The Beginning for the American Church
The Pieces Are Finally All On the Board; Let the Games Begin
Regular readers, or Twitter followers, know that I’m a Lutheran, specifically part of the Lutheran Church-Missouri Synod, the more conservative of the two largest Lutheran denominations in the US. A series of recent conversations with friends made me wonder what trends in religious affiliation look like in the U.S. Plus, there’s been no shortage of recent discussion about the future of Christianity, a topic that interests me very much. Sure, there are lots of “snapshots” of religion, especially in the last 20–40 years, but I wanted something more detailed and with a longer frame of reference. So I decided to build a complete annual dataset for every religious group in America as far back as I could get data, which turns out to be 1925.
Note: for readers uninterested in American religious demography, this post will be boring. For readers who are not theologically conservative/traditional/evangelical/orthodox Christians, this post will be abrasive at times, but please understand, if you’re coming here from that background, you’re eavesdropping on an in-house conversation. My intended audience today is Christians, mostly of the theologically conservative variety.
My first step was to use a resource anybody who works with religion data knows well: ARDA- the Association of Religion Data Archives. They have collected every available estimate of denominational membership for hundreds religious groups from 1925 to 2010. Now, each denomination has data for different years; none have data for all years. So you can’t just drop the data into a spreadsheet and call it a day. To be able to get interesting summary results, you have to come up with estimates or imputations for empty years. Plus, 2010 was 7 years ago now. So I had to look up more recent estimates for every religious group, or come up with plausible extrapolations.
The end result is a spreadsheet that looks something like this, where yellow values reflect imputations:
Even then, there are other shortcomings. Some denominations just BAM appear, with thousands (or millions!) of members. In the vast majority of those cases, it’s due to name changes or mergers/schisms in denominations, so I can build a time series of the pre-existing denominations to get a consistent estimator. But in some cases, I can’t, so I have to just come up with plausible backwards-looking extrapolations of membership.
The problem is even more fraught, however. ARDA data is almost exclusively groups that are either longstanding in the U.S., or Christian. So Muslims, Buddhists, Hindus, Jains, Zoroastrians, Sikhs, etc, are all ignored. Many New Age movements are uncounted. So I have to come up with estimates for those groups. Well, in 2010 the decennial religion census did survey these groups, which is nice, but it’s mostly a survey of adherency, whereas the ARDA data is institutional membership. Institutional membership can be greater than adherency, but is usually lower. Using a range of adherency surveys as well as some plausible adjustment factors to account for the adherency/membership gap, I came up with estimates for groups not included in ARDA.
Finally, it turns out “membership” has an unstable meaning. Some denominations have redefined membership at various times. Some denominations will only count adult membership, while others count kids too. Some only count regular attendees as members, some don’t. So the relationship between “adherency” and “membership” is unstable across denominations and time. In other words, granular comparisons of groups are a crapshoot unless you work to account for these differences!
So the dataset I’ve compiled is not the best dataset for estimating how many people believe in Wesleyan doctrines. It is not the best dataset for estimating how many people will call themselves Presbyterian. Rather, my dataset should approximate the institutional size and strength of American religious movements since 1925.
This matters. Adherency is important for society, as are specific beliefs. But much of the social impact of religions is via institutions and group membership. Institutions support schools, charities, NGOs, political activity, etc. It seems likely that attending non-members are less likely to be active participants in these activities.
While my data isn’t the end-all be all of religiosity in America, it is probably a good indicator of the social influence of religious groups.
And, crucially, to my knowledge, the dataset I’ve built is the most complete and up-to-date such dataset in existence, well, anywhere. Rigorous academics will be skeptical of it because I did tons of freestyle fill-in-the-blank, which is fair. But that’s exactly what we do when we compare decennial census rounds: even though censuses are on different years, we compare them together by extrapolating into a common year. All I’ve done is do this consistently across all groups and years, to allow us to get ballpark estimates of major groups.
