But there’s one category of comment that’s become its own genre. Here’s how it usually goes. I’ll post a graph on mainline Protestants — seven denominations, five election cycles of data, a chart I’m genuinely proud of. And the replies come in fast:

“Where are the Catholics?”
“Why do you ignore the Baptists!?”
“What about Latter-day Saints?”

I get something like 100 of these a day. But one group shows up with a consistency I can set my watch to: people who are upset that Orthodox Christianity isn’t in the graph. And in my experience, it’s almost always young men making this particular complaint.

So this post is for them. Let’s actually dig into the widely held belief that American Orthodox Christianity is experiencing a huge surge in converts. Short answer: that’s a really hard claim to test with survey data, for a simple reason: there just aren’t that many Orthodox Christians in the U.S. to begin with. (Though judging by my comment section, a lot of the converts have found their way to me specifically.)

The dataset I’m using is, by any objective metric, the best survey available for exploring migration patterns around Eastern Orthodoxy: the Pew Religious Landscape Survey. It’s a probability-based design, it was conducted fairly recently (2023–2024), and it has a huge sample size — nearly 37,000 respondents.

Let’s start with a simple calculation of retention across a bunch of different religious groups.

The groups with the highest retention rates are Muslims and Jews. In both cases, over three-quarters of people raised in the tradition still identify with it in adulthood. That’s a strong number — about ten points higher than the overall retention rate across all religious groups. Right behind them are the non-religious: if you’re raised without a faith tradition, there’s a 75% chance you’ll stay non-religious as an adult. That’s one of the more important findings in the Pew data — the “nones” identity has become very sticky.

The Orthodox land right in line with the full sample’s retention rate. For every three people born into the faith, one leaves. That’s roughly consistent with the other larger Christian groups — Protestant retention is five points higher, and Catholic retention is eight points lower, at 58%. So there’s nothing here suggesting Orthodox Christianity is especially good at keeping the next generation. It’s, by an objective measure, pretty average on this front.

Let’s go one layer down and compare retention among older Americans to younger ones. I focused on a smaller handful of groups here because of sample size limitations.

Read more

My current polling place is St. Mary’s Catholic Church, where we attend Mass on Sundays. It’s less than a mile from my house and they have a really nice fellowship hall and a parking lot that is really close to the door. I can leave my house, do my civic duty, and get back to my graphs in less than ten minutes most of the time. Democracy is awesome.

But what makes my life so convenient on Election Day is the fact that the local Catholic church is willing to spare some space for the community. If they weren’t willing to allow the county clerk to set up shop there, I’m not entirely sure where I would vote. I’m willing to bet an alternate location wouldn’t be nearly as convenient.

A couple of months ago, I wrote a post about the percentage of Alcoholics Anonymous meetings that take place in houses of worship. Really what I was interested in was the role that churches play as community gathering spots. Now, I want to extend that question to polling places.

Lucky for me (and you), there was a wonderful repository of polling places accessible on the internet. It was compiled by The Center for Public Integrity. Now, they didn’t grab every single state for reasons that I bet they can explain more clearly, but they got most of them. My analysis will include 35 states in total, and I think this data is more than adequate to address the question: How many polling places are in houses of worship?

Now, the data file didn’t have a specific column that denoted which locations were actually in churches, synagogues, mosques, etc. Instead, I had to build a keyword database of terms that would likely show up when that location actually was a house of worship. I tweaked it a bunch and finally got to a spot where I felt comfortable with the analysis.

I had a total of 62,481 polling places from 2020. Of those, 13,598 got classified as a house of worship using my lookup table. That results in 21.8% of polling places being in houses of worship. I think there’s some wiggle room in that, of course. But I feel very confident that 20-25% of polling locations in the United States are in churches, synagogues, mosques, temples, etc.

It varies a bunch by state, though!

For instance, a majority of polling places in the state of Oklahoma are in churches. That’s the only state where it’s a majority, by the way. You are going to see throughout the rest of this analysis—Oklahoma elections are just different than any other state. The state in second place was Arkansas at 41%. Then, there’s a really big gap before you get to a whole bunch of states that are clustered around 30%.

The states that are the least likely to have a polling place in a church are Nevada at zero percent and then a couple of states up in New England. It was just 2% of polling places in Maine and Vermont, and only slightly higher in New Hampshire at 3%.

But the middle band of entries is this really weird hodgepodge of seemingly random states from all over the country. You have Georgia at 32%, Nebraska at 29%, Iowa at 26%, and Utah at 22%. I just don’t know if there’s any kind of trend there when it comes to the region. I thought that visualizing this as a map would make the spatial component easier to parse out.

I think seeing the darker blue like this does point to the fact that there is a regional concentration. The clustering of Oklahoma, Arkansas, and Missouri in that top tier really stands out from the rest of the country. But I don’t know if we would typically group those three states together in any regional way. I mean, they are kind of adjacent to the South, but not exactly. Maybe if the database included Kentucky, Tennessee, and Mississippi, it would make it clearer. But I can only work with the data I have.

What’s interesting to me is that as you move away from that deep blue in the epicenter, the shades get lighter and lighter. You still see fairly high percentages in places like Alabama, Georgia, and the Carolinas. But then as you get even further away, you can find places like New Mexico at 4% and Wisconsin at 8%.

Again, I am not entirely convinced that there’s some kind of regional thing going on here. And I’m going to poke around on that a bit more later on in the post.

But before I get to that bit of analysis, I wanted to see if I could identify what type of religious organizations are more inclined to host a polling place. This part wasn’t actually that hard when it came to searching for terms like Methodist, Presbyterian, Baptist, etc. A lot of churches (thankfully) still include their denominational affiliation in the name of the church.

The one tradition that stands out immediately is Baptist. About a quarter of all polling places located in houses of worship were in Baptist churches, compared with just 11% in Methodist churches.

I went back to the 2020 Religion Census to see whether that pattern simply reflected the overall size of these traditions. My best estimate is that Baptist congregations outnumber Methodist congregations by roughly 3 to 2. In other words, for every two Methodist churches in the United States, there are about three Baptist churches. That larger congregational footprint helps explain why Baptist churches appear so frequently as polling locations.

Every other group was down in the single digits. Just 9% of polling places were in Lutheran churches and it was only 5% for Presbyterians. The one that I had a really hard time parsing was the Catholic calculation. Most Catholic churches don’t include the word “Catholic” in their official name. So I messed around with “Saint” and “St.” and such. But the problem there is that some non-Catholic churches use Saint names, as well. So I don’t feel as confident about that estimate compared to the others.

And don’t get me started on the non-denominational problem. You see that 24% in “other church” up top there? That’s places like “The Journey” or “Community Church.” Take theology out of the equation – I’m not a fan of non-denominationalism just because it makes classification like this so much harder.

But I know what all of you die-hards really want. You want an interactive county level map of all 35 states so you can mouse over to figure out what percentage of polling places in your locale are in a house of worship. Of course, I can’t disappoint on that front.

I do think that this is a better way of visualizing than just the state level map. I mean, the state of Texas is just fascinating. A lot of darker counties in the eastern part of the state but as you start to move out to the oilfields of West Texas, it just gets a whole lot lighter and almost blends into New Mexico completely.

Looking over that county-level map did trigger something in my mind, though. I was mousing around my home state of Illinois. In Cook County (that’s Chicago), only about 10% of polling places were in houses of worship. But then you go down to Coles County—that’s where EIU is located—and it’s 44% of polling places in churches. You can probably see where this is headed:

Are you more likely to find a polling location in a house of worship in a densely packed urban county or a really sparsely populated rural part of the United States?

This is why doing data analysis is so much fun – I can prove myself wrong with actual evidence. I was convinced that rural America was way more likely to have polling places in churches than a place like Chicago or Indianapolis. Yeah, that’s not even close to true. In fact, it’s the very opposite of my assumption.

Among the least populated counties in the United States, just 15% of polling places are in houses of worship. As we move up the density scale, that share just continues to slowly climb. For the counties that score in the top 20% for density, almost a quarter of polling places are in houses of worship. I guess since I’ve never lived in an urban area, I had no idea about this. For those who read this and live in the big city, chime in for me: Where do you vote?

Okay, so back to that state variation thing. Oh my goodness, did I toil over this so much. I ran about a dozen regression models. I subbed in all kinds of variables. I was trying to figure out just how much state context mattered when predicting the share of polling places in churches. I finally landed on a model that just included: county population, number of congregations, and religious adherence rate.

So here’s what we’ve got: Imagine identical counties with the same population (median ≈26,600), same number of congregations (median = 57), and same religious adherence (median ≈49%). The only thing we change is the state they are located in. What is the predicted probability that the county has at least one polling place in a religious building?

In Oklahoma, it’s a near certainty that this hypothetical county will have a polling place in a church (~95%). It’s also above 90% in both Missouri and Pennsylvania. Try to find any kind of connection between those three places. But there are a whole bunch of states where the predicted probability is still really high. In fact, in 25 of the 35 states I included, the probability was north of 50%.

Then, look at those states at the bottom. That random hypothetical county in Maryland only has a 4% chance that it’s got a voting location in a church. It’s just 12% in North Dakota. So I don’t know if we can even point to anything like a regional factor. It’s probably more about weird little historical quirks in certain places. Oklahoma was probably putting polling places in churches for decades, so when a new one was needed it was just assumed that the county clerk would call up churches and ask for permission to use their fellowship hall. That same culture didn’t exist in Maine.

And I do have to mention Utah. Why is it basically zero percent? Utah is an outlier because religious polling places are concentrated almost entirely in just two counties: Utah County and Salt Lake County. There are 29 counties in the state, but only two have any polling places in houses of worship. Salt Lake County alone accounts for 17 of the state’s 24 religious polling places. But the model is predicting the probability for a typical county with Utah’s characteristics — and the typical Utah county is one of the other 27 counties that have none.

The thing that surprised me most in this analysis is that houses of worship are not simply a reflection of how religious a place is. A state can have lots of churches, high religious adherence, and still rarely use those buildings as polling places. What matters is the relationship between religious institutions and the broader civic life of a community. In some places, the local church has long been understood as a gathering space for the entire community — not just a place for Sunday worship. In others, elections happen somewhere else.

Churches, synagogues, mosques, and temples have always played roles beyond their primary religious purpose. They host recovery meetings, food pantries, community events, and, yes, sometimes elections. The polling place data gives us another glimpse into how deeply religious organizations are woven into the everyday infrastructure of American communities.

Code for this post can be found here.

Ryan P. Burge is a professor of practice at the Danforth Center on Religion and Politics at Washington University.

This post has been unlocked through a generous grant from the Lilly Endowment for the Association of Religion Data Archives (ARDA). The graphs you see here use data that is publicly available for download and analysis through link(s) provided in the text below.

A couple of months ago, I wrote a post about the retention crisis facing the Church of Jesus Christ of Latter-day Saints. The headline finding is pretty straightforward: nearly half the kids who are raised as LDS no longer identify that way as adults. There was a bit of controversy about the best way to compare LDS retention rates to those of other groups like Protestants or Jews, but no one has been able to dispute the simplest conclusion of the data from Pew – 46% of the LDS are not retained.

In the comments for that post, one of my paid subscribers noted that they would like to see me do a more in-depth analysis of retention among American Catholics. And I immediately jotted that down as a note to follow up. Now, I’ve managed to flesh out a full post on this topic.

The Association of Religion Data Archives is awesome for a project like this because I can quickly find the datasets that I need to answer this question from multiple angles. I am going to use two different sources – the General Social Survey, which allows me to trace the contours of American religion all the way back to the early 1970s, and the Pew Religious Landscape Survey, which permits me to do a really deep dive on the state of Catholic retention in the last couple of years.

My biggest impression through looking at all the data is that American Catholicism does seem to be hemorrhaging members at a pretty alarming rate. The Pew RLS found that just 58% of Catholics were being retained into adulthood. But how did we get here? The GSS makes it pretty simple to trace the trajectory of Catholic retention over time.

I am going to do a lot of comparisons of Catholicism and Protestant Christianity in this post because these are the two largest religious traditions in the United States. However, I also recognize that the comparison is not completely apples to apples, which I will discuss at the end of the piece.

Protestant retention has always been higher than Catholic retention. That’s true at every single point over the last five decades. The gap is not that small, either. Back in 1973, 91% of folks raised Protestant were still Protestant. For Catholics, it was 84%. So a difference of seven percentage points. It’s notable to me that Catholic retention was already clearly dropping by the early 1980s, when the same can’t be said for the Protestant trend line for at least another decade or so.

I am a bit surprised by how these lines tend to run in pretty consistent parallel to each other. There are a few dips and dives for one or the other, but they always seem to be in the neighborhood of ten points apart. That’s basically where they stand in the most recent data, too: 72% of Protestants were retained in 2024. It was 62% of Catholics.

So that’s pretty clear and compelling evidence that Catholicism tends to be “leakier” than their Protestant cousins.

The graph above is pretty blunt, though: it just calculates “retained” or “not retained” with no nuance. I wanted to take that a step further by visualizing four possible outcomes: not retained, retained but only rarely attending church, retained but only attending occasionally, and retained and attending weekly or more. This may be the most revealing data visualization of the whole bunch.

