The Washington Examiner’s Timothy P. Carney just released his third full-length book, Alienated America. The basic gist of the book is: the determinative difference between pro-Trump and anti-Trump voters is that the former suffers from a deprivation of social capital and/or social institutions, while the latter does not. This, Carney argues, is why both the very wealthy and the very religious didn’t like Trump — the vision of America that Trump was selling applied to those for whom work, family, church, and community had vanished, and that wasn’t them.
For Carney, church is fundamental to this thesis. He’s a devout Catholic, and one can reasonably infer that he thinks — as any good believer ought—that a society bereft of the spiritual will soon be no society at all. As such, in Carney’s view the collapse of faith is uniquely important in the causal story of Trump. As he put it in the subtitle to a recent American Conservative article, “the lack of faith, not factories, inspired a Middle American movement around one man.”
Carney spends a fair amount of time in his book compellingly arguing that lower church attendance corresponds to greater support for Trump in the GOP Primary, and I have no reason to doubt his analysis. But he goes further in arguing that not only does less church mean greater Trump support, but this is because a decline in churchedness is uniquely important to the decline in social capital, which in turn means greater Trump support.
“The unchurching of the [white] working class is a specific instance—the most important instance—of the erosion of civil society,” Carney writes in the chapter entitled “It’s about Church” (subtitle: “America’s Indispensable Institution”).
“It’s the most important part of this story even if you don’t have much stake in the sacraments or sermons,” Carney adds. “First off, church is the most important part of civil society in America, in part because it is by far the biggest part… white working-class woe in America is the fruit of collapse of civil society; and in America, among the non-wealthy at least, civil society mostly means church.”
Carney reviews a lot of compelling evidence to suggest that churches play an important role in the broader institutional structure of a community. They contribute to neighborliness, are major sources of charitable labor, and generally improve quality of life. All of this is absolutely true — but does it support the claim that church is “the most important part of civil society”?
Before I go further, I want to emphasize that Alienated America is well worth your time and money. Carney combines decades of serious reportorial experience with a clear, firm grasp of the data on America’s social capital crisis. If you don’t know what that means, all the more reason to buy the book.
As such, the subtitle of this post might be “on slightly disagreeing with Tim Carney.” I think Carney is absolutely right that there’s a social capital crisis in the United States (I might prefer to call it a crisis of declining institutional pluralism or polycentrism, but that’s a distinction for another post.)
Obviously churches are social institutions — but so are bars, factories, public schools, and their ilk. But I think it’s wrong to ascribe some special status to religious participation over other factors — falling marriage, fewer manufacturing jobs, fewer children, mass urbanization, etc. Just for example: the proportion of white people over 25 who are married has fallen 17 percentage points from 1962 to 2018; the number of people employed in manufacturing has fallen from its 1979 peak of 19.5 million by more than 40 percent, hitting just below 11.5 million during the depths of the Great Recession.
So the question for me is about magnitude. If the magnitude of the decline in churchedness, especially among the WWC, is bigger than the magnitude of those other declines, then I think it’s reasonable to ascribe it the disproportionate role Carney does. But if—as I think I show in this post—Americans are only kind of more unchurched, and the WWC has always been disproportionately unchurched, then I think it’s not fair to say unchurching is uniquely more important. It obviously does matter — the question is if it matters as much as Carney says it does.
The basic reason I was initially suspicious here is that — in the aggregate— church attendance hasn’t actually fallen that much. We know this because the General Social Survey has asked people about their frequency of church attendance every survey year since 1972.
Respondents to the GSS are asked to fit their frequency of attendance into one of eight slots, ranging from “never” to “more than weekly.” (Look here for the breakdown — if you want to crunch the numbers at home, the variable is ATTEND.) When you look at how that composition has changed over time, here’s what it looks like:
Basically, in 1972/1973, about 12 percent of Americans didn’t attend church; today it’s about 25 percent.
In one sense, that’s a pretty big increase — doubling is a big deal. But in absolute terms, it still means 75% of Americans go to church (70 percent if you subtract those who go to church less than annually). What’s more, the frequency of weekly(+) attendance has remained roughly constant, and a lot of the non-attendance appears to come from among those who didn’t attend church all that frequently to begin with.
This matters because we’re interested in the role that church plays in social capital formation. Church is a communal Schelling point, but it doesn’t play that role—at least not as effectively—if you’re only a C&E attendant. If the higher frequency attendance remains roughly constant, church is probably having a similar impact on aggregate available social capital.
(A critic of my analysis here might argue that the total number of churches is declining, meaning that the availability of church to unique communities is declining even if frequent attendance is not. I don’t know a lot about this topic, but apparently sociologist Simon Brauer found that the total number of religious congregations in the U.S. actually grew by about 50,000 between 1998 and 2012.)
Of course, churches can provide something to less-frequent attendants, and certainly rely on those same people for financial support. So there’s still an adverse effect on social capital. The question here is about size of the effect on social health, as compared against other determinants. From the graph above, in the aggregate, it’s hard to conclude that religious attendance is more in decline than, say, manufacturing employment or marriage.
But Carney’s argument isn’t about the aggregate, it’s about a specific subset of Trump voters. We could delimit these as Republicans, but I think that’s a heterogeneous enough group to look at a different population, which Carney also focuses on: the so-called white working class.
(A quick disclaimer: The GSS variable on race is weird (it just captures white, black, and other???). The GSS has also only tracked Hispanic identity since 2000, so we’re just gonna have to hope that people who are Hispanic mostly check “other” and accept that we’ll have an imperfect measure.)