A note on data access before we get to results: I am happy to share headline data, and if you have specific questions about denominations, then I’m happy to discuss or share a bit at a time. But this data has actually taken a lot of work to put together, and I intend to make heavier use of it in the future, and I’m still hoping to improve it. As such, I won’t be sharing the complete denomination-level dataset. I know, I know, that’s kind of a dick move on my part. But again, if you’re interested in a specific denomination or group, I’d love to share the bit you’re interested in! I just don’t want anybody pirating my work without attribution. So if you’re thinking, “Huh, I wonder what he estimates for my denomination?” just ask, I’ll share! But if you’re thinking, “Gee, I’d love to get that whole dataset and use it for this other project I’m doing…” sorry dude. I’m not game for that just yet.
Finally, this is a blog post on a personal blog. This is not an academic publication. My data almost certainly has errors. My hope is that in publicizing, people will ask me questions that lead to me finding some of those errors and improving on them.
How to Read the Data
Before I get to the actual data, let me explain some classification issues so you know what you’re looking at. It is standard in religious demography to avoid calling conservative denominations conservative or liberal ones liberal. We use categories like “Mainline” and “Evangelical,” which aren’t the same as liberal/conservative, but have some overlap. But this creates new confusion, because, for example, the Evangelical Lutheran Church in America is “Mainline” whereas the Lutheran Church-Missouri Synod is “Evangelical.” I will use classifications I’ve invented: conservative and liberal. The defining characteristics of each relate to their view of scripture: the more tolerant a denomination is of the view that the Bible contains errors, the more likely I classified them as “liberal.”
I also classed some groups as “Heterodox Christian.” I want to be clear, I don’t mean that as pejorative. These groups hold beliefs very different from traditional Christian movements. Most of them are non-Trinitarian. Examples include Unitarians, Mormons, Jehovah’s Witnesses, Oneness Pentecostals, etc. Some of these groups do not even claim to be Christians (Unitarians, for example), but I classify them as such because they historically arose out of Christianity. You’ll also notice that I have liberal and conservative protestants, but then also a third technically Protestant group. This third group is composed of mostly less organized religious groups, or groups that are not structured as traditional protestant denominations. Examples include nondenominational churches (estimated indirectly, for obvious reasons), Pentecostals/Charismatics, the Salvation Army, independent fundamentalist churches, etc.
Finally, we have the non-Christian groups, and then the “Unclaimed.” The “Unclaimed” group would include children in denominations that don’t count kids as members, non-member attendees or adherents of many groups, members of religious groups I missed somehow, as well as true “nones.” Given my focus on institutional data, I can’t estimate these groups separately.
To be clear, many conservative people worship in liberal denominations, and vice versa. But conservative denominations tend to be the ones that will have pro-life political activity, that hold traditional views of sexuality and marriage, that tend to have higher church attendance among adherents, that are less likely to have ordained women, etc.
Some groups are very hard to classify. For example, the United Methodist Church has very strong conservative and liberal wings that are vying for dominance of that group. I have chosen to classify the UM as “liberal,” but thus far “conservatives” have won recent political battles. Nonetheless, I call the UM “liberal” because church discipline is rarely enforced with respect to church doctrines concerning scripture.
Meanwhile, the liberalness of some denominations has changed. Take the Episcopal Church. The Episcopal Church in 1925 may have been fairly conservative. Today, it is very liberal. I classify denominations by their current position, and back-cast that directly into the past. When multiple denominations merge into one denomination, I allow the original groups to have different affiliations, though this is rare. But crucially, when denominations schism, I allow them to have different affiliations.
In the last 40 years, both Anglicanism and Lutheranism have seen major schismatic movements. LCMS had liberal schismatics in the 1970s, then ELCA has had conservative schismatics since the late 1990s. Anglicanism has had conservative schismatics since the 1970s, but especially since the 2000s. When a group breaks away, I count them by their own individual affiliation, not with their parent denomination.
I know this is a lot of explanation, but bear with me. It matters. Let me give an example of why.
