Donald Trump won the Republican nomination for president in large part because of voters who have not graduated from college. In the Indiana primary, for example, Trump won 47 percent of the vote among college grads — but 60 percent among those who had not earned a bachelor’s degree.

Will he be able to rally these less educated voters to win the general election in November? Given his difficulties with other groups of voters, his campaign might view this as an appealing strategy.

But for Trump to win the election this way, he would have to close the U.S. turnout gap between the more and less educated that reappears in every election. And that will be very difficult.

Education now matters more

For years, political scientists have known about the voting gap between college grads and everyone else. In the 2012 presidential election, 58 percent of U.S. voters actually went to the polls to cast ballots. But that overall voter turnout rate disguises a lot of variation among those with differing levels of education. Only about 48 percent – or less than half — of voters who had only a high school degree cast ballots, while roughly 68 percent of voters who graduated from college went to the polls.

Few have noticed how much that the gap has increased – so much so that a college education has become an increasingly important dividing line between voters and not voters. As my research shows, that educational voting gap widened in the 1970s — and has never looked back. The graph below, drawn from data from the American National Election Study, shows voting rates in presidential elections for those who report that their highest educational attainment was either a college degree or a high school degree. I have adjusted the data to correct for overreporting.

That gap wasn’t very wide in the 1950s and 1960s. In 1968, for example, 73 percent of college grads reported voting while 66 percent of high school grads did. By the 1980s, the gap had grown to more than 20 percentage points. In 2012 turnout rates were 73 percent for college grads and 52 percent for those with only high school degrees.

The widening gap appears to be due mostly to declining turnout among high school graduates. Those who’ve gone to college but haven’t graduated vote at lower rates than do college grads — but their turnout hasn’t declined as sharply as those who’ve gone no further than high school. And the educational gap is even wider in midterm elections than in presidential elections.

Here’s why that gap is a problem

Research shows that political opinions differ a lot depending on how far people have gone in school. While the proportion of U.S. citizens who are college graduates has increased over that time, they’re still only about a third of the population. Which means that politicians are disproportionately hearing from, and elected by, people with college degrees – who do not represent the country.

It’s easy to imagine why college would affect voting. Education itself imparts political knowledge. It puts graduates in professions and social milieus that encourage political involvement. Finally, education is at least partly a proxy for demographic factors like family background that affect political participation.

Why did this change in the 1970s?

That isn’t yet clear. It probably has something to with the fact that the 26th amendment to the U.S. Constitution, lowering the voting age to 18, passed and was ratified in 1971. Not only did that constitutional change expand the electorate, but it did so by adding young people, the least likely age group to vote.

That was also the era when a larger share of young people began going to college in the first place. A few decades before that, the great majority of the electorate had no more than a high school degree. That percentage has steadily risen as more people have gone to college. And those who attend college may also be the people who are more likely to vote.

One intriguing possibility is that it’s not about college at all – but an indication that high school no longer does a good job in teaching students civics. Another possibility is that, as college degrees have become both more important and worth more in the job market, larger dynamics are at work that go beyond the political realm. In addition to the ways that education can open doors to social networks and provide direct skills, attending college is a marker of other job-related resources such as income and health, which are known to facilitate voting.

These dividing lines appear to have become more potent over time.

Here’s what that means for the 2016 elections

For Trump to win the presidency, he’d have to get a bigger share of Americans without college degrees to the polls. And that’s going to be hard. A number of overlapping factors hold them back.

Winning the presidency with non-college educated whites will be a long shot in 2016. Even if the Republicans pull it off this year, that will become more difficult as the years go on — because non-college grads are a declining share of the electorate. And neither party appears to have any idea about how to turn that around.

Barry Burden is professor of political science at the University of Wisconsin-Madison and director of the Elections Research Center. Find him on Twitter @bcburden.