Over the past few weeks, I’ve looked at scenarios and non-polling indicators that, if Donald Trump wins, we might look back upon and say, “Aha, the signs were there all along!” These are examples of “what if the polls are wrong again” pieces that proliferate on the Internet.
There are two reasons why there are so many of them. First, after 2016, there’s an understandable aversion to understating Trump’s chances. Second, these analyses are simply more interesting to write. “Joe Biden will win because he is up nine points in the polls” just doesn’t intrigue anyone.
What follows, though, are the reasons to expect a potential a runaway win for the challenger.
In the long run, polls are unbiased.
In 2014, Nate Silver released his database of polls that had appeared in the closing days of an election. Overall, the data set included 6,614 polls taken between 1998 and 2014. There were lots of interesting takeaways from the data, but perhaps the most interesting one is this: Polls are largely unbiased over time.
If we compute the bias of each poll – that is, the amount that it deviated from the actual result, and then average the bias, the overall tilt for polls conducted during this time period is about 0.7 points toward the Democrats. That’s not zero, but it isn’t enough to appreciably alter the outcome of the election either.
Of course, in a given year polls do have systemic biases to them; we saw this in 2014, 2016 and 2018. Maybe we can predict from them?
Poll errors are unpredictable.
In 2014, some analysts thought that perhaps we should expect polls to be biased against Democrats. After all, the polls had, overall, over-predicted Republican performance in 2010 and 2012.
As it turned out, looking at the broader historical context, it was very difficult to predict a poll error in a certain year given the poll error in the year before. As I wrote at the time:
“I mean it in a more general sense. Humans are remarkably adept at discovering and using patterns. We don’t like chaos, and this is part of what has allowed us to advance as a species. Yet our minds aren’t precisely fine-tuned to patterns; we’re overly sensitive, and so we see dragons in clouds, a man’s face on the moon, and images of Mary in a grilled cheese sandwich. If I gave you a page with 15 dots and challenged you to fill in the gaps with what you saw, you’d probably come back with a picture of a Dimetrodon (or at least, that’s what I’d be inclined to draw) or some such; you wouldn’t likely return the page and tell me it is just random noise.
“We do the same thing with data. We do it in very obviously bad ways -- there was a cottage industry of predicting presidential elections based on the winner of the final Redskins football game from 1932 to 2004 (there’s actually a statistically significant correlation between the margins of those games and the margin of presidential elections during this time). But where it’s most dangerous is when we have good reason for believing that there has to be a pattern.”
In other words, it is certainly interesting that we’ve seen the sorts of errors that have occurred the past few cycles, especially when they’ve been concentrated in a key swing region such as the Midwest. It should make us a bit nervous. But we shouldn’t *bet* on such an error repeating itself.
Trump has alienated too many groups.
When I wrote my 2013 series on demographic shifts, including the Case of the Missing White Voters, I outlined a potential path to electoral victory for Republicans that didn’t involve reaching out to non-white Americans as such. Rather, I suggested that a more economically populist Republican Party that was more skeptical of American intervention abroad, aligned against illegal immigration and skeptical of trade deals -- and that, above all else, didn’t nominate a guy with car elevators as its standard-bearer -- might be able to gain the enthusiastic backing of enough blue-collar whites to win elections. This incarnation of the GOP might eventually win over some non-white voters, who vote for Democrats more because of their stances on economic issues than their stances on identity issues.
This is more-or-less the path that Trump took. Indeed, he proved a stronger version of the “Missing Whites” approach than I had thought was possible, by alienating large numbers of whites with college degrees, who had previously been the foundation of the Republican Party.
But there are limits to what can be done with whites without college degrees, who constitute a significant portion of the electorate, but not a majority. While Trump has, in fact, made progress with non-whites, his bleeding of support among whites with college degrees, especially women, and (at least according to polls) older white voters more than offsets that.
In other words, at a certain point you just run out of groups that you can afford to alienate from your coalition, and Trump may well have hit that point.
This time really might be different.
As I’m fond of saying, “ ‘This time is different’ has a poor track record.” That doesn’t mean, however, that it never bears out. Indeed, I can tell stories about why this time may be different.
For example, I take it as almost an article of faith that presidents with 45% job approvals don’t lose landslide elections. Generally speaking, I’d expect a president in that position to get about 46% of the vote, with about 2% going to third parties. That would still be a loss for President Trump, but in the Electoral College it would be close, and Republicans might even be expected to keep the Senate.
But you can tell the story of why this year might genuinely be different as well! In particular, you can see how there may be many Republican-leaning voters who genuinely like his policies but dislike him personally on such a visceral level that they cannot bring themselves to vote for him.
Likewise, presidents who lead on the question of “Whom do you trust more to handle the economy?” tend to fare well at the polls. “It’s the economy, stupid,” is almost a trope in politics. But with COVID-19 dominating the headlines, you can imagine why this time may, indeed, be different.
And so it goes. I do think it is wise to keep an eye on a variety of polling metrics, but not to be too wedded to any one of them. In a strange year like 2020, it seems doubly important to do so..