The “not retained” line should come as no surprise based on the prior graph. It’s basically the reverse of what I did in the first visualization. It’s the other three lines that tell a much different story about how retention works in these two traditions.

The one thing that I want you to focus on is when the red line eventually crosses the green line. That’s the point in which the “not retained” outnumber the “retained and attending weekly” share. For Catholics, this happened over twenty-five years ago. That’s when someone who was raised Catholic was just as likely to leave Catholicism as they were to stay in the Church and be a weekly Mass attender. Now, someone raised Catholic is about twice as likely to leave the Church entirely as they are to be a regular Mass goer as an adult.

For Protestants, the crossing of the two lines didn’t happen until the last couple of years. Now, about 28% of those raised Protestant will leave the tradition entirely and the same share will stay in Protestantism and go to church nearly every week or more.

Let me put a really fine point on this. In the 2024 data, among people who were raised Catholic, nearly three quarters of those respondents were either no longer Catholic or they still identified as Catholic but attended Mass seldom or never. Just a quarter were going to Mass at least once a month. Now, it’s not like Protestants have nothing to worry about in this regard – the same statistic for them is about 55%. But Catholics are 20 points worse than that.

Now, I turn to the Pew Religious Landscape Survey, which is perfect for analyzing the factors around retention because of its recency and huge sample size (nearly 37,000). First, let’s calculate retention of Catholics and Protestants by the decade of birth of the respondent.

The macro trend is pretty easy to parse out – retention is declining among younger Americans compared to their parents and grandparents. For the very oldest respondents to the RLS, 71% of Catholics and 82% of Protestants were retained. But with each successive decade, both numbers begin to erode in a pretty meaningful way. When you get to folks born in the 1970s, Protestant retention is down to 71% and for Catholics, it’s even lower at 56%. For the very youngest adults, retention is about 60% for Protestants and 50% for Catholics.

Let me point out something that I missed on my first pass at this graph, though: Catholic retention has bottomed out. It was 50% among those born in the 1980s and it’s functionally the same for folks born in the 1990s or early 2000s. That’s a hopeful sign that it won’t tumble much further from there. But the same is also true for Protestants, too – except retention is about ten points higher (~60%).

But one thing that is unmistakable: that ten point retention gap that appeared in the first graph is also replicated here, more or less. In older cohorts, it runs as high as 21 percentage points, but among younger adults it seems to have held around ten percentage points or so.

However, here’s where the value of the Pew RLS data comes into play. It asked respondents how often they attended religious services when they were kids. So we can answer a question like, “Maybe Catholic retention is just lower because they were never that engaged in their faith community.”

The older birth cohorts are indicated across the top set of graphs. You can see that there’s a persistent gap between Protestants and Catholics that logically follows from the prior analysis. Even holding childhood religious attendance constant, a person raised Protestant is often 15 points more likely to still be Protestant compared to Catholics from the same birth decade. For instance, among those born in the 1950s who attended church weekly as kids, 84% of Protestants were still Protestants. It was only 64% of Catholics.

The bottom row does complicate this narrative just a bit, though. The gaps are certainly smaller among younger birth cohorts compared to older ones. That becomes most apparent when looking at the trend lines for those born in the 1980s and later. Among the youngest adults who were raised not going to church that much, retention is about the same for Protestants and Catholics (between 45 and 50%). I do need to point out, though, that the gaps shoot back up for weekly attenders. For a Protestant born in 1980 or later who went to church weekly, about two-thirds were retained. For Catholics from the same cohort, retention was just above 50%.

I did want to explore one other question that asked respondents, “Thinking about the type of school you attended when you were growing up, from kindergarten through 12th grade, how many years did you attend a private religious school, if any?” I wondered if Catholics had been able to close the retention gap by sending their kids to private Catholic education.

The answer for older cohorts was clearly “no.” I mean if you compare Protestants who went to at least seven years of Christian education, retention rates were a staggering 84%. (That’s Amish level retention, for what it’s worth). For Catholics, it was a dozen points lower than that. Among the middle birth cohorts, the gap was actually larger (60% vs 77%). So, that’s actually pretty compelling evidence that a Protestant going to Christian school is more effective at keeping kids in the church compared to Catholics sending their children to the local parish school.

Among the youngest folks in the survey, retention is just lower. That’s already been established, but that persistent gap between Catholics and Protestants is still there. For those who attended private Christian schools for at least seven years, Protestant retention was 71% compared to 57% for Catholics. That’s a pretty tough pill to swallow. And let me point a finer point on this:

Among Catholics born in the 1990s or 2000s who attended Mass at least weekly as children and spent seven or more years in religious school, just 57% still identify as Catholic as adults.

When I was pulling together that last stat, I thought it would be helpful for me to control for a bunch of factors to see if I could get that retention gap to close between Protestants and Catholics. So I threw in control variables for just about everything I could reasonably measure about a respondent’s religious upbringing: how often they attended services as a child, whether they attended a private religious school, how personally important religion was to them growing up, how important religion was to their family, and when they were born. Even after accounting for all of that, the gap remained.

If you had two children who were born in the same decade, and raised them in a very similar religious environment, the Catholic kid would be about eleven points less likely to be retained compared to the Protestant. Now, I did try to poke around some variables to try and help us understand what factors were the most predictive for retention for both groups.

My conclusion is: the strongest childhood predictor is not simply attending services or religious school. What matters most is whether religion feels personally important. Compared with adults who said religion was very important during childhood, those who said it was not important have dramatically lower odds of remaining in their childhood tradition. This relationship is evident among both Catholics and Protestants.

Additionally, younger birth cohorts have lower retention even after their childhood religious experiences are held constant. This generational decline is especially steep among Protestants. Compared with Protestants born in the 1960s or earlier, those born in the 1990s and 2000s have about 69% lower odds of retention. The corresponding decline among Catholics is about 41%. Catholics still have lower retention overall, but the traditional Protestant advantage has narrowed considerably among younger Americans.

But I do have to end on this point that I mentioned earlier: comparing retention rates for Protestants and Catholics is not just apples to apples. Someone raised Southern Baptist can become non-denominational and that’s still being retained. Or a United Methodist could join an Episcopal Church and that’s retention, too. For Catholics, it’s pretty binary. Either you are Catholic or you are not.

I don’t know how to account for that because there’s really no way to throw that difference into a statistical model. But I’m not even sure that comparing things is that useful. Instead, I think that I will end this by reiterating a point I made earlier.

If someone is raised Catholic, there’s a 75% chance that as an adult they are no longer Catholic or they are still a Catholic but hardly ever attend Mass. Among people who were raised Catholic, just 15% are now weekly Mass attenders. For Protestants, it’s nearly double that rate (~28%).

That’s about as apples to apples as I can get. And on that metric, the difference is a chasm.

Code for this post can be found here.

Ryan P. Burge is a professor of practice at the Danforth Center on Religion and Politics at Washington University.

But one of the most valuable lessons came at the very end of my methods training. After years of learning regression analysis and research design, we felt ready to explain the world. Our professors offered two words of caution.

The first was an old saying: when all you have is a hammer, everything looks like a nail. We had learned a powerful statistical tool, but that didn’t mean every interesting question could—or should—be answered with a regression model.

The second warning has stayed with me even longer: don’t use those tools to predict the future. Forecasting is a dangerous business. Even if you’re right most of the time, no one remembers your successes. They remember the one prediction that turned out to be spectacularly wrong.

Well, I was reminded of that lesson a few weeks ago when I stumbled upon a book written in the early 2000s by the Canadian sociologist Reginald W. Bibby. In 2002, he published Restless Gods: The Renaissance of Religion in Canada. His thesis can be derived from the title: despite huge drops in religious attendance and involvement in Canada, there was a deep well of spirituality present in the population. It would just require churches to adapt a bit to bring in these seekers.

You can probably guess where this post is going. The Canadian General Social Survey is fielded every year, but it doesn’t ask about religion in every wave. However, it did include a battery of questions about religious affiliation, importance, and attendance in 2003, 2008, 2013, and 2020.1 This gives me ample data to test the theory of a religious revival in Canada.

Bibby couldn’t have been more wrong in his forecast.

In the full sample, 80% of Canadians identified with some kind of religious tradition back in 2003. Today, that number is closer to 66%—a fourteen-point drop in seventeen years. The share of Canadians who say religious beliefs are important to them was 71%, but that has dropped to 56%. This is exactly what projection models would have predicted by simply tracing the trend lines from previous decades: a slow and methodical decline in overall religiosity.

In 2003, 26% of Canadians indicated they attended religious services at least once a week. By 2020, that had dropped by more than half to just 12%. Actually, the largest decline in terms of percentage points is the share who had attended church at all in the prior year. It was 75% in 2003 and just 40% in 2020. That may be the most telling statistic of all: a significant majority of Canadians have not darkened the door of a church in the last year.

There are two ways to make sense of this decline. One is simple generational replacement: older Canadians were more devout, and their children and grandchildren are more likely to claim no affiliation. People die and new generations come of age, moving aggregate trends downward. The other is that religiosity has declined within birth cohorts. Perhaps Boomers, Gen X, and Millennials have simply walked away from religion as they have aged.

We can test that by tracing religiosity across these four waves of the General Social Survey. Remember that in the full sample, religious affiliation dropped from 80% to 66% during this period.

A quick glance at this graph reveals that the primary driver in the rise of the “nones” is generational replacement. There hasn’t been that much religious defection within the existing population. For instance, among Boomers, the share who were religious dropped just four percentage points between 2003 and 2020 (81% vs. 77%). For Generation X, the drop was only three points (74% vs. 71%).

Millennials buck this trend in a pretty significant way. In 2003, 70% of them aligned with some type of faith tradition. By the last wave of the GSS, that was down to 55%. That 15-point drop among Millennials was actually larger than the decline in the three older birth cohorts combined. Furthermore, while I didn’t include it because I only had one data point, just 51% of Gen Z were affiliated in 2020—nearly 20 points lower than Millennials when they were entering adulthood.

Let’s do the same thing for religious attendance and I used both metrics in this analysis: going to church in the prior year and attending weekly or more.

Here, the inter-cohort decline of religion gains much more empirical support. For the Silent Generation, the share who had attended at least once in the last year dropped by a staggering 27 percentage points between 2003 and 2020, and weekly attendance rates fell from 37% to 24%. For Boomers, the drop is even larger: in 2003, 73% had gone to church at least once in the prior year; by 2020, it was 39%. Weekly attendance rates among Boomers also fell from 22% to 13%.

For the younger two cohorts, it’s basically the same story. Yearly attendance rates dropped by at least thirty percentage points for both groups, while weekly attendance saw a more modest decline. Among Gen X Canadians, just 11% were weekly attenders in 2013, similar to the 10% found among Millennials.

But what about Gen Z? Are they still heading lower than their parents and grandparents on these religiosity metrics?

In some ways, the answer is very clear: yes, Generation Z Canadians are significantly less religious than Millennials or Gen X. As previously mentioned, only 51% of Gen Z claims a religious affiliation—four points lower than Millennials and twenty points behind Gen X. There’s also a huge gap in religious importance: just 43% of the youngest adult Canadians indicate that religious beliefs are important to them, six points lower than Millennials and sixteen points behind Generation X.

But on attendance measures, the picture is a bit muddier. For instance, 39% of Gen Z attended a religious service in the prior year, which is actually three points higher than Millennials and about the same rate as Boomers. Meanwhile, weekly attendance rates seem to have hit a hard floor. You can see that 10% appears to be about as low as these metrics go; that’s where Millennials and Gen Z are now, with Gen X just a single point above.

So while affiliation and importance measures still have room to fall, I think the decline in religious attendance has stopped. I believe this is a “floor effect.” Once you get to a certain point, you hit bedrock. That’s what is happening in Canada right now. The share who are deeply religious is about one in ten, and it’s not likely to move much from there, even with generational replacement.

I know what most readers are thinking: how does this Canadian data compare to their neighbors to the south? Well, the U.S. has its own General Social Survey, and while the years don’t line up perfectly, they are close enough to make solid comparisons.

On affiliation, the United States has always been slightly more likely to be part of an established faith tradition compared to its Canadian cousins. Numerically speaking, the differences aren’t always large—some years show only a one or two point difference. But in the most recent data, the gap has widened. Among Canadians, 66% are part of a faith tradition versus 71% in the United States.

Attendance is trickier to compare because the Canadian survey offers five options while the U.S. instrument offers seven. I decided that weekly attendance would be defined using each survey’s standard measure: “at least once a week” in Canada and “nearly every week or more” in the U.S. General Social Survey.

With that caveat, you can clearly see that attendance rates in Canada are much lower than in the United States and the gap is widening. It was an 8-point gap back in 2003-2004 (26% vs. 34%), but by 2020-2021, it had grown to 13 points (12% vs. 25%). Secularization appears to have been more rapid up north.

Let’s do a generational comparison between the two countries, as well.

On affiliation measures, it’s striking how older folks in both countries are nearly as likely to be religiously affiliated. If you combine all respondents from the Silent Generation, Boomers, and Gen X, the share aligned with any religious tradition is basically the same. The gaps emerge with those born in 1980 or later. Among Canadian Millennials, 55% are part of a faith tradition compared to 60% of American Millennials. But the divide becomes a chasm with Gen Z: an American young adult is 14 percentage points more likely to be part of a faith tradition than their Canadian counterpart (65% vs. 51%).