Say there are 100 churches. 80 are in a liberal denomination, 20 in a conservative denomination. Say there are 100 people per church. Let’s say all 20 churches in the conservative denomination are conservative, while 25 of the liberal denomination churches are conservative. Let’s say every church is shrinking at the same rate. And let’s say that after a few years, 20 conservative churches break away from the liberal denomination.
If you estimate “conservative religion” based on church-level membership, you get a very different estimate and very different trend than if you estimate “conservative religion” based on denomination-level membership. Here’s the results:
As you can see, the denomination-level estimates are quite different from the church-level estimates. And you could erroneously think that conservative churches are growing if you use denomination-level estimates, even though conservative churches are shrinking! Over the last few decades, schismatic movements have indeed tended to be led by conservatives leaving liberal denominations, so this could artificially inflate conservative growth!
But hold on. Is it really artificial? It may be artificial in the sense of wrongly-estimating membership, but a new conservative denomination unencumbered by liberal compatriots (and vice versa!) may be more free to pursue the kind of church life it wants to pursue, may be freer to exert its political muscle with less compromise or dissension, etc. So even if denominational estimates wrongly present trends in the ideological traits of adherents, they may accurately present trends in the public presence and expressiveness of denominations.
And more to the point, denominations with more unity of belief and agreement on their distinctives (either liberal or conservative!) might flourish more and grow faster. That’s speculative, but think of it this way: if you forced stodgy Lutherans to share a service with a bunch of Pentecostals, nobody will be happy. Nobody will get to worship the way they find most rewarding. In all likelihood, that denomination will fail at holding onto people. It is therefore plausible speculation that schisms that result in more freedom for both new denominations to pursue their own distinctives could be growth-positive. If indeed schisms yield more growth-positive denominations regardless of ideological or theological conviction, the graph could look more like this:
I have only denominational, not church-level data. If somebody has an “in” with Lutheran Churches in Mission for Christ, let’em know I’d loooove to look at their church-level membership and attendance data to see what happened pre-post schism from ELCA. But for now, this question is theoretically ambiguous.
Anyways, let’s finally get to the data!
Douthat Was Right: Liberal Protestantism Is Dying
On Sunday, Ross Douthat’s column argued that American liberals need to save mainline Christianity and go back to church. The chart below will make abundantly clear what he’s talking about.
As you can see, the blue line for Liberal Protestant denominations is headed steadily downwards. Conservative Protestantism is mot stable, nudging downwards bit by bit, but has been buoyed in recent years by conservative schismatics within Anglicanism and Lutheranism. In the 2000s, non-traditional Christian denominations surpassed Conservative Protestantism, and they may soon surpass Liberal Protestantism. Meanwhile, non-Christian religions and Heterodox Christianity are both growing, and may, within a decade or two, be equal to or greater than Conservative Protestantism.
Jumping up to the top of the chart, we can see that Catholicism and Orthodoxy are both growing. I should note, I estimate a lot more Orthodox members than most surveys; over 4 million. I suspect a very, very big share of these are “on the books,” but never, ever show up to church, probably don’t even identify as Orthodox.
And then, of course, we have the unclaimed. As I mentioned, the unclaimed aren’t all “nones.” Many are just kids who aren’t old enough to be “members” in certain denominations. Some are religious groups I missed. Some are non-member adherents.
Americans not formally tied to or claimed by any religious group are the fastest-growing segment of the American religious landscape. Liberal Protestants are the fastest-declining.
These two trends go hand in hand. Let’s assume for a moment that Liberal Protestantism had continued growing at the 1925–1965 rate until the present day, and all other traditions remained the same.
The unclaimed population remains steady until 1990 in this scenario, then rises. Since 1990, Liberal Protestant weakness has spread to Conservative Protestantism. And since 2005, it has begun to spread to Catholicism and to nontraditional Christian movements, though their growth seems to be continuing for now.
