Donald Trump’s chances, write Rothenberg, probably depend on his winning white voters by a landslide. (Photo by Spencer Platt/Getty Images)

Race isn’t the only demographic factor influencing the outcome of the 2016 presidential election. But, along with gender and education, it’s likely to be a powerful way to understand the contours of the electorate and the overall contest.

For that reason, the most important question of 2016 is: Can Donald Trump turn out enough white voters to beat Hillary Clinton?

In every presidential election since 1968, the Republican nominee has carried whites, sometimes with a plurality, sometimes narrowly, but increasingly by a comfortable margin. (see 1976-2012 exit poll data from the Roper Center for Public Opinion Research at Cornell University; and 1968 and 1972 data from Gallup’s final pre-election surveys.)

Ronald Reagan won white voters by 20 points (56 to 36 percent) in 1980 and by over 30 points (66 to 34 percent) four years later. George H.W. Bush carried white voters by 20 points against Michael Dukakis in 1988.

Bush and Bob Dole each carried white voters narrowly in their losing three-way races against Bill Clinton and Ross Perot in 1992 and 1996, but George W. Bush carried white voters comfortably, 55 to 42 percent, in 2000. Four years later, Bush improved his showing among whites, carrying them by 17 points.

In 2008, John McCain carried white voters by a dozen points, 55 to 43 percent, while Mitt Romney did much better, winning the group by 20 points (59 to 39 percent).

[GOP worries rise amid hostile Trump comments on Latinos, Muslims]

Romney’s margin with whites was the same as Reagan’s in 1980 and Bush’s in 1988. Of course, Reagan and Bush won those races, while Romney lost by almost 4 percentage points. That is all you need to know about the changing nature of the American electorate.

Whites constituted 88 percent of the electorate in 1980, 85 percent in 1988 and 81 percent in 2000. But they were just 72 percent of all voters in 2012. If the recent historical pattern holds, they will account for no more than seven in ten November voters.

What does this mean for November?

Right now, major national media polls show very different snapshots of the general election. The May 16-19, 2016 Washington Post/ABC News survey, showed Trump leading Clinton by 2 points, while the May 13-17 CBS News/New York Times survey found Clinton ahead by 6 points.

Part of the reason for that difference is that the Washington Post/ABC News survey found Trump leading Clinton among whites by 24 points, 57 to 33 percent, a wider margin than Reagan or Romney had in their competitive contests.

On the other hand, the CBS News/New York Times survey showed Trump holding a much narrower 12 point- lead, 50 to 38 percent, among whites.

Most national polls show a competitive contest, but Clinton’s narrow lead is likely to grow after Bernie Sanders exits the race and his supporters get behind her. Given that, how can Trump improve his odds? There are a few ways.

First, he could strongly outperform Romney and McCain among African-Americans, Latinos and Asian-Americans. But that’s unlikely given Trump’s non-inclusive rhetoric so far, as well as the nature of his supporters. Gallup found that 77 percent of Latinos have an unfavorable view of Trump, and the wealthy businessman doesn’t seem like a natural fit for black voters or Asians.

Alternatively, he could win landslide levels of white voters, as Reagan and George W. Bush did in their re-election victories. Given the 2012 electorate, the polarized political environment, and Trump’s own problems with some GOP voters, that seems unlikely. He will win whites comfortably but almost certainly not by 30 points.

The Republican’s chances would improve if black voters and younger whites don’t turn out as they did for Obama in 2008 and 2012. That is a possibility, given Obama’s unique appeal and Clinton’s problems with younger voters in the Democratic nominating contest. But it isn’t something that Trump can control.

Ultimately, the best chance for Trump – maybe the only chance, even with his current competitive standing – is to change the make-up of the electorate by bringing millions of new white voters to the polls and winning them by a large margin. That would change the election’s arithmetic.

Trump claims he is doing just that, and there is anecdotal evidence that he has energized some non-voters and former Democrats. But so far, it isn’t clear that he will fundamentally change the electorate.

The Pew Research Center has done plenty of work on non-voters, and some of them certainly fit the profile of Trump supporters. Non-voters are more male than female, and a plurality of them are political independents. They have less formal education and lower incomes.

But, as Pew wrote in 2014, “A much higher proportion of non-voters identify as racial or ethnic minorities than voters,” and Pew’s 2012 study of non-voters found that they were much more likely than voters to view President Obama favorably and to prefer Obama to Romney in the election. Non-voters also skew young, and younger voters have been a problem for the GOP recently, as they were for Trump in the GOP primaries.

So, while there certainly are alienated whites who will turn out for Trump in November, many of them already vote Republican regularly, and the number of new voters is uncertain. Just as important, there is an even larger pool of likely anti-Trump, less well-off and minority nonvoters that Democrats can target in the fall. (See Philip Bump’s piece in early May.)

Given what he has accomplished so far, it would be foolish to dismiss Trump’s ability to change the make-up of the electorate or even to win the presidential election. But he faces a difficult demographic challenge, since he needs to alter the trajectory of the last 50 years of American elections.

Clearly, the odds are against him.

Stuart Rothenberg writes about the politics of the presidential and congressional races.