The gaps in weekly attendance are more persistent across every generation. What is noteworthy is that the largest differences are among the oldest respondents. Among Americans from the Silent Generation, 43% are weekly attenders, compared to just 24% of Canadians from the same generation. In subsequent cohorts, the differences are smaller: 10 points among Gen X (23% vs. 13%) and 9 points among Millennials (19% vs. 10%). However, it opens up again for Generation Z: 24% for Americans versus 10% for Canadians.

By any objective metric, Bibby’s 2002 forecast was as wrong as it could be. Canadian religion has completely collapsed in the last twenty years, and there are no signs of revival. Interestingly, Bibby wrote a book in 1987 called Fragmented Gods: The Poverty and Potential of Religion in Canada, which was much more dour in its assessment. It earned him the nickname “Bad News Bibby.”

I can commiserate with Dr. Bibby. Tablet Magazine dubbed me “The Antiprophet“ back in May. But don’t expect a book about an American religious revival from me any time soon. Projection is always a fool’s errand, and I try my best to learn from the mistakes of others.

Code for this post can be found here.

Ryan P. Burge is a professor of practice at the Danforth Center on Religion and Politics at Washington University.

I’ve written about the Outreach 100 list before – it’s the best dataset of the largest churches in the United States. Among those, half describe themselves as non-denominational, 15% are Southern Baptist, and 20% ID as independent Christian or “other Baptist.” The Assemblies of God has about 5% of churches on the list, but the next in line are the Methodists at just 1.8%.

Yeah, there’s just not a whole lot of discussion of really big United Methodist Churches for a very good reason – there aren’t that many of them. But that gives me even more motivation to study this subset of American Christianity. Thankfully, the United Methodists have an easily accessible table listed on their website with the title “Largest 100 Churches by Membership.” I wanted to use this list because it offers a glimpse into a problem that plagues a lot of churches, especially in the mainline: their membership rolls are bloated with folks who haven’t attended a service in decades.

I’m going to explore that in just a minute, but let me start by just showing the 25 largest UMC congregations in terms of total membership.

The very largest United Methodist Church in terms of membership is in Dallas, Texas – it’s Highland Park. They report a total membership of just over 15,000 folks. Next in line is a church that you all might be hearing about quite a bit more in the months to come: Resurrection Church in Leawood, Kansas. The reason for that is that their senior pastor is Adam Hamilton, who is running for the United States Senate in Kansas. Their church’s membership boasts 13,419 members.

There are a total of four churches that have at least 10,000 on their membership file but most of these churches on the list are in the 5,000-7,000 range in terms of membership. I also color coded these churches based on their current jurisdiction. The United Methodists love their bureaucracy and there are multiple layers of organization in the UMC. There are five total jurisdictions in the United States and you can see that South Central shows up quite a bit on this graph.

To give you a better sense of geography, here’s a count of the number of times each state shows up in the top 100 list.

It’s pretty apparent where the biggest United Methodist Churches are located: deep in the heart of Texas. Over a quarter of the largest churches can be found in the Lone Star State. The biggest concentration, by the way, is around the Dallas-Fort Worth area. But then you have another seventeen churches that can be found in Georgia. Every single one of them is a short drive from downtown Atlanta. You don’t find a lot of big churches (UMC or otherwise) in the rural parts of the United States.

The rest of this list is pretty concentrated in the South. You’ve got states like Tennessee, North Carolina, and Alabama. The states that are the furthest north are Virginia and Missouri. But you don’t see any states in the top ten that are clearly in the North or the West. The locus of power of the UMC is actually pretty similar to the Southern Baptists – the Bible Belt.

To make that even more plain, I geocoded the location of every single church in the top 100. Then I used those coordinates to build an interactive map of the biggest United Methodist Churches. Each church is represented by a circle that you can click on to provide additional information. It gives you the name of the church, the total membership as well as their average attendance.

Read more

The Bible was such an integral part of my upbringing that it’s hard to describe. We did Bible quizzing drills and scripture verse memorization. By the time I got my driver’s license, I had a pretty good grasp on the meta-narrative laid out in the 66 books that make up the Old and New Testaments. I had Bibles from all kinds of translations, too. Of course the old stalwart was the New International Version (NIV), but I also had the ESV and the NRSV on my bookshelf. I even remember getting several copies of Eugene Peterson’s The Message Bible as swag at a conference I attended.

You don’t know what’s weird until you see your life through someone else’s eyes, though. I distinctly remember that I had one of my friends come by my house when I was a junior in high school. He walked into my room and the first thing he encountered was my book shelf. It had a full row of Bibles. He looked at me with the most puzzled expression on his face and said, “Why do you have SO MANY BIBLES?” I was completely dumbfounded by this question. I knew my buddy was a Christian, a Catholic more specifically, so I just didn’t know where he was coming from.

I asked him how many Bibles he had. His answer was simple, “I have one and it was given to me by my grandma when I had my First Communion.” I just didn’t even know what to say to that. Protestants (and more specifically evangelicals) love the Bible. They are instructed to read it daily, to memorize the words, and to constantly wrestle with it throughout their lives. Other traditions just don’t have that same emphasis.

This is where a very smart theologian would go into a discussion of Sola Scriptura and authority and all that. You won’t get that from me. You get graphs. So, let me show you the genesis of this entire post and how it got me thinking very carefully about how Catholics think about the Bible compared to Protestants.

It all started (as it usually does) with a graph. It’s the one below. It was for another post where I basically make the claim that there are two “camps” in Christianity. Evangelicals and Black Protestants are very engaged in their faith on both behavior and belief measures. Mainline Protestants and Catholics are in the other camp. They just don’t believe or behave as much as their cousins.

I wanted to make it plain that even if you compare the most active mainliners/Catholics with the most engaged evangelicals/Black Protestants, there are still big gaps on belief.

This graph visualizes the share of each tradition that believes in God with total certainty, believes in both Heaven and Hell and also says that the Bible is extremely important to them at each level of attendance. In essence, I was trying to parse out if the belief gap was really just an attendance gap and when I control for that, the differences fade away.

As you can tell, that’s not at all what happens. Among weekly attending evangelicals and Black Protestants, 57% hit a perfect score on all four of the belief items. That was basically twice the rate as their mainline and Catholic counterparts. Even when you restrict the sample who attend multiple times per week, the gap persists.

But then I got to thinking about the construction of my belief index. Remember it had four items: belief in God, heaven, hell and then another question about the importance of the Bible. Maybe that last question is sort of biased against Catholics because they don’t make the Bible as central as evangelicals, for instance.

So, I reran the analysis but just dropped that last question and made the index only three items.

And guess what? The gap between Catholics and evangelicals/Black Protestants disappears among those who attend church at least once a week. In the first graph, just 28% of weekly attending Catholics were maximally religious on the four item test. When I dropped the importance of the Bible, that share shot up to 72%.

I think we can all assume that Bible importance rises with religious attendance, right? I felt like I needed to test that before I went any further.

And it pretty clearly does for both evangelicals and Catholics. There’s a distinct linear progression going from left to right. But there’s also a huge gap between these two types of Christians. Even among evangelicals who never attend in person religious services, they are twenty points more likely to say that the Bible is extremely important compared to never attending Catholics. And what’s interesting is that the gap between the two groups only widens with each step up the attendance ladder. It’s 41 points when comparing weekly attenders.

However, Catholics’ view of the Bible does begin to rise really quickly once you get to the most active in the faith. For instance, just 15% of monthly Mass attending Catholics say that the Bible is extremely important. It doubles when moving to weekly attenders (31%) and then doubles again among those who attend multiple times per week (64%). So it does look like some Catholics do place more emphasis on the Bible.

But there was also a question about how often each respondent reads the Bible, too. Let me show you that alongside a more expansive version of the Bible importance question.

It’s interesting to me how there’s a very consistent gap in each tradition between the share who say that the Bible is extremely or very important and the portion who actually read the scriptures on a weekly basis. The gap is actually the smallest among mainline Protestants – 46% say that the Bible is extremely/very important while 18% read it weekly. A difference of 28 points. For Catholics, the divergence is 32 percentage points, which is exactly the same as evangelicals.

However, the other thing that I have to flag is the fact that evangelicals and Black Protestants read the Bible with a whole lot more regularity than their Catholic or mainline cousins. For the former groups, about half are reading the Bible once a week. For the latter groups? It’s 15-20%. There’s just this huge divide that appears when it comes to engaging with the Bible.

But when you look at this graph and the one right above it, I know what you are thinking: Ryan didn’t control for church attendance in the dumbbell plot. We know that Catholics just don’t go to church as much as evangelicals, so the next logical step is to do an apples to apples comparison.

Everytime I look at this graph the first thing that my eye is drawn to is those stubby little bar graphs for Catholics and mainline Protestants on the left side of the plot. For those who attend church monthly or less the share of each tradition who reads the Bible on a weekly basis is less than 10%. For evangelicals and Black Protestants with low attendance, at least a quarter of them are still reading their Bible on a consistent basis.

But the real action is on the right side of the graph – that’s when I am just looking at Christians who attend church once a week or more. If we compare devout vs devout, that has to close the “Bible gap”, right? Kind of, but not completely. For instance, among weekly Mass-attending Catholics, 78% say that the Bible is extremely or very important to them. That’s still about twenty points lower than evangelicals.

The bigger gap, though, is when it comes to reading the Bible. The question explicitly asks respondents to exclude reading the scriptures as part of religious services, so this is the share who pick up the Bible on their own during the week. Even among Catholics who go to Mass weekly or more, only about a third of them read the Bible on a regular basis. That’s forty points lower than evangelicals and Black Protestants.

That created a bit of a challenge for me: could I figure out what factors would actually close the “Bible gap” between Catholics and evangelicals? That’s what I did in the graph below. The vertical dashed line is the share of evangelicals who say that the Bible is extremely/very important and those who read the scripture weekly. What factors lead Catholics to “catch up” to their evangelical counterparts?

Comparing all evangelicals to all Catholics makes the gap pretty plain – it’s about thirty points on both metrics. But just adding one filter to Catholics (weekly Mass attendance) does a whole lot to bring the two groups together. A regular attending Catholic is only about 3 points lower than all evangelicals on the importance of the Bible, but it still leaves a 16 point difference on Bible reading.

If we compare all evangelicals to Catholics who both attend Mass weekly and pray daily, that gap basically disappears when it comes to their view of the Bible, but there’s still a pesky behavior gap. The only way that I can sort the sample to bring Catholics into line with the average evangelical is just limiting my sample to Catholics who: attend Mass weekly and pray daily and have at least seven years of religious education and also say that religious teachings are extremely important to them. That gets Catholics to 56%, which is a couple of points higher than the average evangelical.

If I compared the same level of devotion for evangelicals, though, it wouldn’t even be close. Evangelicals are “people of the book” in a way that Catholics are not.

This is where the pastors and theologians who subscribe to this newsletter will have very strong feelings about the “right way” to do church. I’m not going to indulge in any of that. On a typical Sunday I attend both a Methodist Church service and a Catholic Mass. Both have strengths and weaknesses. My faith is edified in different ways in both gatherings.

Catholics’ posture toward their faith is different from evangelicals. That is empirically indisputable. What implications that has for both types of Christians is a much more open question.

Code for this post can be found here.

Ryan P. Burge is a professor of practice at the Danforth Center on Religion and Politics at Washington University.

This post has been unlocked through a generous grant from the Lilly Endowment for the Association of Religion Data Archives (ARDA). The graphs you see here use data that is publicly available for download and analysis through link(s) provided in the text below.

Let’s say we did a random sample of the adult population of Ohio in 1985. In that poll, we found that 55% of people identified as politically conservative and 35% were liberal, with the remaining 10% indicating that they were ideologically moderate. Then, we did another poll of adults in Ohio forty years later. This time we found that conservatives were just 35% and liberals had jumped up to 55%. What could possibly explain that kind of shift?

Well, there are two primary drivers of that change.

The first could be the fine people of Ohio just changed their minds about political stuff. Maybe they got fed up with Republican politicians. Or it’s possible that a really charismatic liberal rose to national prominence and convinced the rural Midwest that a pathway to a better life was to shift to the left.

Or, it could be that the actual people living in the state of Ohio changed a whole lot over that forty-year window. Of course, when you are looking at a four-decade span, you are going to have a whole lot of generational turnover. But you also have migration happening. Columbus has exploded in population, for instance.

Either (or a combination of both) could lead to a change in the ideological composition of a place like Ohio.

But the same basic process always looms behind the scenes when I am doing analysis of how religious groups have changed over time. In my most recent book, I provide a lot of data that points to the fact that white evangelicals are way more conservative today than fifty years ago. What’s interesting about that is that evangelicalism is about the same size today as it was back in 1972 (around 20% of the population).

Other religious groups have gotten smaller over time, though. Mainline Protestants are down by at least two-thirds since 1975 (from 30% to less than 9%), and Catholicism has also seen a noticeable decline in the last fifteen years (27% → 22%). So, given all these changes and shifts, one has to wonder: are those who still identify as Christians actually more devout today than the same group from a couple decades ago? A “we may be smaller in number but we are mighty in Spirit” kind of thing.

