One question in the tumultuous 2016 presidential campaign was whether white evangelicals would “come home” to the GOP and vote for Donald Trump, given his history of divorce, crude language and lack of familiarity with the Bible.

We now know from exit polls that they did — in droves. As shown in the graph below, Trump did better among white evangelicals (81 percent) than Mitt Romney in 2012 (78 percent) or even George W. Bush in 2004 (78 percent), and far better than John McCain in 2008 (74 percent). This is a critical constituency, as white evangelicals made up 26 percent of the electorate in 2016.

But perhaps an equally significant story line is the vote of nonreligious, or secular, Americans. In 2016, 15 percent of all voters did not identify with a religion, the so-called “nones.” In the overall population, an even higher percentage — one in five — are nones, and their numbers are growing. Traditionally, this group leans Democratic.

But Hillary Clinton’s 68 percent among secular Americans was lower than Barack Obama’s in 2012 (70 percent) and 2008 (75 percent). Her performance was similar to John Kerry’s 67 percent in 2004.

Why did Trump improve on past Republicans’ support among evangelicals while Clinton lost ground among secular voters? One clue comes from voters’ strength of commitment to their religious faith — or lack thereof.

Although we can’t use the exit polls to examine how evangelicals’ religious commitment affected support for Trump, we can draw on data from the 2016 American National Election Studies (ANES) Pilot Study. The survey was conducted during the primaries, but its questions about religious commitment, secular identity and feelings toward the candidates provide insight into how committed evangelicals and secularists reacted to Trump and Clinton.

For evangelicals, their level of religious commitment can be measured by their frequency of church attendance. Evangelicals who report attending church regularly are more committed than those who rarely attend.

Respondents to the pilot study were asked to rate Trump and his GOP competitors on a 100-point “feeling thermometer.” Among Republican and independent evangelicals who rarely attend church, Trump was viewed more favorably than Sen. Ted Cruz of Texas, himself an evangelical.

But even among regular churchgoers, Trump trailed Cruz only by a tiny margin (62 to 66). Regardless of evangelicals’ church attendance, they rated Clinton an extremely cold 15.

Thus, Trump had solid support from evangelicals across the board even as early as the spring. Clinton, however, was strongly disliked.

But Clinton’s problem was not so much in failing to win over right-leaning evangelicals. After all, those were votes she was not expecting to win. Instead, she struggled among some secular voters.

Clinton held a major advantage among secular Democrats and independents who called themselves atheist or agnostic, a group that might be described as having a stronger commitment to secularism. Her average rating was 56, while Trump’s was 13. (This group gave Bernie Sanders a rating of 80.)

But compared with Trump’s strength among evangelicals, Clinton was relatively weak among the secular Democrats and independents who described their religion only as “nothing in particular.” In the ANES pilot study, those respondents rated Clinton 53 on the feeling thermometer, with Trump trailing — but not by much — at 34.

The consequences are evident in head-to-head matchups between Trump and Clinton. In the graph below, Trump had a large advantage over Clinton among evangelicals regardless of their religious commitment. While Clinton had strong support among atheists and agnostics, she lagged — and Trump closed the gap — among other secular voters.

Trump’s support among evangelicals is no doubt partly explained by the widespread antipathy toward Clinton (and Obama, whom evangelicals also dislike). Some were also likely voting for conservative Supreme Court justices, who Trump has promised to appoint. Despite concerns about Trump’s moral character, evangelicals appear to be the base of the Republican Party, at least as long as the Democrats keep nominating candidates in the Obama-Clinton mold.

For Democrats, the lesson may be that they can’t take the nonreligious vote for granted. Seculars might become the bulwark of the Democratic electoral coalition someday, but they will have to find the right issues to mobilize them.

One possibility is to appeal more strongly to seculars’ staunch progressivism on economic, cultural and environmental issues. Democrats might also argue that seculars’ values are under attack by the GOP, borrowing evangelicals’ rhetoric of cultural embattlement.

Whatever the specific message, seculars need to be courted and mobilized, as the Republicans have done with evangelicals. Clinton’s performance raises questions about how well the Democrats have done that.

Mark Brockway is a PhD student in political science at the University of Notre Dame.

David Campbell and Geoffrey Layman are professors of political science at the University of Notre Dame.