Confused by Contradictory Polls? Take a Step Back

http://www.nytimes.com/2016/07/18/upshot/confused-by-contradictory-polls-take-a-step-back.html

Version 0 of 1.

It has been a rocky few weeks all over the world. Only a few surprising presidential poll results have broken through in a news cycle dominated by shootings by and of the police, the F.B.I.’s decision not to indict Hillary Clinton, Donald Trump’s vice-presidential pick, more terror in France and an attempted coup in Turkey.

In truth, around 50 state and national polls were released this past week. Individually, they have sometimes felt as chaotic as the news around the world. But together they tell a far clearer story: Mrs. Clinton holds a modest but clear lead heading into the conventions.

While the occasional poll has looked good for Mr. Trump, it has been two months since he led a national survey that included voters without a landline telephone. Mrs. Clinton has led dozens.

Her lead is smaller than it was last month, before the director of the F.B.I. said he would not indict Mrs. Clinton, though he excoriated her for using a private email server as secretary of state. But she retains an advantage — perhaps by 4 percentage points nationwide, and a similar margin in the battleground states that are likely to award the electoral votes needed to win the presidency.

This straightforward story can get lost in the headlines, which tend to give the most attention to the most surprising results — whether it’s a predicted Clinton landslide or a narrow lead for Mr. Trump in key states.

The truth is probably somewhere between those extremes. Pollsters aren’t joking about the “margin of error”: the inevitable random variance in polls that exists simply by chance. If Mrs. Clinton leads by 4 points, you should expect polls that show her with a big lead or locked in a tight race, with others clustered around the average. That’s more or less what we saw this past week.

It’s a lot like baseball. Even great baseball players go 0 for 4 in a game — or have rough stretches for weeks on end. On the other end might be a few multihit nights with extra-base hits, or a spectacular few weeks.

Sometimes, these rough stretches or hot streaks really do indicate changes in the underlying ability of a player. More often, they are just part of the noise inevitable with small samples. Taking more polls is like watching more at-bats, and you need many if you want to be confident about whether a candidate is ahead or tied.

The “margin of error” doesn’t even include the various distinctions among pollsters, like whether a pollster screens for likely voters, calls cellphones, conducts polls online, or makes many other choices hard to discern from opaque public releases.

For whatever reason, the Quinnipiac polls consistently show Mr. Trump doing better than NBC/Marist polls do, even though they appear fundamentally similar. Add these differences on top of the inevitable random error and it’s surprising that there haven’t been more polls showing Mr. Trump ahead.

The polls tell a very clear story about the country’s divisions in an era of sweeping economic and demographic shifts. For white voters with a college degree and nonwhite voters, the 2016 presidential election must look and feel like a landslide. Mr. Trump trails by as large or larger margins among these voters as John McCain did in 2008.

But the story is very different for white voters without a college degree, who remain a very large bloc of voters in the electorate. Here, Mrs. Clinton is doing far worse than President Obama last time. On balance, these two shifts have canceled out — leaving Mrs. Clinton ahead by roughly the same margin as Mr. Obama was in pre-election polls from 2012.

These big demographic divides have added confusion to the polls in battleground states. Those states have such different demographic characteristics that Mrs. Clinton has seemed to fare very well in some states while struggling in others.

On balance, Mrs. Clinton is excelling in diverse and well-educated states like Virginia, Colorado and even North Carolina. But she is struggling to match Mr. Obama in less educated or less diverse places in the Midwest and Northeast, like Iowa, Ohio, New Hampshire, or even Maine’s Second Congressional District. Yet Mrs. Clinton appears to be tied or ahead even in these states. The combination gives her a strong advantage in the Electoral College.

The volatility could grow this week and next week, when the two parties hold their conventions. Typically, the party convention helps its candidate by a few percentage points in the polls, and sometimes more. The bounce typically fades in the few weeks that follow, but sometimes it doesn’t, permanently altering the race.

In general, the polls are far more predictive of the final results a few weeks after the conventions than they are immediately before. No modern presidential candidate who trailed in the polls several weeks after the convention has come back to win the popular vote. This isn’t to say it couldn’t happen, especially if the post-convention polls were fairly close. There have been instances in which the polls have moved considerably after the conventions, even if they didn’t contradict the result of the popular vote (as in 1976 and 1980).

But it does indicate that a post-convention lead is quite meaningful, even with weeks or months to go until the election. By that time, both candidates will have had many and fairly equal opportunities to unify their parties and make their cases to the broader electorate. If a candidate holds a clear lead at that point, it means a lot.