ANALYSIS — With the convention season ending and the sprint to November about to begin, it might be useful to look once again at the 2016 presidential exit poll. The survey offers a window into that year’s electorate and will allow us to compare where we are now and where we are headed over the next two months.

First, a couple of caveats. The exit poll is a poll, which means it has a margin of error. Don’t treat the numbers as if they were given to Moses on Mount Sinai.

Second, the exit poll is a national survey, which is why it is called a “national” exit poll. As we all know, electors, chosen by each state, pick the president, so a national exit poll is merely a way to measure national public sentiment. Please don’t ignore the key states.

As you watch the presidential contest and hear about the latest poll, keep in mind what happened in 2016 — how groups turned out and how they voted. Hillary Clinton won the popular vote by 2.1 points four years ago, according to tallies of votes cast. Is Trump doing better or worse than he did four years ago?

White voters

White voters went for Donald Trump by 20 points four years ago, 57 percent to 37 percent. They also accounted for 71 percent of exit poll respondents.

So if the electorate is less white than it was four years ago, or if Trump falls well below his 20-point margin among whites, the president is in deep trouble.

Sure, he could improve his showing among Blacks (8 percent of whom backed him in 2016, according to the exit poll), Latinos (28 percent) or Asian Americans (27 percent). But that is unlikely. Whites remain his base, and he cannot afford significant defections from that group.

[jwp-video n=”1″]

How important are white voters to Trump?

In a June 7, 2016, article for The Washington Post (“Will there be enough white voters to elect Donald Trump?”), I noted that two national surveys offered vastly different views on the question.

A May 16-19, 2016, ABC News/Washington Post survey found Trump ahead of Clinton by 2 points, primarily because he led her by 24 points (57 percent to 33 percent) among whites. A May 13-17, 2016, CBS News/New York Times poll showed Clinton leading by 6 points, in part because Trump held a narrower 12-point lead (50 percent to 38 percent) among whites.

So if you can figure out the percentage of whites in the electorate and get Trump’s margin with that group correctly, you probably know the outcome of the 2020 election.

Many analysts are looking at smaller demographic categories, such as white women with a college degree or white men without a college degree, which could offer interesting insights into the electorate. But smaller subsamples can be tricky.

For example, the Sept. 11-14, 2016, Fox News poll interviewed 1,006 registered voters and 867 likely voters. According to information supplied by the survey, the margin of error for likely voters for whites was 3.5 points, for whites with a college degree 4.5 points, and for whites with no college degree 8 points. The smaller the sample, the larger the margin for error.

College graduates

According to the 2016 national exit poll, half of respondents were college graduates and half were not.

Clinton won college graduates by 10 points, 52 percent to 42 percent, while Trump won voters without a college degree by 7 points, 51 percent to 44 percent.

Any improvement in Biden’s performance among college graduates should raise alarms in the Trump campaign. On the other hand, an electorate with a much larger percentage of non-college graduates would make Democrats nervous.

White evangelicals

White evangelicals have been among the president’s strongest supporters. They accounted for 26 percent of exit poll voters in 2016, and Trump carried them by a stunning 64 points — 80 percent to 16 percent.

Trump should again carry the group overwhelmingly, but any defections to Biden will come right out of the GOP/Trump base. Biden has talked a great deal about faith and character, so it will be interesting if Trump suffers any weakness with the group.

Suburban/rural/urban

Suburban voters constituted almost half of the 2016 exit poll respondents (49 percent), and the category is widely viewed as an important swing group. After all, Trump won suburban voters by only 4 points in 2016, 49 percent to 45 percent, and suburban voters split evenly (at 49 percent each) in the 2018 midterm exit poll. That change undoubtedly helped Democrats make significant gains in the House.

But small changes from 2016 to 2018 in both rural areas and urban America could also benefit one party or the other.

Clinton won urban voters 60 percent to 34 percent in 2016, but two years later, the exit poll showed Democrats improving on that performance by winning 65 percent of urban voters.

And while Trump carried rural voters handily in 2016 (61 percent to 34 percent), the GOP margin slipped to 56 percent to 42 percent in 2018.

How the suburbs vote — and whether urban or rural voters are changing their votes — should tell us something important about this November’s elections.

You will see lots of polls with ballot tests and subsamples over the next two months. Keep your eye on the quality national polls and the demographic categories that can tell you how voting groups have changed (if they have changed at all) since 2016.

And don’t forget about those state polls, which can be equally helpful if they present accurate snapshots of the races in states such as Pennsylvania, Michigan, Wisconsin, Florida, Arizona and Minnesota.