Just to start, here’s the same chart from above, but filtering out only self-identified whites.
Doesn’t look all that different, which makes sense given that whites make up the majority of the country. The never percentage is slightly higher as of 2016, and the weekly percentage is slightly lower.
But we want to look at the “working class” specifically. The GSS’s income variables are wonky (because of inflation, I think), so I’m going to look at education, which closely tracks job market position and socioeconomic status. The GSS asks number of years of education, but you can chop that up into rough educational attainment with relative ease.
This figure revisualizes church attendance as an area plot — percentages of attendance by each of the possible responses I’m comparing a) education level and b) 1972 versus 2016. (It’s among adults over 22.)
There’s a lot of information here, but to try to summarize: between 1972 and 2016, the proportion of weekly attenders fell and the portion of non-attenders rose among whites across every education group. The number of weeklies dropped a little bit and the number of yearlies rose a little bit on average.
This facially confirms the Carney claim that the WWC is unchurched — but it also suggests that all whites are more likely to be unchurched than they used to be. The magnitude of the increase in non-attendance is greater among those with less than a H.S. degree, although not notably greater among those with only a H.S. degree. But is it enough to merit placing church at the center of our account of social decay? I’m not so sure.
As Howard Wall noted on Twitter, there’s an important difference between the two comparison years — Americans, including white Americans, have gotten more educated between 1972 and 2016.
For our purposes, we don’t actually care about why churched attendance has dropped — even if it’s entirely explained by “everyone who would attend church shifted into a higher education bracket,” that’s still a social trend that affects the WWC’s relative social capital standing. Either the white working class is attending church less, or people who attend church are less likely to be in the working class than they used to be. Either way, there’s a noteworthy change.
This chart does show us, however, that the definition of “working class” might shift over time. The wage premium associated with a college education has only grown — it might be appropriate to think about the working class as increasingly including many Americans with high school degrees as the education mix shifts.
The first chart above is just two points in time. What happens if we look at the trend in church attendance over time by education group? Carney stresses a “growing irreligiosity among the white working class” — but has it grown in a way that deviates from historical trends? He further claims that “among the non-wealthy at least, civil society mostly means church” — but is that actually the historical case?
Here’s weekly attendance among Americans over 22 (smoothed with a five-year moving average), which I think is a good proxy for “gets social capital out of church attendance.”
The interesting result? Whites with less than a high-school education have always attended church less than their higher-educated peers. The divergence of the college-and-up set from the rest of society is interesting here — supports the thesis that social capital clusters among the rich in a mutually reinforcing cycle. Relatedly, as the education mix shifts upwards overall, those with just a H.S. degree start to look more and more like those without a high school degree — indicating that as the definition of working class expands to include H.S. degree-holders, that same group attends church less.
The main takeaway here, vis-a-vis Carney, is that white non-attendance isn’t actually new. The same story shows up if you look at those who never attend church:
The gap here is even more pronounced — whites with less than a high-school degree are substantially more likely to never attend church than basically everyone else, and whites with a high school degree have started to look like those without. And that’s as recently as the mid-1980s: well before the drug/suicide crisis, or offshoring, or any of the other things that definitely are big drivers of the recent social capital crisis.
This flew in the face of my intuitions. I naively expect church attendance to be more common among people of lower social status. I was suffering from what Carney amusingly and aptly labels the “Lena Dunham fallacy” of thinking that the upper classes are all bourgeois atheist nihilists.
But the opposite is actually true, at least among whites — if you’re bereft of social/physical capital in one area, you’re probably bereft of it elsewhere too. And that’s been the case *at least* since 1972 — which suggests that it’s been persistently true as other indicators of the WWC’s social capital have gotten consistently worse.
— — — — — — — — — — — — — — —
So what have we learned? 1) in the aggregate, the frequency of white people not attending church has only increased 10 to 15 percentage points over the past 45 years. That’s a notable change, but less pronounced than, say, the drop in marriage or fertility. 2) the change has been more pronounced among low-status whites, but there’s been a change across all groups. But, importantly, 3) lower-educational-status white people have always been less frequent church attenders, indicating that they haven’t suddenly lost out on a social advantage they otherwise had over their more-educated peers.
I’ll offer the caveats that I’m a) not big on the sociology of religion lit., and b) only looking at data from 1972 on. Carney cites Brad Wilcox, who indicates that the association between religion and wealth has waxed and waned over the past century-plus. I could absolutely be convinced that Carney is right, and religion is the most important thing to the social capital crisis — you’d just have to show me another way to read the GSS data, or other data, which make the case more compellingly than Carney does.
I want to emphasize: I don’t think Carney’s wrong that religion is an important piece of the puzzle; I just think that it’s only one piece. Family, jobs, norms around childbearing and childrearing, extended kin networks, etc. all play an important role — some, I suspect, more important than church. If we want to address the social capital crisis, we can’t imagine that any one factor will be the silver bullet. We have to care about solutions on all fronts, including but not exclusive to religion.
That all said: if you liked this post, or if you didn’t, you should go ahead and buy the book to learn more about this vitally important issue.
 Timothy P. Carney, Alienated America (New York: HarperCollins 2019), p. 134.
 ibid., pp. 134–135
 Current Population Survey Annual Social and Economic Supplement, via IPUMS.
 Alienated America, p. 132.