Well, let’s test that idea. The Association of Religion Data Archives (ARDA) hosts the entirety of Pew’s Religious Landscape Survey. It was fielded in 2007, 2014, and 2023-2024. But before we dig into compositional changes, I wanted to start by visualizing how the three largest religious traditions — evangelicalism, mainline Protestantism, and Catholicism — have changed across generations over time.

The left panel above is the share of the population who identifies with an evangelical tradition. You can pretty quickly figure out that evangelicalism was at least a quarter of every generation from the Silents to Gen X. But then Millennials are a whole lot lower on this metric — just 19% are evangelical. For Gen Z, it’s even lower at 16%. That’s a ten-point difference in just two generations.

But one thing evangelicals have going for them is that there’s really not a consistent slope in the lines. They don’t seem to be losing people as they age. The same is not true for mainline Protestants. Every single generational line points downward — often in the two-point range. What is even more worrisome, though, is the clear drop from one generation to the next. Among Boomers, 16% are still mainline. For Millennials and Gen Z, it’s just 7%.

The Catholic story is a whole lot messier. For Silents, Boomers, and Gen X, the three lines are a rat’s nest. They are all really close to each other in a tight band from 21% to 24%. I just don’t think there’s a lot of movement there in either direction. For Millennials, though, the story is incredibly clear. Back in 2007, 22% of Millennials identified as Catholic. Now, it’s 14%, which is the exact same share of Gen Z in the most recent data.

Now that we have a good sense of the broad trends around affiliation, let’s look inside each generation to see how they have changed on a bunch of metrics. The first thing I calculated was weekly attendance rates.

The trend among evangelicals is really easy to spot. Between 2007 and 2014, weekly churchgoing rates didn’t move in a meaningful way for any group. Attendance was clearly higher for the oldest respondents, but from Boomers on down to Millennials, 56% were at church every weekend. However, for every generation from the Boomers and younger, there’s been a big dip in attendance. It’s fallen between six and nine percentage points. Now, a majority of evangelicals aren’t showing up every week.

That same pattern of stability between 2007 and 2014, followed by a big dip in 2023-2024, is there among mainliners and Catholics, too. The big difference, though, is that these last two groups of Christians just don’t go to church nearly as much as evangelicals. The typical mainliner has a weekly attendance rate that is 25 percentage points lower than their evangelical cousins. Catholics stand somewhere between the two.

What about the political viewpoints of each group?

One thing that is absolutely impossible to miss is how clearly those evangelical lines are moving upward. Every generation of evangelicals is more ideologically conservative now compared to 2007. For Gen X, it’s a twelve-point bump (51% to 63%). It’s seven points for Millennials. But note that Gen Z evangelicals are at 46% conservative in 2024. That’s where Millennials were seventeen years ago.

The mainline and Catholic numbers are just completely different in levels and trends. The simplest way to describe the trajectory is pretty flat, really. There does seem to be a macro-level move in the conservative direction, but not nearly the same amount of movement here compared to evangelicals. For instance, among Gen X mainliners, the conservative share is up five points. It’s seven points for Catholics from the same generation.

But the other thing I need to flag is how there is a clear gap between each generation. Boomers are more conservative than Gen X. Gen X is more conservative than Millennials, and Gen Z is less conservative than the prior generation. And I would be derelict in my duty if I didn’t point out Gen Z Catholics. Just 17% identify as ideologically conservative. That’s half the rate of Millennial Catholics. Certainly something to watch.

The graph is focused on the issue of religious importance and it has a lot of echoes of the analysis of religious attendance. There was a general sense of stability between 2007 and 2014. The only difference here is that there are some signs of slight increases in importance in those first two waves. However, across the board you can see that religious importance dropped in a huge way between 2014 and 2024. For Boomer evangelicals, it went from 81% to 72%. It was down a dozen points for mainline and Catholic Boomers, too. That’s pretty incredible to think about.

For Gen X, the drops were substantively large as well. Ten points for evangelicals, thirteen points for mainliners, and a whopping fourteen points for Catholics. But the Millennial drops were pretty large, too, especially among mainliners. In 2014, 45% said religion was very important to them. A decade later, it was just 29%. That may be the most revealing thing — how important religion is to younger evangelicals compared to the other two groups. I mean, look at Gen Z. Among evangelicals, 66% rated it as very important. It was just 33% among young adult Catholics and 25% for mainline Protestants.

However, let me show you one more metric — this one focused on how certain respondents were in the existence of God.

While the last couple of graphs were relatively stable between 2007 and 2014 before a big drop in 2024, this belief item bucks the trend because certain beliefs have consistently eroded across the full time series. For Boomer Catholics, those figures went: 74% → 68% → 65%. That’s a really common progression you see across the bottom two rows — a dip of ten points across seventeen years.

That’s not what emerges from the evangelical subgroup, though. Their lines do drop, but the magnitude of the decline is modest. For Boomers, it’s just three points. For Gen X, it’s no change at all (89% straight across the board). You do see a little dip among Millennial evangelicals (88% to 82%), but that’s easily the most change across the entire top row.

After looking at all of these, I get the sense that secularization has not eroded the devotion of these three groups at the same rate. Evangelicals consistently score the highest on all these metrics, while mainliners and Catholics started out lower and have seen their numbers continue to decline over time. That means not only has the share of Catholics and mainline Protestants gone down, but even the ones who remain are not as religious either.

So are evangelicals holding strong against the forces of secularism? I tested that by creating a religiosity index that averages three measures — whether religion is very important in daily life, weekly-or-more worship attendance, and absolute certainty in God’s existence — into a single 0-100% score per respondent. Anyone who scores religious on all three components sits at 100%; someone who scores religious on none of them sits at 0%. Then I compared two groups: evangelicals vs. all other Christians.

The first conclusion from this graph is the least surprising one based on the prior analysis: evangelicals are more religious than other Christians. That’s true at all three sampling periods and across all generations. Evangelicals take their faith seriously, and it shows up consistently. The gaps are often pretty sizable, too. The average evangelical scores about twenty points higher on the religiosity index compared to other Christians.

But are evangelicals managing to push back against secularization? The answer is no — not entirely. Look at any single line on the graph and it moves the same direction: down. Silent+ evangelicals dropped from 80% to 75% between 2007 and 2024. Millennial evangelicals fell from 71% to 65%. Every generation of evangelicals is less religious today than the same generation was a decade or two ago. Secularization isn’t skipping over them.

What’s actually happening is more subtle than “holding strong.” Evangelicals are declining more slowly than other Christians, not staying flat while everyone else falls. Millennial evangelicals dropped six points over that span; Millennials in other Christian traditions dropped eleven. That’s the pattern across every generation — the evangelical line has a gentler slope, and the gap between the two groups actually widens as you move to younger cohorts, from about 13 points among the Silent+ generation to 20+ points among Millennials and Gen Z.

So evangelicals aren’t immune to secularization. They’re just secularizing more slowly than the Christians around them.

Remember the example of ideology in Ohio I mentioned in the introduction? If liberals and conservatives completely flip-flopped, why is that? Did people change their minds, or did new populations of liberals move in while a lot of the conservatives died off? Well, if you can control for the age of the respondent, you can shed a lot of light on how shifts like that took place.

Why are Christians today less devout than Christians from 20 years ago? The answer is twofold: younger Christians today are less religious than younger Christians from 2007. But we also know that Christians, regardless of tradition, are secularizing as they age as well.

How do aggregate changes happen in a population? It’s usually a bit of column A and a bit of column B.

Code for this post can be found here.

Ryan P. Burge is a professor of practice at the Danforth Center on Religion and Politics at Washington University.

The system empirical social science has coalesced around is called RELTRAD, and it was first introduced in 2000 in a paper by Steensland et al. called “The Measure of American Religion: Toward Improving the State of the Art.” It uses the General Social Survey’s question structure to create a seven-part typology of American religion. I wanted to focus on just the first four categories in RELTRAD for today’s post: evangelicals, mainline Protestants, Black Protestants, and Catholics. Those four groups together encompass about 60% of the American population.

I don’t want to take a whole lot of time explaining all the contours of each of those three Protestant groups and justifying why they are distinct categories because it would take up too much space in this post. But I have written (with the help of Claude) a pretty nice little one-page website that describes and visualizes the three Protestant groups. For those who are new around my corner of the world, I would strongly urge you to take ten minutes to read it over. There’s also a fun little widget that lets you type in the name of a denomination, and it will tell you how RELTRAD classifies it.

With that out of the way, let’s get to the point of this post. I actually think that if social scientists were only concerned with measuring religiosity when it came to creating a typology of Christianity, then there would be just two groups – not four. My argument is that evangelical Protestants and Black Protestants are almost exactly alike and form one cluster of American religion, while the other category consists of mainline Protestants and Roman Catholics.

Let me just show you what it looks like by taking all of us on a tour of a number of questions that are featured in Pew’s Religious Landscape Survey that focused on religious beliefs and behaviors of those four groups mentioned above.

To begin, we have a really basic question about the existence of God that is asked in two separate parts. The first one is a simple yes/no query: do you believe in God or a universal spirit? Once they answer that, they are asked to express how much certainty they have in that belief, ranging from absolute certainty to being not at all certain.

After taking just a quick glance at the heatmap, you can already pretty easily figure out that two camps are starting to form. The share of evangelicals who say that they believe in God with absolute certainty is 86%, and for Black Protestants it’s 84%. If you combine the top two categories, you get 98% for both. You get the point here, right? Nearly everyone in both traditions expressed a highly certain belief in God. Less than 1% can be found on the right-hand side of the visualization.

But contrast that with the folks who identify as mainline Protestants or Catholics. Just looking at the far left-hand column tells a big part of the story. The share who believe in God with complete certainty is about 20 points lower than the first two groups we just discussed. For mainliners, that figure is 61%, and it’s nearly the same for Catholics at 63%. Instead, there are much bigger shares of these two traditions who slide down the scale just a bit. About 30% of both groups end up in the “fairly certain God exists” category, which is double the rate for evangelicals and Black Protestants.

There are also two questions about respondents’ views of heaven and hell.

And, again, you can see that two really clear typologies begin to emerge here. The group that is the most likely to believe in the existence of both heaven and hell are historically black Protestants – 84% of them can be found in the top left hand corner. Among evangelicals, the top left square is a bit smaller but not by that much – 79%. It is interesting to me that evangelicals are the group that is the most likely to believe in hell but not heaven (6%) while every other group hovers around 1%.

You can see a bit of a break between mainline Protestants and Catholics on this one, though. Among the latter group, 71% indicate a belief in both heaven and hell. That’s a full ten percentage points higher than their mainline cousins and it’s only 8 percentage points lower than evangelicals. So, the Catholics sit in a sort of middle ground on these afterlife questions. I do also need to flag the fact that nearly a quarter of mainliners believe in heaven but not hell—that’s five points higher than Catholics and 15 points higher than evangelicals.

There were also some questions on the Pew Landscape survey that focused on how people made decisions about right and wrong. There were four sources to come to a decision: logic and reason, practical experience, religious teachings, and scientific information. I calculated the share of each tradition who said that each source was “extremely important.”

The first thing that should jump out to you is that the top two categories are extremely popular with all four religious groups. Whether someone is an evangelical or a Catholic, they tend to rely on both logic and reason alongside practical experience to puzzle their way through difficult decisions in their lives. It’s kind of amazing how rock-steady those figures are across all four columns, really. There is some deviation across the bottom row in the question about using scientific information, but it’s really only among evangelicals. They are about 7 points lower than the other three groups.

The big divergence emerges when you take a look at the column that focuses on religious teachings. Among evangelicals, 56% say it’s an extremely important source of guidance for them. It’s 48% among Black Protestants. Then you have mainliners and Catholics at 25%. That’s about half the rate of the other two groups. You can see those two camps becoming even more justified here, right?

Let me show you just one more question on religious beliefs to drive that point home. It asked people to indicate how important the Bible was to them.

And this bar graph is about as clean as it gets in reinforcing the idea that there are two types of Christians. For both evangelicals and Black Protestants, 58% said that the Bible was extremely important to them. And if you add the top two categories together, it’s 83-84% for both as well. This data was so clean I had to check it twice, but these numbers are accurate. Evangelicals and Black Protestants put almost exactly the same weight on the Bible in their faith traditions.

Meanwhile the breakdown for mainline Protestants and Roman Catholics is eerily similar, too. For mainliners, the top two categories comprise 46% of the sample. It’s the exact same share for Roman Catholics, too. Among both groups, 18% say that it’s “not too important” as well. It’s eerie how much these two groups look alike.

I don’t want to belabor the point much more. Evangelicals and Black Protestants have a very similar posture toward their faith. They have a very high level of belief in God, and place a lot of emphasis on the Bible. They lean on the teachings of their religion to help them make difficult decisions. On the other hand, mainliners and Catholics just score lower on all of these metrics, and often in a pretty significant way.

But what about metrics related to religious behavior like church attendance, prayer frequency, saying grace before meals, and reading scripture? That’s what I calculated in the graph below.