But the point is, the vast majority of the rise in the unclaimed population is pretty straightforwardly a question of Liberal Protestantism failing to maintain its growth.
The data bears out as well that poor Liberal Protestant retention is the actual direct source of the growing unclaimed population. Pew Research finds that only about 45% of people born into Mainline denominations remain there today, vs. 65% for Evangelical Protestants, 59% for Catholics, 53% for Orthodox, 70% for historically black protestants. Jews, Muslims, and Hindus all have higher retention, while Buddhists and Jehovah’s Witnesses are quite low. Of those raised in each group, Mainline Protestant kids are the 3rd most likely to end up Unaffiliated, at 26%, behind Jehovah’s Witnesses (35%) and Buddhists (40%). Evangelical and Black Protestants are the least likely to become unaffiliated. So, yes, it really is the collapse of mainline denominations that gives rise to the large unclaimed or unaffiliated population and, to a lesser extent, reaffiliation by Catholics. For Millennials, just 37% of Mainline-raised remain in the Mainline, vs. 61% for evangelical protestants.
So to be clear: this isn’t just a question of demographic transition and aging. This is a question of some denominations doing systematically worse at retaining and attracting people over the last few generations.
Looking at Lutheranism
I’d like to walk through one detailed example to help people understand the data. Because I’m Lutheran, I’ll look at Lutheranism. Plus, Lutherans, being who they are, have kept relatively complete and consistent data on membership for a long period of time, and the denominations all define membership in roughly similar ways, which means we can speak more confidently about trends.
I’ll start by showing all Lutheran denominations.
It’s chaotic. I didn’t place labels because there were too many to make meaningful distinctions, but if you download the data, labels are included.
But this data is clearly useless due to obvious mergers and schisms. So let’s reclassify by groups. I’ll reclassify all of these denominations into five groups: ELCA and its precursor denominations, LCMS and its precursor denominations, ELCA Schismatics, the Wisconsin Synod, and Other. For those who don’t know, ELCA is the most liberal Lutheran group, LCMS is conservative, Wisconsin Synod is somewhat more conservative, and the “Other” Lutheran groups are an ideological mix. The ELCA breakaways are more conservative than ELCA, but less conservative than LCMS. For purposes of my major categorizations, I have treated ELCA as a liberal protestant group, but all other Lutherans as conservative protestants, though in truth the ELCA groups could be recast as Liberal Protestants reasonably enough.
As you can see, every Lutheran group except for the ELCA Schismatics is shrinking, and that group is growing by adding new ELCA breakaway churches each year. Remember the example we showed earlier of church/denomination growth differences? This is a prime case for that!
Because Lutheran denominations are very different sizes, however, assessing relative growth rates can be hard from the raw numbers. So instead, we’ll index to 1970.
As you can see, LCMS has modestly outperformed ELCA since 1970, but the difference was very small until the 2000s. What happened then? Simple! ELCA had some schisms, and LCMC and NALC both broke off from the ELCA! I didn’t index these groups on their own because, well, they didn’t exist in 1970! But I did recalculate LCMS + Schismatics and ELCA + Schismatics. Basically, if you count the schismatics as being part of ELCA, you can see that ELCA membership has actually tracked LCMS almost exactly. So if you call the Schismatics “Liberal Protestants,” then the main Liberal and Conservative denominations have the same growth performance. But if you call the Schismatics “Conservative Protestants,” the LCMS+ group does quite well, and shows membership stability for a number of years. As we discussed earlier, it’s really not clear what the best way to count these groups actually is.
But more to the point, the most conservative group, the Wisconsin Synod, has actually seen the best membership performance. To me, this gives credibility to the idea that you shouldn’t just assign ELCA Breakaways to ELCA: in fact, there do seem to be growth differences along theological lines within Lutheranism.
Now, I also have family who are Anglican, Presbyterian, and Methodist. So for completion, I’ll show headline liberal/conservative estimates for liberals and conservatives within Lutheranism, Anglicanism, Presbyterianism, and Methodism.
