Keeping Polls in Perspective

The latest numbers suggest that Trump’s convention bounce has dissipated, and Hillary is enjoying hers. Most polls released this week put her up by 5 to 9 points. Crucially, though, her favorability ratings and voters’ opinion of her truthfulness and trustworthiness have not improved significantly, suggesting that her overall numbers will fade again soon.

But before we get too excited over the latest ups and downs, here are a few things to keep in mind.

Tune out until mid-September. The closer we get to the election, the more accurate polls are likely to be. Right now, the vast majority of Americans are paying little or no attention to politics. Consider that 34.9 million Americans watched Donald Trump’s acceptance speech, and 33.3 million watched Hillary’s. There was undoubtedly a huge overlap. But even if there was no overlap, that means half of those who will vote this November did not watch either one. And we are about to head into the Olympics, which will dominate the news and drown out the candidates’ messages for the next three weeks. Positions may be more hardened this year, because both candidates are so well known and so polarizing. Their partisans are unlikely to change their minds. But for the rest of the electorate, there’s a lot yet to come: the debates, non-stop advertising, world events. Who knows what Trump might say next, or whether there’s another scandal awaiting Hillary?

Is it a two-way, or a three– (or four-) way race? Some polls offer Trump and Clinton as the only options, forcing the binary choice. But when voters actually enter the voting booth, they will have other choices. Libertarian nominee Gary Johnson will be on the ballot in all 50 states, and the Green Party’s Jill Stein is already on the ballot in 23 states and D.C. and has filed in 8 more. Polls since the convention have generally shown Johnson taking 8 to 12 percent of the vote in a three-way race. Traditionally, third-party candidates have faded as the election gets closer, but, as we have seen, this is far from a traditional year, and those numbers have held remarkably firm so far. Given that Johnson is still largely unknown, a great deal of his support may just be a proxy for none of the above. As such, he seems to be taking support about equally from both Trump and Clinton. But what happens if he picks up a couple of points and gets into the debates, or attracts enough funding to make a significant advertising effort? Would that change the campaign dynamics? Keep an eye out.

While most polls are based on national samples, we don’t elect a president through a national vote, but through 50 state contests.

National vs. state polls: While most polls are based on national samples, we don’t elect a president through a national vote, but through 50 state contests. If Hillary is running up her numbers in California, or Trump is winning big in Texas, that can distort the results. One can overstate this, of course, since 2000 was the first election since 1888 in which the winner of the electoral vote lost the popular vote. Still, there are bellwether states that can tell us a great deal about the election outcome. How is Trump performing in states like Pennsylvania and Ohio? Is Clinton outperforming in states like Arizona and North Carolina? Who has the advantage in Florida? And it is in state-by-state races that third-party candidates can make a big difference. Could Johnson, or Stein for that matter, tip a state one way or the other? Remember, it was likely a combination of Ralph Nader and Pat Buchanan that gave Florida to Bush in 2000. Is it possible that a candidate like Johnson could even win a state in the libertarian West?

Don’t cherry-pick your favorite poll. Confirmation bias is never stronger than when we look at polls. We can always find a poll that looks good for the candidate we support. And we all know that certain polls skew Republican, while others skew Democratic (usually by 2 or 3 points). There will undoubtedly be outliers leading to excited headlines about a “shock poll.” And, yes, the pollsters have certainly gotten it wrong more than once in recent years. Still, if you look at the aggregate, such as the RealClearPolitics average, you can get a pretty good idea of where the race is headed.

Watch the internals. The Trump campaign is based on a gamble. He expects to lose overwhelmingly with women and minorities. Some recent polls show him receiving as little as 1 percent of the African-American vote, and less than 20 percent of the Latino vote. He is also losing college-educated whites, a group that Republicans have carried since modern polling began in 1952. He hopes that he can offset those losses by driving up both the turnout and his margin among working-class, non-college-educated whites. Mitt Romney won 61 percent of those voters in 2012. Trump will have to do quite a bit better than that.

Like a car accident along the highway, the latest poll always grabs our attention. And Republicans should have learned from 2012 that you ignore them at your peril. But before we get too excited by the next tick up or down, everyone should take a deep breath and keep it all in perspective.