The top two categories are always the same: Black Protestants and evangelicals are more likely to engage in religious activities than mainliners or Catholics. Sometimes the gaps between the first two groups are pretty small. For instance, 72% of Black Protestants and evangelicals pray every day and the share who read scripture weekly or more is just about the same, too. But evangelicals are much more likely to go to church weekly (50% vs 34%), while Black Protestants are eleven points more likely to say grace before meals.

In most of these graphs you can also see that clear break between the top two and bottom two groups. On prayer frequency, mainliners and Catholics are at least twenty points lower than evangelicals and Black Protestants. On reading the Bible weekly, the gap is thirty percentage points. There’s no other way to say this: mainline Protestants and Roman Catholics just don’t do as much “religious stuff” compared to their evangelical and Black Protestant cousins.

To drive that point home in a single graph, I just wanted to visualize all four groups of Christians on two axes. The horizontal one focuses on religious behavior and includes the four items we just discussed. The vertical axis is a composite metric of the belief items that have been previously described in this post.

The white dot that appears in each scatterplot is the group average on both dimensions and the ring indicates where 50% of each tradition falls on both dimensions. Where the two dashed lines cross is where the average Christian fell on both dimensions. So anyone in the top right quadrant scored higher on both religious behavior and belief than the average. Someone on the bottom left is below average on both metrics.

I think this makes it pretty plain—the average evangelical and Black Protestant are in that top right corner while mainliners and Catholics are typically found in the bottom left corner. But I also wanted you to take note of the size of the circles. A larger circle denotes less concentration of respondents in one single spot on the scatter plot. That’s certainly the case for mainliners and Catholics. There’s just a whole lot more diversity of beliefs and behaviors there. For Black Protestants and evangelicals, conformity is way more widespread. For both groups, over half of all respondents are in the top right corner.

And just for fun, a little game which is pretty interesting to play if you are a big time religion nerd. I pulled all the evangelicals and Black Protestants from the dataset and built a simple web app that displays their gender and age along with how they answered a bunch of questions about their religious belief and behaviors. Your task is simple – guess if they are an evangelical or a Black Protestant based on the information provided.

So, if Black Protestants and evangelicals are so incredibly similar on all these religiosity measures, then why do we separate them out? I bet you can probably guess why: it’s politics. RELTRAD tries to recognize all the differences between groups, not just religiosity. The fact that 90% of Black Protestants voted for Harris in 2024 compared to 80% of white evangelicals who supported Trump makes it pretty hard to justify collapsing them into one category. That would obfuscate more than it would reveal.

But, as you can pretty easily tell – evangelicals and Black Protestants may strongly disagree at the ballot box but they often walk in lockstep during Sunday services and midweek Bible study.

Code for this post can be found here.

Ryan P. Burge is a professor of practice at the Danforth Center on Religion and Politics at Washington University.

Some people are incredibly religiously devout, it’s just that they can’t make it to church services for a variety of reasons. They may have to work every weekend, or they have some type of disability that makes it difficult to leave the house. That’s why it’s always helpful to try and measure a concept like religious behavior through a number of ancillary measures. This is what makes the Pew Religious Landscape Survey so helpful. It circles around religious behavior with lots of different approaches.

After working with the survey a whole lot over the last couple of months, I have really come to believe that there are two dimensions of religious behavior that are included. One is a collection of five questions that seem to focus more on private activities like praying, saying grace before meals and reading holy scriptures or other devotional texts. The other is a set of three questions that are centered around the social side of religious behavior: attending religious services in person, going to a prayer group meeting, and talking about religion with someone from another faith background.

The table below should help orient us to both dimensions and give a good sense of the actual survey questions that appeared on the instrument.

I think these are pretty defensible from a measurement perspective. None of the “private” acts require being in the same space as another individual. All three “social” acts necessitate being in the same physical space with another person. It’s really that simple of a dividing line.

Let me start my analysis of both of these dimensions by visualizing the distribution of values across both indexes in the full sample so you can get a sense of which ones are more or less prevalent among a typical American.

The distribution of values on the social dimension is clearly leaning toward the left side of the distribution. Huge shares of Americans can be found between 0 and 2 on this social behavior index. The average in the entire sample was just 3.2 out of 10. As you move further to the right hand side of the graph, it’s noticeable how far this dimension lags behind the private manifestations of religious behavior.

For the private religiosity index, the overall mean is undoubtedly higher – 3.9/10. A key reason for that is that despite the fact that there’s a pretty high peak in this distribution right around 0, it quickly drops off and lags behind the social distribution at the lower ends. As you move above 5, the blue line is consistently higher than the orange one. It’s undoubtedly the case that a larger share of Americans practice private acts of religious activity than public ones.

Let’s do a scatter plot now of all the religious traditions in the dataset so you can see how they line up on both private and social religiosity.

Read more

And expect some of the posts on Graphs about Religion in 2027 to use data collected through this research initiative.

We are excited to announce that January 11th, 2027 will be the next deadline for submissions to the Survey Experiments for the Scientific Study of Religion (SESSR) project. SESSR is supported by a grant from the John Templeton Foundation and provides the opportunity for social scientists to field religion-related survey experiments on a probability sample of U.S. adults at no cost to them. For more information, see www.sessr.org

What do we do with the time that we have? That’s one of the most important questions we face, in my estimation. It’s certainly something I struggle with on a day-to-day basis. I have a really hard time taking a vacation that lasts more than four or five days because I feel a deep sense of guilt that I’m not doing enough with my life. It’s that pesky Protestant work ethic that I will never be able to shake. And don’t try to lecture me about it in the comments, either. Many have tried and many have failed to convince me to work less.

I am, for better or worse, very deliberate about how I spend my time because it’s such a precious commodity. And my fatal flaw is that I think other people are as thoughtful about such matters as I am. So whenever I hear someone say, “I just can’t make time for that,” what I actually hear is, “I am choosing to use my time on something I prioritize more.” I want to believe that people are parsing out their values when they make a decision about going on vacation, attending a concert, or staying at home with their kids.

And I want to believe that lots of people are very thoughtful when they make a decision of whether to attend a local house of worship each weekend. But I also know that this decision is not made purely randomly by each person. There are certain factors that make someone more inclined to get up on Sunday morning, put on some nicer clothes, and make the drive to their local church.

Educated people are more likely to be weekly attenders than those with a lower level of education. That comes through when looking at this data from Pew.

Now, I must admit that the positive association between education and attendance is age-dependent, to a certain degree. For the very oldest Americans, the relationship between the two factors is actually negative: educated folks are less likely to be weekly attenders than those who stopped at high school. But that’s the only age cohort where that’s the case.

For those born between 1950 and 1979, the line goes up in a pretty clear way. From left to right, the increase is six points for those born in the 1950s, nine points for those born in the 1960s, and four points for those born in the 1970s. The last two birth cohorts are a bit more of a mixed bag. For those born in the 1980s, the line is flat. It may be tilting up for the very youngest adults, but I think I am going to reserve judgment on that until more data comes in.

You can see the overall trend, right? Educated folks go to church more often than those with less education. I’ve posted this finding many times on social media. It garners a lot of discussion, as you can probably guess. One common refrain is that those at the low end of the educational spectrum don’t attend as much because they are working and thus there’s a scheduling conflict. It’s a possibility worth some serious reflection, and trying to examine that claim will make up the rest of this post.

Lucky for us, the Pew data included a question about employment status and gave five options ranging from working full-time to being retired.

For full-time workers, those with graduate degrees are just as likely to be weekly attenders as those who stopped at high school. But for part-time workers, the most educated are much more likely to go to church on an average weekend—a full ten percentage points. That’s a massive difference. That ten-point gap is also there among folks who are not working at all. For those with a high school diploma or less, 21% are weekly attenders. It’s 31% for those with a master’s degree.

The graph for those who were disabled is really interesting, but you need to know that part of the problem is a small sample size—less than 800 respondents out of nearly 37,000 in the full survey. The retired sample is much larger and the result is much less spectacular: there’s absolutely no change in weekly attendance rates based on education for those who have retired.

What complicates all of this, of course, is a number of factors. One big one is that education is deeply intertwined with employment status. Simply said: less educated folks are less likely to be in the workforce compared to those working full-time.

Among those who are working forty hours a week or more, 45% have at least a four year college degree. That’s ten percentage points higher than the full sample. But that’s really the only group where the mean level of education is above average. For part-time workers, just 31% have a college degree and 36% stopped at high school.

But look at those who aren’t working: 46% have no more than a high school diploma. That’s double the proportion of those who have earned a four-year degree at 23%. It’s even more severe among the disabled, where 66% stopped at 12th grade and just 7% have a college degree. Among retirees, the educational breakdown doesn’t deviate that much from the full sample. Retired folks are just slightly less educated than the average American.

But the reason for that deviation is easily explained: people who retire are typically older. They were raised at a time when not that many people were being encouraged to attend college. You see how all these things are compounding on each other now? It would be simple to test a theory about education and attendance by just looking at those who aren’t working. Those folks are not impacted by the “my work schedule makes it tough to go to church” problem. But those two subgroups have unique characteristics that could muddy our results.

Here’s why just focusing on the impact of education on attendance among retired folks is a bad idea. The most significant predictor of weekly attendance is the age of the respondent.

This is a really simple regression model where I held education and employment status constant and just focused on the likelihood of each generation reporting weekly attendance. This graph is about as clean as it gets: older folks are much more likely to be weekly attenders than those who are younger.

Among the very oldest respondents in the survey, nearly 40% are going to church on a weekly basis. You get to people born two decades later and attendance already dips to 25%. Among the very youngest cohorts in the data, weekly attendance rates are down to just 17%. So, if you only run your model on retired folks and you see that education does or doesn’t have an impact, you’ve got to realize that it’s age confounding the whole thing. We really need to control for age more than anything else.

So we can’t focus on retired folks because there aren’t many young retired people in the sample! It would show us nothing of value, really. So instead, I wanted to look at just three levels of employment: those working full-time, those working part-time, and those not working at all. But I am also throwing in a control for age here, so we are taking out the biggest confounding variable.

Notice which employment situation has the smallest gap between the two estimates of weekly attendance? It’s those who are working full-time. A respondent with a graduate degree working full-time is only two percentage points more likely to be a weekly attender than someone who works full-time but only has a high school degree or less. That’s not a huge substantive difference. But that’s not the case for the other two scenarios.

Among people with a high school diploma working part-time, 28% are weekly attenders. It’s 36% among those with a graduate degree. For those who are not working at all, a highly educated person is ten percentage points more likely to be a weekly attender compared to someone who stopped at high school (33% vs. 23%). These are gaps that are both statistically significant and substantively large.

Let me just reiterate this point: when we got rid of the problem of scheduling, there’s still an 8–10 percentage point gap in weekly attendance rates among the most and least educated. So, the logistical explanation just can’t explain that away. That’s not to say there aren’t other variables to consider. For instance, someone with a high school diploma who is working part-time or not working at all may have a problem with transportation. I can test that by including income as a control variable in the same model.

But that explanation doesn’t actually close the gap in any meaningful way. Once I added a control for income, the differences in the estimates between this model and the previous one barely changed. For those who aren’t working at all, the difference in the two estimates is still huge. For those with graduate degrees, 39% are weekly attenders compared to 29% among those with a high school diploma who are not working at all. That ten-point difference is the same as the prior graph.

And what may be even more interesting is that someone with a high school diploma working full-time is just as likely to be a weekly attender as someone who isn’t working at all. And that’s while holding income constant. So I just don’t see how we can chalk this up to an inability to get to church.

There’s one other explanation that could make all of this wash out, of course: maybe these two groups have a totally different level of enthusiasm about religion. Directly stated, “Maybe educated folks are going to church more because they are just more religious than those with a high school diploma or less.” That’s easily testable using other questions from the Pew data.

This possibility falls flat, too. If you look at the share of each level of education who say religion is very important to them, it’s essentially the same. If you compare prayer frequency between the two groups, there’s no substantive difference at all. You do see that those with less education are seven points more likely to say the Bible is very important to them, but then those with graduate degrees are seven points more likely to say they are very religious compared to those with a high school diploma.

Doesn’t that seem kind of like a wash to you? Because that’s the impression I get from this data. I don’t think we can really say that one of these two groups is just more enthusiastic about religion than the other one. I just don’t think religiosity is a ready-made explanation, either.

None of the obvious explanations hold up. It’s not that less-educated Americans have less free time. In fact, there’s a bunch of time-use data that actually shows the opposite: the less educated actually have MORE leisure time than those with college degrees. It’s not that work is keeping them away; the gap shows up (and grows) the moment work stops, not the other way around. And it’s not because the less educated are just not that religious, because on belief, prayer, and self-described devotion, the two groups look almost identical.

What’s left is something harder to measure and probably harder to fix: showing up. Whatever it is that turns private faith into public practice, it isn’t distributed evenly across education levels and that gap seems to widen, not narrow, once the excuse of a job is gone. That’s worth some very serious reflection.

I opened this by saying “I don’t have time” usually means “I’m choosing something else.” I expected that logic to hold here—that the education gap in attendance was really a gap in priorities. It isn’t. These two groups want the same thing, believe the same thing, and pray at the same rate. The gap isn’t in what they value. It’s in what they do about it. That’s a stranger, and frankly more unsettling, finding than the one I went looking for.

Code for this post can be found here.

