How did American evangelicals vote in the 2016 election?

Based on polling data and news sources, you might be under the impression that an overwhelming number of evangelicals—more than 80 percent—voted for Donald Trump. But this isn’t quite accurate. There isn’t any way to truly know what percentage of evangelicals voted for our president-elect. But using a more nuanced analysis we can reasonably estimate that somewhere between 35 percent and 45 percent of all evangelicals in America voted for Trump.

Why are the media reports so off the mark? Here are four reasons:

1. Exit polls do not capture the ‘evangelical’ vote, only the ‘white evangelical’ vote.

All conclusions about 2016 voting patterns reported by the media are based on a single survey conducted by Edison Research. (Edison collected the survey for the National Election Pool, a consortium of ABC News, Associated Press, CBS News, CNN, Fox News, and NBC News.) While there are reasons to be skeptical of exit polls in general (e.g., they don’t use a random sample), let’s assume that this poll is sufficiently reliable as far as the factors it was able to measure. For this reason alone, we should be leery of claims made about “evangelical” voters.

This exit poll survey asked people to self-identify their religion from a range of choices. You could, for instance, choose to identify as evangelical on the survey—but only if you are white. If you’re an evangelical of non-white race or ethnicity—Latino, black, Asian, and so on—your closest option was to identify as “Protestant or other Christian.” As far as this exit poll is concerned, the label “evangelical” is reserved for whites only.

This means the exit poll literally has no way to determine how evangelicals voted. It doesn’t even try to do so. Like the media that commissioned the survey, it is merely interested in the subset of evangelicals who happen to be white.

UPDATE: A reader who had worked as a survey taker sent me a picture of the questionnaire. It turns out the actual question on the form is, “Would you describe yourself as a born-again or evangelical Christian?” Race is not included on that question. And yet in every media report I’ve found (Pew research, New York Times, CNN) they all imply the question is only about “white born-again or evangelical Christians.” If the media has data on how black and Latino evangelicals voted, why aren’t they releasing that info?

2. The exit poll conflates ‘evangelical’ and ‘born-again.’

For more than a decade, observers of religion in America have attempted to point out to both media and pollsters that the terms “evangelical” and “born-again” are not synonymous. It’s a subtle, but substantial, distinction: While almost all evangelicals would describe themselves as “born-again,” not all who identify as a born-again Christian would say they are evangelical. For example, some Mormons even consider themselves to be “born again Chrisitians,” yet no evangelicals (that I’ve ever known) would consider Mormonism a branch of evangelicalism.

Yet on this exit poll voters could choose to identify as a “white evangelical or white born-again Christian.” Because the two groups have been lumped together into one category, it’s impossible to determine how many non-evangelical, born-again Christians are being counted.

3. Many cultural Christians who never go to church identify as ‘evangelical’ or ‘born-again.’

Polling companies aren’t to blame for how respondents answer questions about their religious identity. Still, any poll or survey that merely asks someone to identify as “evangelical” or “born-again” without any additional clarifying questions should be viewed with skepticism.

“There's a form of cultural Christianity that causes people to respond with 'evangelical' and 'born-again' as long as they're not Catholic, even though they haven't been in a church since Vacation Bible School as a kid,” said Russell Moore, a TGC Council member and president of the Ethics and Religious Liberty Commission of the Southern Baptist Convention.

To the media, such distinctions may be unimportant. But if we are seeking a fair and accurate representation of actual evangelicals, it’s important to distinguish them from those who do not truly subscribe to evangelical beliefs and practices.

4. Exit polls only tell us about the people who have voted.

While this point may seem obvious, misleading extrapolations based on exit poll numbers are the norm. For example, many people will look at the exit poll data and claim that 81 percent of (white) evangelicals voted for Donald Trump. But this is not what the exit polls reveal. The data merely show that 81 percent of (white) evangelicals who voted voted for Trump. Again, the poll doesn't even attempt to represent a survey of all evangelicals, merely a subset of a subset.

According to the United States Elections Project, there were an estimated 231,556,622 Americans eligible to vote, but only 131,741,000 voted. That means only 56.8 percent of the eligible population voted, and 43.2 percent of the eligible population did not.

Let’s look at a simple model based on these figures and built on the assumption that the number of evangelical voters/non-voters matches the general population.

Imagine there are only 100 self-professed white evangelicals in America. Based on the exit data we could say that out of those 100 only 57 voted in the election, and that out of those 57, only 46 (81 percent of 57) voted for Trump. Therefore, instead of saying 81 percent of white evangelicals voted for Trump, a more accurate claim would be that 46 percent of white evangelicals who were eligible to vote did so. In other words more evangelicals did not vote for Trump (or Clinton) than voted for him (or her).

Although the accuracy of this 46 percent figure is debatable, it provides an upper-limit estimate for how many evangelicals actually voted for Trump. While it may be lower, it cannot be (much) higher.

More Reasonable Estimate

Indeed, if we add in the number of non-white evangelicals (about 20 percent), the number of evangelicals ineligible to vote because of a felony conviction (since 28.9 percent of Americans identify as evangelical and 6.5 million Americans have a felony conviction, we can estimate that nearly 1.7 million would be ineligible), the number of “culturally Christian” voters who identified as evangelical, and so on, the actual number of evangelical Trump voters would be even lower, likely between one-third (roughly 35 percent) and two-fifths (about 40 percent).

Whether you consider that final estimated number to be too high or too low, one thing is certain: it is substantially less than the 81 percent figure that is being touted as representing the voting figures for our faith community.

Before we opine on what evangelical voting behavior means, we should first make certain our claims are based on reasonably accurate assumptions about how evangelicals voted—or didn’t vote.

Addendum: One more category from the exit poll that is worthy of notice is the “Best description of vote.” While the majority of Democratic voters said, “I strongly favor my candidate” (53 percent), only 42 percent of Republicans said the same. The majority of Republicans said the best description of their vote was “I dislike the other candidates” (51 percent).

Assuming the same percentage is true for white evangelicals who voted Republican, we can make a rough estimate and conclude that the majority voted for Trump because they did not like Clinton. We can also assume that approximately only 1 in 5 of all evangelicals (about 18 percent) strongly favored Trump—about the same as the number that strongly favored Clinton (an estimated 19 percent).