives ends up electing Trump instead of the general population ― that might lead him to the White House. But these scenarios are unlikely. The polls have mostly stopped moving after Comey’s announcement last week, and so far, it looks like the damage has been mostly controlled. Clinton’s maintained her modest lead, and a number of different forecasts (The New York Times currently has her at an 84 percent chance of winning; FiveThirtyEight’s has her at 70 percent) still give her great odds.

And in a last bright spot for Clinton, news broke on Sunday that Comey was again closing the investigation and that he would stick by his decision not to bring charges. This could give her a (probably small) bump tomorrow, but we probably won’t ever know the full effects since few major polls will have time to capture it.

There have been upsets in presidential elections before, and it would be premature to say that the race is, by any means, over; but at this point, Trump’s argument for the existence of some hidden “silent majority” remains purely theoretical.