Ryan P. Burge is a professor of practice at the Danforth Center on Religion and Politics at Washington University.

This post has been unlocked through a generous grant from the Lilly Endowment for the Association of Religion Data Archives (ARDA). The graphs you see here use data that is publicly available for download and analysis through link(s) provided in the text below.

I don’t listen to podcasts that much anymore, but I still have a couple that I check out on a regular basis. One that’s still in my queue is from the New York Times and it’s called The Interview. You can probably guess the format of the program. What I appreciate about it, though, is that their guests don’t just fall into simple categories. I mean, they do have actors and musicians like everyone else. But they also interview people that may not be as famous. For instance, they had Father James Martin appear on the program back in February.

I really appreciate when they bring on someone who is really well known in a specific corner of the world, but is completely foreign to me. That was the case about 18 months ago when they interviewed Dr. Lindsay C. Gibson – a name I had never heard of. She wrote a book called Adult Children of Emotionally Immature Parents: How to Heal from Distant, Rejecting, or Self-Involved Parents. Here’s what is astounding about the book: it has sold more than 1.5 million copies. And I had no idea it had even existed.

The book is more than a decade old, but it’s seemed to get a second life on social media. Some accounts will post quotes from the book and then overlay them with text exchanges with their parents that perfectly illustrate what Dr. Gibson is talking about.

It seems like a huge part of the discourse surrounding people my age is how they are dealing with the relationship they have with their parents. That book and Dr. Gibson came to mind when I was scrolling through a recent survey posted on the Association of Religion Data Archives website – the Pew American Trends Panel Wave 137. The stated purpose was, “to evaluate the relationships between parents and their adult children.” That seems fun, right?

What I really like about this panel is its overall design. There are basically two subgroups inside the survey: U.S. adults who are either a parent of at least one 18- to 34-year-old or who are themselves ages 18-34 and have at least one living parent. So we can actually check perceptions on both sides of this debate using one data source.

For instance, one question to the younger sample was: “Overall, to what extent do you think your parent(s) prepared you to be an independent adult?” Parents were asked how well they had prepared their children to be an independent adult, too.

The broad finding is that parents overwhelmingly felt like they did a really good job in preparing their kids to be adults. I mean, across the board that sentiment is shared by at least 80% of parents. Which makes sense, doesn’t it? Very few parents want to admit their helicopter tendencies made them feel better at the expense of their own child’s independence. Atheist and nothing in particular parents score a bit lower than this, but they are still at 80% in the “fair amount” or “a great deal” options.

On the other hand, the young adults in the sample are not as inclined to feel like they were raised in a way that taught them independence. The group that scored the absolute lowest on this was those who claimed no religion in particular. Just 53% of these people said that they were well prepared to enter into adulthood.

I have a theory about why that’s the case based on my other work, The Nones, where I did a deep dive into the non-religious. The “nothing in particular” folks are often at the very bottom of the socioeconomic spectrum. And I am wondering if these young adults felt that their parents were largely absent from their childhood because they were constantly working or worrying about making ends meet.

However, the biggest takeaway of all for me is the gap between parents and children on the level of preparedness for being an independent adult. You can see it in the first graph, but I wanted to make it even simpler.

Among evangelical parents in the sample, 53% of them said that they did “a great deal” to make their children independent adults. Evangelical children were not so enthusiastic. Only 28% chose the “great deal” response. That’s a total gap of 25 percentage points, which is the largest of any religious group in the data. What’s even more interesting are the two groups in second and third place: agnostics and nothing in particular with gaps of 17 and 13 points respectively. It’s very rare to see evangelicals be so close to the nones on any metric.

Among mainline Protestants and Catholics, the gap was much smaller: 9 points for the former and 7 points for the latter. The big reason for that is that the young adult sample was much more inclined to say that they were very well prepared for being an adult. One interpretation from this data is that evangelical young adults may have felt sheltered by their parents. However, I don’t think that same concept carries over into agnostics and nothing in particulars. So I really don’t know what to make of this.

There were other ways to explore the friction between parents and kids, though. The survey included a battery of questions to the young adults that started with this preamble, “How often do you and your parent(s) disagree about each of the following?” and then included eight options ranging from financial choices, to religion, to raising children.

The overarching conclusion that I came to when looking over all these results is that kids just aren’t arguing with their parents over that much. There’s only one square in the entire heatmap where more than 20% of respondents chose the extremely or very often option. In most cases it was in the low teens. So, the narrative about a lot of conflict between parents and their adult children just doesn’t find empirical support here.

There are a couple of findings that I do want to flag, though. One is that evangelical young adults aren’t fighting with their parents that much. In most cases it was 10% or less across the full range of issues. For mainline Protestants, there were a number of questions where the percentages were in the mid-teens like raising children, career, or romantic partners.

It’s pretty clear that the nones tend to have a lot more friction with their parents. Especially on the left half of the heatmap. Topics like religion, politics and career elicit a lot of disagreement among atheists and their parents. There’s also friction when it comes to the choice of romantic partners for agnostics and nothing in particular.

The results from the prior analysis got me wondering if a lot of the friction among the nones was because they were raised religious but then became atheist, agnostic, or nothing in particular as they moved into adulthood. Unfortunately, there weren’t any questions on the survey to assess that line of inquiry. But there was a question that could hint at that possibility. It asked young people, “Would you say you can be your true self around your parents?”

It’s absolutely fascinating to me which groups score the lowest on the “all the time” metric. It goes in this order: atheist, other religion, evangelical, agnostic. That’s pretty usual, right? I do think that this is pretty clear evidence that young adult atheists probably don’t feel super comfortable about their parents because there’s a religious mismatch. That makes sense for agnostics, too.

But how do we make sense of that evangelical result? I don’t really know what to make of this one. One possibility is that their brand of evangelicalism may be a bit more moderate than their parents? When I was at an evangelical college a lot of my friends would tell stories about how they would have to hide their new tattoos or proclivity to say a swear word or two from their parents who were much more buttoned up. Maybe that’s what is going on here. It’s surely not the case with mainliners or Catholics.

I wanted to finish up here by focusing on a three question battery that asked those young adults how often they actually communicated with their parents in three ways: phone/video calls, text messages, or seeing them in person. I thought this provided a good proxy to answer the question: do these young adults have a good relationship with their parents?

I was really surprised to see which of these questions generated a clear difference in response rates and which ones didn’t. For instance, there’s just not a ton of variation when it comes to texting parents once a couple times a week. About two-thirds of Catholic young adults text on a regular basis compared to 56% of agnostics. Those differences aren’t statistically significant.

But notice which groups are always on the top and bottom of each graph? Catholics are always at the top when it comes to communicating and agnostics are at the bottom. The gap on phone/video calls is huge between the two groups: a total of 26 percentage points. There’s also a bit of a divergence when it comes to visiting in person. A Catholic is almost three times more likely to see their parents on a weekly basis compared to an agnostic.

I tried to poke around on the question of why Catholics were so much more communicative. At one point I thought it was because a huge share of Catholic young adults in this sample were Hispanic. That drove a big difference in seeing their parents in person but Hispanic Catholics were actually less likely to call or text their parents compared to white Catholics. So it doesn’t explain everything.

Looked at in totality, there are some clear patterns that emerge. Evangelicals, agnostics, and nothing in particular young adults are way less likely to say that they were well prepared to be an independent adult compared to their parents. These three groups were also more likely to say that they couldn’t be totally comfortable around their parents, either. That’s something to really think about.

If I were to use these results to point toward a religious group where parents and children have a better relationship, it’s Catholics. As the last graph showed – Catholic young adults always lead when it comes to talking to their parents on a regular basis. Catholics run neck and neck with mainline Protestants on feeling comfortable around their parents and not having lots of friction around difficult issues, too.

Does religion cause this directly? It’s hard to know. It’s easy to posit that certain religious beliefs are intertwined with specific parenting styles. Which can sometimes generate hard feelings and distance when those children move out of the house. But switching also has to play a role here, too. Kids who were raised evangelical but became agnostic are likely going to clash with their parents more frequently than young people who have been Catholic their entire life.

But, above all, I don’t see any strong evidence here that animosity is widespread between parents and children. Young adults generally feel that they can be themselves around their parents and the amount of conflict I see in the heatmap is pretty minimal.

This is probably another case of social media amplifying the minority view. Think about a video where the person says, “I generally like my parents. We fight a bit here and there, but I really do feel grateful for having a relationship with them.” That’s probably not going to get a lot of shares, right?

Code for this post can be found here.

Ryan P. Burge is a professor of practice at the Danforth Center on Religion and Politics at Washington University.

Said simply, about a quarter of Americans report that they attend a house of worship at least a couple of times per month, and about forty percent are unlikely to go to a church, synagogue, or mosque at all in the next 12 months.

We measure American religion through three dimensions: belief, belonging, and behavior. It’s something I feel like I write about constantly, but if you need a quick primer, I’d recommend you peruse this piece: What Does It Mean to be ‘Religious’? — then come back to today’s post, which is going to be a deep dive into religious behavior.

Social scientists almost always measure behavior through the lens of religious attendance. It’s just the simplest, most straightforward way to do it. Nearly every survey includes the question, and it’s easy for a respondent to answer. But there are other metrics that tap into behavior, too. That’s what I want to explore today.

The Pew Religious Landscape Study has a wealth of questions about religious behavior in all kinds of contexts beyond just going to church on Sunday.

According to Pew’s data, about a quarter of Americans indicate weekly religious attendance in person. (That’s exactly the same as the General Social Survey, by the way.) At the same time, the share of Americans who indicate that their attendance is never/seldom is 50% of the sample. Online attendance is even lower: two-thirds report that they never or seldom attend a religious service via the internet. In contrast, 16% are watching every week.

Prayer frequency is another way to measure religious behavior, and this one specifically asks folks to exclude praying as part of religious services. Over a quarter of Americans say they pray multiple times per day, and another 16% say they pray by themselves every single day. In total, a majority of Americans say they pray at least a couple of times per week.

In contrast to religious attendance, there aren’t that many adults who say they don’t pray much at all — 16% say they never pray and the same share describe their prayer frequency as seldom. So a third aren’t praying much at all.

But then we’ve got all these other kinds of religious behaviors that offer a fascinating window into other ways of religious practice.

It’s readily apparent here that the vast majority of Americans are not engaging in religious dialogue with people outside their faith. In the full sample, less than one in five say they do this on a monthly basis or more. About the same share (19%) report that they attend a prayer or scripture group at least once a month. I feel like I have to point out that three quarters of all adults say they never or seldom attend a prayer or scripture group.

There are other behaviors that happen with a bit more frequency. When it comes to reading holy books or religious devotionals, 30% of respondents said this happens at least once a month. However, about 40% say this never happens in their daily lives.

It’s hard to look at this graph and think the average American is engaging in any type of religious behavior on a regular basis — just the opposite, in fact.

To try and simplify this, I created a “religious behavior index,” which is just a metric that combines all the above behaviors into a single scale running from 0 (meaning no behaviors at all) to 10 (those who indicated participation in each behavior at the highest frequency). I visualized the distribution of responses across all 37,000 survey participants below.

I broke the sample down into quintiles (those who score in the bottom 20%, next 20%, and so on). You can see those distributions based on the dashed lines and color changes. For instance, 20% of Americans scored between zero and .6 on this ten-point scale. In fact, about 7% of the sample scored exactly a zero. The mean score for the entire sample was 3.73 (weighted), which gives the clear impression that the typical American isn’t engaging in a whole lot of behaviors.

As you can probably tell, it doesn’t take a high score to end up in the top quintile — just 6.8 out of 10. In fact, only 3.7% of the full sample scores at least a nine on this index. The average American is just not that religious at all.

I’m sure you’d like to know where you’d land on this distribution, right? To that end, I’ve put together a simple online quiz you can take in about 90 seconds. It’s just the eight questions you see above.

Let’s do some data work, then, with this religion index. I first wanted to show you how the quintiles are distributed based on the birth decade of the respondents.

Anyone who has hung around here very long should have a good sense of how this graph would look before I pulled it together. Older Americans engage in a whole lot more religious behaviors than younger Americans. For the very oldest respondents, about a quarter landed in the top 20% for religious behavior, and about the same share landed in the next highest quintile. Those percentages drop with each subsequent decade of birth.

You can see that the blue bars get shorter in a methodical and consistent way as you move down the graph, and it really hits rock bottom for respondents born in the 1990s — just 14% are in the top category, and double that share are in the bottom grouping (29%). We do see a bit of a rebound among folks born in the early 2000s, but let’s not overinterpret this. A lot of these respondents still live with their parents or still have a high level of attachment to their parents’ religion. I’d expect their distribution to look a lot like the older cohorts in the next 5–10 years.

What religious groups have the highest level of religious behavior? That’s below.

Without a shadow of a doubt, it’s Jehovah’s Witnesses. That’s a group I’ve written about before with this title: Politically Neutral, Religiously Intense. That fits them to a T. According to the data from that piece, about 70% of JWs say they go to church every week (with nearly 60% indicating weekly+ attendance). The Jehovah’s Witnesses are followed by Latter-day Saints at 6.6 out of 10. If you want to know which groups are the most religious — LDS and JWs are the most empirically defensible answer.

