The New York Times’s Thomas Edsall raised an interesting point in a column published last week: The extent to which the electorate was growing more liberal was itself bad news for President Trump’s reelection bid. Buried in that article was a bit of data that, by itself, speaks to a broad shift in American culture — and probably its politics.

Eastern Illinois University’s Ryan Burge looked at data released last month as part of the General Social Survey and determined that the number of people who identify as evangelical Protestants was about equal to the number who say they are members of no religious faith. Those numbers were about equivalent to the percentage of identified Catholics.

Most of those evangelical Protestants are white. White evangelical Protestants make up about 19 percent of the population, according to Burge.

This trend doesn’t come out of the blue. In 2015, the Pew Research Center estimated that a quarter of Americans would be religiously unaffiliated by 2050, up from 16 percent in 2010. Burge’s analysis found slightly more nonreligious Americans in 2010 and suggests that the group will make up a quarter of the population much sooner than 2050.

So how does this overlap with politics? Well, fairly directly. While white evangelical Protestants are heavily Republican — and among Trump’s most fervent supporters — nonreligious Americans are more likely to vote Democratic. In the 2018 midterm elections, exit polling suggests that 75 percent of evangelical voters preferred the Republican candidate in their local House elections. Seven in 10 voters without a stated religion cast a ballot for the Democrat.

That doesn’t mean, however, that as many votes were cast by nonreligious people for Democrats as were cast by evangelicals for Republicans.

Since 2004, here’s what exit polling tells us about voting in presidential (2004, 2008, 2012 and 2016) and House elections. (The data for 2006 suggests a big surge in evangelical turnout that favored the Democrats more than in any other election, so let’s take that with a grain of salt.)

The first thing to notice here is that while the size of those nonreligious circles is growing — from 10 percent of the electorate in 2004 to 17 percent in 2018 — evangelicals still make up a higher proportion of the electorate. In 2018, more than a quarter of voters identified as evangelical.

Notice, too, that these are white evangelical Protestants, the group that Burge estimates make up only 19 percent of the population. In other words, evangelical voters make up a much heavier portion of the electorate than of the population, while nonreligious voters make up less of the electorate than of the population. That skews things in favor of evangelical voters.

Nonreligious voters are also less heavily Democratic than evangelical voters are Republican. On average over these elections (excluding the weird 2006 data), evangelicals backed the Republican candidate in each election by a 57-point margin. Nonreligious voters backed the Democrat by 45 points, on average. That’s in part a function of nonreligious voters voting more heavily third party in presidential election years. On average, about 2 percent of evangelical voters in each election and nonreligious voters in midterm elections voted third party. In presidential election years, about 8 percent of nonreligious voters voted third party, according to the exit poll data.

What’s particularly interesting about the data revealed by Burge is how it mirrors another central demographic shift that the United States is undergoing: the increase in the percentage of the population that isn’t white. By 2050, non-Hispanic whites will make up less than half of the population. Nonwhite voters tend to vote more heavily Democratic but also tend to turn out less heavily overall.

In other words, Edsall’s point about 2020 may be generally accurate, though we’re in something of a gray zone on the effects of religion in particular. Over a longer time period, however, the electoral shifts stemming from these demographic changes should become much more obvious.