The polls are coming in and evangelicals are backing Donald Trump. Despite major evangelical leaders rejecting Trump, evangelicals appear to be Trump's strongest supporters. This support, however, is inflated by 15 to 20 points because of how polls survey evangelicals.
A recent Pew survey found 78 percent of evangelicals support Trump. This is (slightly) more supportive than they were of Mitt Romney, who received 73 percent of their support in a similar poll four years earlier.
Sociologists and other social scientists have found the best approach is to ask what type of church a person attends. By knowing what type of congregation a person attends, social scientists can group Christians into religious traditions. Evangelicals are those who attend churches that have similar histories. They are churches that are distinct from Catholics, mainline Protestants, and historically black denominations.
Pollsters often don't have this kind of time and resources to ask about specific denominations. They need a quick and easy measure. They can't ask if someone is an "evangelical" because many people don't know what this term means. Pollsters ask instead if a person is a "born-again or evangelical Christian." To make sure that they don't include Catholics, Mormons, and others who say they are born-again, they include only Protestants. They also count only white, non-Hispanics as "evangelical" (more on this decision later).
So, when pollsters talk about "evangelicals," what they really mean are "white, non-Hispanic Protestants who identify as a born-again or evangelical Christian."
This decision to define evangelicals as white, self-identified born-again Christians inflates the support for Republicans.
Using the 2012 General Social Survey (sociology's gold standard survey), I compared how measuring evangelicals based on denomination differs from the white-only-born-again measure used by pollsters.
In 2012, 75 percent of "white, non-Hispanic Protestants who identify as a born-again or evangelical Christian" voted for Mitt Romney. This is close to the 78 percent found in exit polls using the same measure.
But if we look at those in evangelical churches, then Romney vote drops to 61 percent. There are two reasons for this decline. Some is due to evangelicals who do not identify as "born again." Most of the difference is due to race and ethnicity. There are increasing numbers of blacks, latinos, and other minorities who attend historically white denominations. If we count only white, non-Hispanics in these churches, the vote for Romney goes up to 73 percent.
The vote for Romney goes up even further if the two measures are combined. Around 87 percent of those who identify of white born-again Christians who attend an evangelical church voted for Romney.
Support for Romney plummets if we used the born-again question without taking into account race. Among all born-again Protestants, only 52 percent voted for Romney.
So, which is it? Was the support for Romney 52 percent, 61 percent, 73 percent, or 87 percent?
Most experts would choose the 61 percent figure. It is based solely on the type of church a person attends. This keeps it distinct but does not exclude blacks or latinos who attend an evangelical church.
Using just the born-again question brings in many blacks who are in churches that are African Methodist Episcopal, National Baptist, Church of God in Christ, or other denomination that developed separately from other evangelical churches. There is a black Protestant tradition in America that remains distinct from historically white denominations.
That said, there are minorities in evangelical denominations. Excluding all minorities from the evangelical category goes too far. Race and religion overlap, but they are not the same.
Pollsters can't tease this distinction out because they rely on one question. In an effort to keep those in historically black denominations distinct, they exclude all minorities. The result conflates race and religion and thus inflates support for Republicans.
Every poll I have seen on evangelicals has been transparent about how they are measuring evangelicals. The tables clearly state "white evangelicals." The additional reports give the questions used. The problem comes when this nuance is lost in headlines and public debates.
If the goal is to talk about only evangelicals that are white, then the polls are fine. But if you want to talk about the voter in the typical evangelical church, then the polls probably inflate the actual Trump support by 15 to 20 points.