The cascade of the rest of the groups follows along with what many would have guessed: Black Protestants and evangelicals land in basically the same spot (around 6 out of 10). Catholics and mainline Protestants are way lower than that, though — they score 3.8 out of 10. What’s interesting to me is that the average Catholic or mainline Protestant is essentially as religiously engaged as the average American (which includes a whole bunch of non-religious folks).

Speaking of the nones, I thought it would be helpful to pull them out and visualize the distribution of their behaviors separately. When I calculated the mean score for each group, it was less than 1 for both atheists and agnostics, and only 1.5 for the nothing-in-particulars.

But the quintiles give us a much better sense of where these groups land. I like this visualization because it gives us a nice sense of how not all nones are created equal. Atheists are clearly the least likely to engage in religious behaviors: almost 70% of them score below .6 on this scale that runs from zero to ten. Among agnostics, 56% are found in the bottom bucket. For nothing in particulars, it’s still pretty high at 45%, but it’s less than half.

But here’s what I want you all to think about on this graph — the right side of the distribution. Among atheists, just 3% score “moderate” or above on the religious behavior index. For agnostics, it’s just a little higher: 12%. There’s a strong contingent of nothing-in-particulars who score in one of the top three quintiles, though: 26% of them. And remember this — about 60% of all the “nones” in the United States are nothing in particular, not atheist/agnostic.

I’ve become fond of saying that religious behavior is the type of religiosity that goes first. Remember — half of all Americans report that they attend religious services seldom or never. The share who identify as non-religious? About 30% of the country. And the percentage of people who indicate they have an atheist or agnostic belief in God? Only 15% in most surveys.

The unfortunate part of this equation is that religious behavior is more than likely the most beneficial type of religiosity that exists. The data indicates that those who engage in religious activities frequently have a longer life expectancy, are less likely to suffer from mental illness, are more politically engaged, are more likely to acquire civic skills, and are more tolerant of views they disagree with (among all other kinds of positive benefits). Yet it’s the type of religiosity that’s fading the fastest.

In a future post, I’m going to do an additional layer of analysis on religious behaviors by splitting them into personal vs. communal activities. Going to a Bible study is something entirely different than just reading the Bible by yourself, right?

I’m going to sort all of that out really soon.

Code for this post can be found here.

Ryan P. Burge is a professor of practice at the Danforth Center on Religion and Politics at Washington University.

This post has been unlocked through a generous grant from the Lilly Endowment for the Association of Religion Data Archives (ARDA). The graphs you see here use data that is publicly available for download and analysis through link(s) provided in the text below.

Who do you talk to when you’re struggling? When it feels like you’re carrying the weight of the world and there’s no way out of the mess you’re in—who can you turn to for an honest conversation? I’d bet that for a lot of us, the answer is “no one.” Many Americans are suffering in silence, without a spouse, a close friend, or even a co-worker they can lean on when times get tough. The loneliness epidemic has real casualties. Of that, I am absolutely sure.

I saw a video on Reddit a couple of weeks ago that got me thinking about this. It was posted in “HumansBeingBros,” a subreddit that exists to make you feel a little better about the world. A motorcyclist comes upon a man walking down the middle of the road, clearly in distress. The rider stops his bike, turns around, and engages with the man—at one point even asking if he needs a hug.

Later in the video, the two of them are sitting against a fence on the side of the road. Here’s the part I love: whoever uploaded it blurred the man’s face and captioned the moment with just a few words—”talked about his struggles.” No details. I can really appreciate that. We don’t need to know what the man was dealing with. He deserves his privacy. The video ends with the police arriving to offer him a ride home.

But it left me with a question: who can people actually talk to in an open and honest way? I’ve got some data from Pew that speaks directly to this. It’s a simple battery of questions from American Trends Panel 142, hosted on the ARDA. The preamble is straightforward: “How comfortable would you be talking about your mental health and emotional well-being with each of the following?” Then respondents rated several options:

• Close Friends

• Spouse/Partner

• Immediate Family Member

• Mental Health Therapist

• A Co-Worker

• Neighbor

• Religious/Spiritual Leader

Let’s start with the share of each religious group who said they were extremely or very comfortable talking about their mental health with each of those individuals.

One thing that is crystal clear from this data is that people truly rely on their spouses or partners when it comes to being honest and open about what’s happening in their lives. Among every religious group, the spouse scores the highest, often by a wide margin. The response levels are remarkably consistent, too: most are within a few points of 80%. An interesting outlier is the LDS community—nearly 90% say they are very or extremely comfortable sharing mental health issues with their spouse.

The other groups follow a consistent rank ordering. Close friends, mental health therapists, and immediate family members all rank similarly, typically in the 50-60% range for most religious traditions. At the bottom are co-workers and neighbors, who score poorly; people generally aren’t willing to be vulnerable around them.

But there was one blip in the data worth digging into: the responses regarding comfort levels with religious or spiritual leaders. For most respondents, this doesn’t rank high. Only about a third of non-evangelical Protestants, Catholics, and LDS members feel very or extremely comfortable speaking with their pastor about their struggles. But for evangelicals, that number jumps to 56%—putting clergy in the same neighborhood as immediate family or close friends.

That outlier led me to compare two specific options: speaking with a mental health therapist versus a member of the clergy. I wondered if certain religious groups favored one over the other. The graph below explores that comparison.

It’s clearly the case that the average person of faith feels more comfortable talking with a counselor than a member of the clergy. Among Catholics, half feel very or extremely comfortable talking to a mental health professional, while only 33% say the same about talking to a priest. For LDS members, those numbers are 52% and 25%, respectively. (This may be related to how the LDS Church is organized, as they do not have a traditional paid clergy). Across almost every group, there’s a clear throughline: people of faith aren’t always keen on talking to their pastors about their troubles.

However, evangelicals are the exception. They are actually slightly more comfortable talking with their pastor (56%) than with a therapist (51%) regarding their emotional well-being.

Part of this may stem from the fact that evangelicals attend church more frequently than other Christians. It stands to reason that being around someone more often makes you more willing to share your struggles. I can test this by breaking down the three largest groups by attendance levels.

One consistent finding is that church attendance has very little impact on how comfortable someone is speaking to a mental health professional. Across Protestants and Catholics, about half say they are very or extremely comfortable speaking with a counselor. Aside from low-attending Catholics, these differences fall within the margin of error.

However, frequent church attendance does increase comfort levels when speaking to a pastor. For evangelicals, comfort increases from 44% among infrequent attenders to 63% for weekly attenders. For non-evangelical Protestants, it rises from 27% to 49%. For Catholics, the jump is even larger: 19% vs. 57%. Clearly, more religiously active people are more likely to see their pastor as a resource for their mental health journey.

It is worth noting that among weekly attending non-evangelicals and Catholics, the comfort levels for therapists and clergy are nearly identical. For weekly attending evangelicals, however, the preference for their pastor over a therapist is clear.

I also wondered if age played a role in whether Christians are comfortable speaking with mental health professionals. Recent commentary suggests young adults are “The Therapy Generation”—seeking counseling at much higher rates than previous generations. This would suggest that younger Christians might feel more comfortable seeking out a counselor.

For evangelicals, there is a higher level of comfort among young adults. Of those aged 18-29, 55% are very or extremely comfortable talking to a mental health professional, compared to 48% among those aged 50-64. It appears the stigma is weakening within this group.

However, this trend isn’t as clear for non-evangelical Protestants and Catholics. There isn’t a smooth age gradient in either group. Among Catholics, for example, those aged 50-64 are the most comfortable with therapy. For non-evangelicals, it is the 30-64 age range.

The variation across these religious groups is difficult to pin down.

Many factors could influence openness toward a therapist or clergy member, including gender, race, education, income, and partisanship. To test this, I used a regression model to control for these factors simultaneously, looking at comfort levels with both mental health professionals and religious leaders.

There’s a lot to dissect here.

Regarding gender, men are slightly less comfortable talking to a therapist, but it makes no difference when talking to a religious leader. Race also has a differential impact: white respondents are generally more comfortable speaking with a therapist than with clergy. Age works similarly in both models: older individuals are more open to discussing mental health with both counselors and clergy (though the effect is more pronounced for the latter).

The most significant factor is religious attendance, which strongly correlates with comfort in speaking to clergy. Surprisingly, it remains a positive and significant factor for talking to a therapist as well. While identifying as evangelical increases openness with a pastor, it has zero impact on comfort levels with a therapist.

The variable that stands out most is Republican identification. It works in opposite directions depending on the confidant. A Republican is significantly more comfortable talking to a member of the clergy about their mental health, but significantly less comfortable opening up to a therapist. In fact, partisanship is the strongest negative predictor of therapist comfort in the model—surpassing gender, race, and education.

This holds true even after controlling for religious attendance and evangelical identification. It isn’t just that Republicans are more religious; even two people in the same pew every week will differ—the Republican remains measurably more wary of the therapist’s couch than the Democrat. Comfort with mental health care has become another facet of political sorting in the U.S.

Why Republicans are more averse to therapists is a question for another time. My more pressing concern is for the man in that video. Did he have no one in his life he could talk to? What would have happened if that motorcyclist hadn’t driven by?

Social atomization has real consequences, both seen and unseen. And as this data shows, the question isn’t just whether Americans have someone to open up to — it’s that even the answer to “who would you trust?” now depends on their politics, their education, and their pew. The doors people are willing to walk through are different. What matters is that they walk through one.

Code for this post can be found here.

Ryan P. Burge is a professor of practice at the Danforth Center on Religion and Politics at Washington University.

But regardless of the actual provenance of the words – the ideas here are incredibly powerful because the sentiment expressed is the backbone of pluralism. Under this model, tolerance is the glue that holds societies together. We may not agree on everything (or almost anything), but what we do have to hold in common is the view that people have the right to speak their minds without fear of violence or criminal reprisal.

In other words, free speech needs to be free. And this is a concept that is tested on college campuses thousands of times per academic year when student groups invite speakers to give addresses, lectures, or colloquiums on hot-button issues. Should someone be able to come to an institute of higher learning and advocate for a very fringe belief? Or is it necessary and prudent for students to stop that event by whatever means necessary?

That’s tested with a short battery of questions that were included in FIRE’s 2026 survey of over 68,000 college students from more than 250 colleges and universities all across the United States. Let me just start by giving you a broad sense of where students stand on the acceptability of suppressing free speech.

When they were presented with the statement about “blocking other students from attending a campus speech”, 22% of respondents said that action was sometimes or always acceptable. On the issue of using violence to stop a speech, that figure was 15%. So there is certainly a small but not marginal group of college students who are willing to engage in what I would describe as extreme measures to stop a speaker from voicing their opinions.

The third statement was, “shouting down a speaker to prevent them from speaking on campus” and received much more support than the other two. In fact, only 28% of respondents thought this was never an acceptable course of action and 35% said it was rarely acceptable. That leaves 37% of college students who believe that shouting down a speaker is sometimes or always acceptable.

But let’s break those three statements down by the political ideology of the college student who took part in the survey.

One thing that comes through in this analysis is that conservatives are more likely to say that blocking students from attending a speech or shouting down a speaker is never acceptable on their campus. For instance, in the “shouting down” scenario, 39% of conservative students said that this should never happen, it’s only 21% of liberal students. Just 28% of conservatives thought it was a behavior that is sometimes or always acceptable. Those two response options were chosen by 42% of liberal students.

What is noteworthy, though, is how ideology mattered so little when it came to the question about “using violence to stop a campus speech.” Whether I analyzed this among liberals or conservatives there was no real difference in the outcome: about two-thirds said that this kind of behavior was never acceptable. That was easily the highest share for any of the three scenarios. It’s pretty strong evidence that college students see a clear distinction between other forms of obstruction and actual physical violence when it comes to stopping a campus event.

Okay – I want to really zoom in on that question about “shouting down” a speaker, though. It clearly has the largest variation of the three. And, from my perspective, it’s the one that I seem to see the most often on social media. And let’s bring religion into the mix, now. Which traditions are the most willing to shout down a speaker and which ones aren’t?

What we see in the graph is that atheists are the most likely to say that it’s sometimes or always acceptable to shout down a speaker, nearly half of them (45%) fall into these two camps. That’s about eight percentage points higher than the full sample. Then, as you go down the list you have groups like agnostics, Jews, and Muslims. Each group is above the overall mean on this question.

Which groups are the least supportive of shouting down a speaker? It’s Latter-day Saints. Among the LDS, 38% say it’s never an acceptable behavior – that’s the highest of any religious group. Then you have the two biggest groups in the survey, Protestants and Catholics. For each, at least two-thirds say that shouting down a speaker is never or rarely acceptable.

But if you have been around this newsletter for any length of time you can make a pretty easy inference about the graph – it’s really political ideology that is doing the “work” here, not really religion. Atheists tend to be the most left leaning, followed by agnostics and so on. Then, LDS are more right leaning and so are Catholics and Protestants. So, this isn’t really about religion, it’s more about politics. So let’s just pull both those factors apart in the graph below.

This is where some really interesting patterns begin to emerge for me. Yes, it is certainly the case that liberals are more inclined to say it’s acceptable to shout down a speaker compared to conservatives. Oftentimes the gaps between the two subgroups are sizable. The largest one, interestingly enough, is among agnostics. Among the liberals, 46% say shouting down a speaker is acceptable. It’s half that rate among conservatives at 23%. But there are big gaps for Catholics and Protestants, too. It’s only among the smallest religious groups where the differences aren’t significant, and that’s more about statistical power than anything else.

