By Paul A. Djupe and Ryan P. Burge
[photo credit: Fr. Peter Preble.]
Apologies off the bat – this is some inside baseball. If you care about sorting through how people would classify themselves in broad religious groupings, there are now two strategies to parsing out Protestants (or those we academic types think should be called Protestants; it’s amazing how many “Protestants” don’t understand their Reformation roots). Researchers either sort through a thicket of denominations or they ask for a broad affiliation (e.g., “Protestant” or “Other Christian”) and then ask if they identify as “born again or evangelical.” This provides a positive embrace of evangelicalism, but it leaves the rest of Protestantism as the remainder. Is the former juggernaut of the Protestant establishment, “Mainline Protestantism”, comprising such denominations as the ELCA, UMC, ECUSA, PC(USA), and a few others, the same thing as a “non-born again Protestant”? We know that we have felt some hesitation at making this equation before, so it was time to sort out whether this is indeed an appropriate move.
We turned to the usual suspect data, the Cooperative Congressional Election Study to find out. The CCES is particularly useful because it has appropriate measures for both schemes — the affiliation route and the pan-identity route — by taking question text from the Pew Research Center. We’ve used these data in a prior post to point out how deeply problematic these two schemes are since they both mis-classify very large numbers of people. Large numbers of affiliates do not identify coherently and large numbers of identifiers do not affiliate in a consistent way. In the 2016 CCES, 75% of white mainline affiliates would consider themselves non-born again Protestants. Putting that aside for the moment, we can assess whether the white non-born again Protestants return similar estimates about political attachments as denominational affiliates. Let it be clear, though, that these will draw on different people to arrive at their estimates.
In the figure below, we can see that the WNBAP (white non-born again Protestant ) measurement scheme makes some sense in terms of what denominational affiliates compose its ranks, though it has a good number of non-denominational Christians and Pentecostals and others who just don’t know how to classify themselves. Still, most all WNBAPs make sense.
So, what about their politics? While the mainline has a reputation for being social justice warriors, that appellation is earned more from the denominational elite than people in the pews. However, just like the majority of Americans, the WNBAPs and the component denominations are majority in support of same-sex marriage. Only United Methodists come close to majority opposed and that portion has shrunk since 2012. WNBAPs fall right in the middle as we would expect if it is a valid substitute for a denominational approach.
The figure below shows the fully appropriate comparison — white mainline Protestants classified by denomination compared to WNBAPs in their views on same-sex marriage. The two are distinguishable in 2010 and 2012, but the gap is consistent across yearsand averages about 2 percentage points. For instance, in 2016, a denominational approach yields an estimated opposition to same-sex marriage of 33.1%, while the WNBAP approach yields an estimate of 31.1%. Those confidence intervals overlap.
The distribution of partisan affiliation in these denominations belies unity — each houses a wide diversity of partisan affiliation among individuals affiliates as well as a range of congregational partisanships. Those splits have not changed much over the last decade as the figure below shows. There is some variation by denomination that parallels the same-sex marriage support numbers and the WNBAP estimate is dead center — independent.
The following figure provides the best comparison – showing how partisanship estimates look for WNBAPs and denominationally-categorized mainline Protestants. Clearly, they are indistinguishable for all but 2010 and these are very small differences only made significant by very large sample sizes in the CCES.
CONCLUSION – If you are interested in aggregate statistics about religious groupings and have little room on your survey, then using a two step religious affiliation and born again/evangelical question set will yield estimates that are very close to those using an elongated affiliation and denomination approach.
While we have had occasion to use this approach, it is clearly chock full of variation. These are group averages that strain at the reins of the label “central tendency.” On top of that, the equation of these two measures begs some uncomfortable questions about just what is going into their production. Some of these folks are in evangelical denominations but do not identify as such, just as there are evangelical identifiers in mainline denominations. They clearly net out so the aggregate estimates are similar. But that fact just pushes us away from thinking of either of these schemes as coherent and therefore pushes us back to the congregational and individual levels to figure out why their politics are so diverse and why their identifications does not match their denominations. These measures are functionally interchangeable at the population level, but all of the interesting dynamics are underneath the hood.
Paul A. Djupe, Denison University Political Science, is an affiliated scholar with PRRI, the series editor of Religious Engagement in Democratic Politics (Temple), and co-creator of religioninpublic.blog (see his list of posts here). Further information about his work can be found at his website and on Twitter.
1. In the 2016 CCES, the PCUSA is 95.3%, ECUSA is 92.3%, UCC is 93.2%, UMC is 93.9%, and the ELCA is 97.6% white.
2. It’s important to note that the CCES changed how the question was asked. 2008 and 2010 are like this: Constitutional Amendment banning Gay Marriage: Support or Oppose. 2012-2016 like this: Do you favor or oppose allowing gays and lesbians to marry legally? Favor or Oppose.