It should be noted that the hashed lines for liberal churches are larger than the solid lines for conservative churches for every single group. It should also be noted that conservative denominations tend to be more uniformly conservative, whereas liberal denominations have more internal diversity of theological views. And of course, many people may disagree with specific denominational classifications, which is fine. In particular, I will note again my choice to identify United Methodism as liberal. It is probably one of the most “conservative” bodies I identify as liberal.
Broadly speaking though, the conservative wings of each denomination family seem to be having the strongest performance. The starkest contrast is probably for Reformed/Presbyterian traditions.
The Next Christian Century
But alas, this post is already long, and I’ve probably irked or offended enough people for one day. If I mischaracterized your denomination, please let me know. If you think I’m misinterpreting the data, let me know. If you know of better data, seriously, let me know! I’d love to improve this dataset. But for now, I leave you with one final note. The unclaimed population is, of course, a hard-to-define group. But while many factors explain them, it is undeniable that a large share of them are probably truly unaffiliated. Here’s the unclaimed, Christian orthodox, and heterodox Christian + non-Christian religion shares of the population:
In the early years of my sample, I am probably undercounting Christians somewhat. My guess is Christians actually amounted to more like 45% of the population. But even then, American history has seen far less Christian-dominated times than the present day. If you go back and read religious history from the late 1800s and early 1900s, then you’ll know that internal U.S. missionary societies were a huge phenomenon because there were so many unchurched people! This was largely thanks to immigration, which separated people from their national churches in Europe, but was also just because Americans were a footloose people often unwilling to be tied down somewhere. As migration fell in the 1920s and 1930s, as immigration declined, and as American Christianty launched aggressive internal evangelism efforts, the Christian membership share of the population rose, and it remained high through the 1990s. But today, Christian membership probably remains more prevalent than in the 1920s and 1930s. This will be jarring for many people to read.
Now true, in the 1920s and 1930s, the Christian moral vision was less challenged. Even non-members basically recognized Christianity as being morally authoritative, and “getting religion” would basically mean “becoming Christian.” Today, our society is more pluralistic: non-Christian or heterodox Christian faiths are growing quickly, and even orthodox Christianity is increasingly divided, with traditional denominations declining and new ones rising.
But nonetheless, Christians considering withdrawal from cultural engagement due to some defeat should stop and count the score better. Organized Christian religion remains more prevalent than at many times in American history. Most decline has occurred among the least theologically distinctive and rigorous denominations, and many theologically conservative denominations continue to grow. The fact that secularists feel free to more loudly proclaim their views does not mean Christianity in America is finished or defeated. Rather, it means that the debate over the soul of the nation is only now beginning. Christianity has had hundreds of years of uncontested opportunity to evangelize and catechize an enormous nation. Now, other viewpoints have the freedom to compete as well. A Christianity which cannot compete is a Christianity which would disappoint and embarrass its apostolic fathers, and rightly so.
So gird yourselves for the next century, Christians of orothodoxy. We have an opportunity to engage with a wider range of worldviews than ever before. Personally, I think it’s gonna be fun.
Check out my Podcast about the history of American migration.
If you like this post and want to see more research like it, I’d love for you to share it on Twitter or Facebook. Or, just as valuable for me, you can click the recommend button at the bottom of the page. Thanks!
Follow me on Twitter to keep up with what I’m writing and reading. Follow my Medium Collection at In a State of Migration if you want updates when I write new posts. And if you’re writing about migration too, feel free to submit a post to the collection!
I’m a native of Wilmore, Kentucky, a graduate of Transylvania University, and also the George Washington University’s Elliott School. My real job is as an economist at USDA’s Foreign Agricultural Service, where I analyze and forecast cotton market conditions. I’m married to a kickass Kentucky woman named Ruth.
My posts are not endorsed by and do not in any way represent the opinions of the United States government or any branch, department, agency, or division of it. My writing represents exclusively my own opinions. I did not receive any financial support or remuneration from any party for this research.