Senate candidate Roy Moore speaks at a “Drain the Swamp” campaign rally at Jordan’s Activity Barn in Midland City, Ala., on Dec. 11, 2017. (Photo by Jabin Botsford/The Washington Post)

Editor’s note: Alabama’s recent Senate race raised some intense debates among white evangelicals, a faith group that — when voting as a whole — tends to vote in high numbers for Republican candidates. Moore’s candidacy tested whether those evangelicals would turn out in high support of a candidate who was accused of sexual misconduct by several women.

When exit polls revealed that white evangelicals who voted cast their votes for Roy Moore with the same percentage as for President Trump (80 percent of those who voted), the reaction was swift. Some responded with a sense of shame and self-reflection. Others defended Moore’s voters, since voting for a Democrat, many believe, would be a vote in support of abortion. And some believe that Alabama evangelicals, who make up about half of the state, don’t reflect evangelicals nationwide. But another group emerged: those who felt newly curious about the many white evangelicals who didn’t vote at all, and about the possibility that this group might have been less-than-passionate about Moore — leading to Jones’ win through their absence.

Here’s what Scott Clement, polling director for the Washington Post, suggests we can glean from limited exit poll data compared with previous elections.

The network exit poll finds 80 percent of white evangelical or born-again Christians supported Roy Moore, 10 points lower than the share of this group that backed Mitt Romney in 2012. The shift is statistically significant and would have been enough to overcome Jones’s 1.5 percentage-point victory margin. The exit poll also suggests that those evangelicals made up a slightly larger share of voters in previous elections  they reached 47 percent in 2012 and 2008 compared with 44 percent today.

Some 37 percent of eligible Alabama voters cast ballots, according to an estimate by Edison Research, which conducted the exit poll. The turnout was closer to levels in previous off-year governor’s races than in presidential years. But while it is clear that a majority of Alabamians sat out the election, the exit poll is not conclusive on whether white evangelicals were more or less likely to do so.

The exit poll has sampling error similar to other random sample surveys, which can mean that small differences are not statistically significant, even if they imply substantively meaningful findings in a close election. In Tuesday’s exit poll, the total sample size for the result on white born-again vs. all others is 2,225 interviews. The sample size for the 2012 and 2008 Alabama exit polls (which showed 47 percent of voters were white evangelical) was smaller: 1,021 and 1,075 respectively.

Taking these sample sizes into account, the three-point shift in evangelical identity to 44 percent is not statistically significant. The difference would need to be about 3.7 points (4 points rounded) in order to be significant. (For interested readers, the survey company Langer Research has a handy online calculator for statistical significance when comparing poll results; use the “comparing groups” calculation for this.)

So the question of whether evangelicals’ share of the vote altered the election cannot be answered with great confidence. The sample size is sufficient to say there was a slight shift, but it’s really hard to say much more than that. The other point on turnout is that in an election decided by 1.5 percentage points, nearly all small differences are consequential, not in one particular group.

Even if white evangelical turnout did drop sharply, it’s difficult to divine whether this was driven by discomfort with Moore’s candidacy rather than more traditional reasons for low turnout. Lack of attention and news coverage can be ruled out for most people, but even so, people tend to be more motivated to vote in presidential election years, driven by constant and prolonged news coverage and the stakes of the election.

History suggests that much of the turnout drop-off is due not only to levels of enthusiasm for particular candidates but cyclical factors. Our graphics team highlighted the big shift in presidential and midterm election turnout in 2014 here.

Moore’s support among white evangelicals is historically low for a Republican. At the same time, the drop-off in Moore’s support among other white groups from previous elections (particularly non-evangelicals, white women and whites with college degrees) is far larger, indicating that evangelicals were far less likely than other typical Republican voters to alter their party support with Moore as a candidate.

It’s very possible that many evangelical voters stayed home because they lacked interest in either candidate. But the exit poll is not conclusive on this front. One thing the exit poll does suggest is that if the election were limited only to white evangelical Christians who cast ballots, Moore would have won with 80 percent of the vote.