But one thing I did want to flag here is comparing liberal Protestants and Catholics to liberal atheists and agnostics. Note how the latter group is more inclined to believe that shouting down a speaker is acceptable compared to the Christians? So, yes, there is something going on here with political ideology, but it is tempered (at least in some measure) by the religious tradition of the respondent.

I wanted to go a layer down on this, though. I was specifically interested in exploring the idea that Christians who go to church more often may hold to views of free speech that lean toward toleration instead of suppression. The very first paper I ever published found just that—regular churchgoers expressed higher levels of political tolerance compared to those who attended less frequently.

The data in that regard is pretty mixed. The throughline through this analysis is still replicated here, though: liberals are more inclined to believe it to be socially acceptable to shout down a speaker on campus compared to conservatives. The gaps between the two groups run between ten and fifteen percentage points among Christians. So, this is really a story about ideology.

But what about attendance? Well, among college students who are conservative, Christian, and attend church weekly – they are least likely to find it acceptable to shout down a speaker on campus. Only 23% indicated it was sometimes or always acceptable. But for liberal Christians, there’s really no discernible pattern when it comes to attendance. A never attender is at the same spot on the scale as a weekly attender. So I’m not sure that going to church more often automatically leads to more pluralistic views. It’s certainly conditioned on the ideology of the respondent.

There are just so many moving parts here that I’ve not explored to this point. What about the race of the respondent, their gender identity or their sexual orientation? I wanted to try (as much as I could) to really sort out what mechanisms lead to students being more open to free speech on college campuses. To do that I created an additive index of three questions from the battery I described at the beginning. Higher scores indicate a greater level of support for free speech.

Let’s work through the results of this regression model in sections, starting with religion at the very top. The reference category here is Protestants, so if the coefficient is positive that translates to “more protective of free speech compared to Protestants.” There’s only one group that this applies to: students who identify as nothing in particular. Muslims, Catholics, Buddhists, and Hindus are all less protective than Protestants.

When it comes to politics, the result is pretty simple – when compared to Democrats, Independents are less supportive of free speech. But here’s a big surprise: in this model, there’s no difference in support for free speech when comparing Republicans to Democrats. But there’s a very good reason for that when you look down to the next section. It’s not Democrats who are less protective of free speech, it’s liberals. In all the models I ran, a conservative ideology was the most powerful predictor of protecting free speech rights. When you put both party and ideology in a model, it’s ideology that emerges.

Now, the bottom half of this graph really helped to clarify my thinking on this topic. Let me explain. Non-binary people are negatively signed – they are not as protective of free speech rights as men. The next set of coefficients indicates that people of color are also signed negative compared to white respondents. Then the section below is on sexual orientation. A non-straight person is also more inclined to find it acceptable to shout down speakers/stop someone from attending a speech than a straight person.

Do you see the pattern that emerges here? It’s incredibly consistent. The “majority” identity is more apt to protect free speech than the minoritized positions. A white, straight, cisgender person is the most protective of someone’s right to give a lecture on campus. Anyone who doesn’t fit into that mold is less supportive.

One has to wonder if the subtext of all this goes back to the old phrase, “It all depends on whose ox is being gored.” In other words, when a transgender college student is thinking of these scenarios in their minds, they are picturing that hypothetical speaker as one who says things like, “transgenderism is a mental disorder” and that’s why they are more willing to restrict speech.

So I’m not sure it’s liberalism per se that is what is leading to less support for free speech. It’s also probably tied up (at least in part) with the fact that lots of liberals are from these “minoritized” groups or at least allies to people in these groups.

I am reminded of Oliver Wendell Holmes’s dissenting opinion in United States v. Schwimmer. The case, decided in 1929, involved a woman who was seeking citizenship. However, she was a pacifist and was unwilling to affirm the section in the citizenship oath that required her to “take up arms personally” in defense of the country. Her application was denied, and she appealed to the Supreme Court. In a 6-3 decision, the Court affirmed that she could not become a citizen of the United States. Holmes vehemently disagreed, writing, “Some of her answers might excite popular prejudice, but if there is any principle of the Constitution that more imperatively calls for attachment than any other it is the principle of free thought—not free thought for those who agree with us but freedom for the thought that we hate.”

May we strive to live up to that ideal.

Code for this post can be found here.

Ryan P. Burge is a professor of practice at the Danforth Center on Religion and Politics at Washington University.

Others could hear that same sermon and can’t wait for the altar call to begin — not so they can head down front and make a public commitment to Jesus, but so they can get out of there as soon as possible. As they hear the pastor pound the pulpit and speak with such resolve, the question that keeps coming up in their head is, “How is this guy so sure that he’s right?” These are folks who are often consumed by doubts and second-guessing, and they struggle to know what “right” is in many situations because of their inability — or unwillingness — to just embrace certainty.

I think about the nature of belief a lot and have always wished I could wrap my head around it in an empirical way. Asking questions in this area is hard. No one ever seems to like the response options to a query about what folks believe about God, or evolution, or the existence of the soul. We all hold positions that seem to land between two response items instead of being perfectly captured by one of them.

Well, I was tinkering around in the Pew Religious Landscape Survey the other day and came to the conclusion that it contains more questions about belief than anything else I have available, and it was fielded to a larger sample to boot. I was wondering if I could begin to model dimensions of faith — hoping to clearly identify whether there are two, three, or four ways that people seem to orient themselves to belief items.

I think I’ve stumbled upon something that may be helpful in understanding belief among American adults.

Here’s how I did it. I pulled ten belief items into an exploratory factor analysis, which is a statistical technique for understanding how a group of items “hang” together. What I ended up with is two dimensions that made the most sense.

The first dimension is what I’ve come to call “doctrinal commitment.” The items with the darker blue circles all loaded on the same factor and it was pretty clear they were focused on propositional claims — the kinds of questions that show up in creeds and catechisms. Folks who are sure that people who don’t believe in God can’t be moral will score high on this dimension. There were six items that fit the bill here.

The second dimension focuses on what I’d call an existential understanding of faith. To be a little colloquial, these are the “woo woo” aspects of religion and spirituality. They tend to focus on the idea that we all have spirits and that we can’t be entirely certain God really exists in a concrete, definable way. When I looked over the data, I got the distinct impression that these are folks who would say things like, “I just feel that we are all connected somehow.” People who load high on this dimension have a softer and more diffuse perspective on faith. They aren’t bound by a specific doctrine or creed.

Read more

We fielded our survey through Qualtrics in August of 2025 and we managed to collect a total sample of 557 respondents. Not bad, right?

One concept that kept bubbling up during our discussion was the idea of self-improvement. Young folks have a term for this: “maxxing.” For instance, some young men are devotees of “jaw-maxxing,” which is a series of facial exercises to make them have a more defined jawline. There is also a corner of the internet that is obsessed with sleep-maxxing. These folks track their sleep patterns obsessively, always looking for ways to catch a few more minutes of shut-eye.

It’s pretty clear that the typical American is always on the lookout for ways to improve their lives on a whole bunch of different dimensions. We wanted to focus on that topic in a more philosophical way in our questionnaire. We asked them: How close do you feel you are to being the kind of person you ought to be, morally speaking?

Our goal here was just to get a sense of the level of self-reflection going on in the average American.

Because our sample size isn’t that big, we have to be careful about how we subdivide this data, so breaking it up into three age categories is about as good as we are going to do. But you can clearly see that there’s a distinct age effect on the question when it comes to people reflecting on their moral character.

Among the youngest adults, a significant portion (38%) say that they are “very far” or “somewhat far” from being the person that they ought to be from a moral standpoint. In contrast, almost none of them feel like they are “already there.” But as you move down the graph, the bars begin to shift in the opposite direction. Among the middle age bracket, only 28% place themselves in the bottom two categories.

But look at those who are 55 or older! Just 6% say that they are “very far” from the person that they want to be and 8% said “somewhat far.” In contrast, about half of them said that they were “pretty close” or “already there” when it came to the process of moral refinement. Older folks seem pretty content with the person that they’ve become.

However, there were other questions in the survey that focused on this process of self-improvement.

How often do you make a deliberate effort to improve your character?

How often do you reflect on whether your actions were right or wrong?

The graph on the left is the share of each age bracket who said that they made a deliberate effort to improve their character on a daily basis. Again, there’s a clear upward trend. Older folks are about 11 percentage points more likely than younger adults to say that they are trying every single day to improve their character. Think about that in conjunction with the prior graph where older folks were likely to say that they were closer to the person that they were supposed to be.

Here’s a possible explanation: people who’ve spent decades consciously working on their character feel closer to the person they want to be. Trying harder and feeling more arrived can genuinely go together. In other words, all that hard work is actually translating to a perception that they are making moral progress.

But what’s also interesting is that older folks are clearly less likely to ruminate on their actions frequently. Among those who were 55 or older, only ⅓ of them said that they were reflecting on their behavior most of the time or always. So it seems like something important happens as we age: we try hard to be better people, and lots of us feel like we are getting closer to that goal. However, older folks just let things go more easily. They don’t dwell on every single thing they said or did during the course of the day.

To this point, though, we haven’t mentioned religion. There does seem to be ample reason to believe that religiously engaged people would reflect on their actions, right? I don’t think it takes a huge leap in logic to assume that people who sit in a house of worship on a regular basis will have more opportunities to think through their behaviors during the previous week.

Because the sample size wasn’t that big, we could only divide our sample into three groups: those who are practicing religious folks (attendance at least monthly), non-practicing religious (those attending yearly or less), and then the non-religious.

And you can see an unmistakable pattern in each of these three graphs – active religious folks are much more likely to think about the morality of their actions than those who either are non-practicing or non-religious. The gaps among the youngest adults are pretty large. Among the non-religious, 27% say that they think about the morality of their actions “most of the time” or “always.” It was 53% of those who were religiously engaged. It’s pretty hard to miss the conclusion that an hour in a sanctuary that includes some preaching does lend itself to moral reflection.

What we do want to highlight, however, is that self-reflection does tend to wane among older religious respondents. For instance, among the practicing religious people who are under 35, 53% were thinking about their actions constantly. It was 44% among those who were 55 or older. There was also an 8-point drop among the non-practicing religious respondents, too.

Let’s do the same basic analysis but now make the dependent variable that first question we analyzed: How close do you feel you are to being the kind of person you ought to be, morally speaking? This is the share who said that they were “pretty close” or “already there.”

Now this is fascinating based on the previous bit of analysis! Among the youngest adult respondents, it was the most religious who were the most inclined to say that they were closer to the person that they ought to be. And the gap is pretty large – 19% among the non-religious and non-practicing for 34% among the young adults who attended church at least once a month. The one big change in the middle-aged bucket is that the non-religious are much more likely to say that they are closer to the person that they ought to be than their younger counterparts (19% → 30%).

However, the data for the oldest adults is just an entirely different story. The percentages are much higher across the board, regardless of religiosity. But the bigger shift is that now it’s the non-religious who are more likely to feel like they are the person that they are supposed to be. Among older nones, nearly 6 in 10 say that they are “pretty close” or “already there.” It’s 47% among the other two groups.

To finish up this post, we thought it would be fun to combine two metrics into a single bit of analysis: moral confidence minus actual reflection. A negative number means people are doing more soul-searching than their confidence would suggest; a positive number means their confidence has outrun the habit of checking their own behavior.

This is where it really becomes apparent who is engaging in a whole lot of soul-searching while not feeling super comfortable with their moral position. Among young adults, the bars are very much in the negative territory for the religious respondents. They are far away from the person that they want to be and they are reflecting on their actions frequently. Among young nones, those two metrics are pretty equally balanced.

For middle-aged respondents, the metrics for the religious are basically in the same spot. There’s a lot of moral dissatisfaction happening there. They are constantly monitoring their behavior and not really feeling like they are landing in a good spot. Meanwhile, the number for middle-aged nones moves in a much more positive direction. They are growing content with their moral character and not scrutinizing every action.

The results for the oldest people in our sample indicate that everyone seems to be growing more content with who they are, but it’s not the same level of contentment across all three groups. Among the religiously active, the number is still solidly negative. However, among the oldest nones, it’s easily the most positive. No one seems to be closer to feeling that they have “arrived” in a moral sense than a non-religious retiree.

Now, of course, the issue looming over all of this is that everything is self-reported and subjective. The first question, “How close do you feel you are to being the kind of person you ought to be, morally speaking?” does not presuppose that there’s some type of ideal moral plane that we can achieve through self-improvement. Everyone’s ideal for morality is probably just a bit different. And that may be why older nones are in a different spot on these metrics compared to religious folks.

If you’ve ever been a part of a church community, you have probably been admonished hundreds of times to become more Christ-like. That same theme runs through basically every religious tradition. Does that create a different (possibly higher standard) to a question about becoming the person we are supposed to be? It’s hard to know.

What is self-evident from this data, however, is that self-improvement is a young person’s game. Older folks just seem happy to be where they are in life. They probably aren’t going to get obsessed with jaw-maxxing or sleep-maxxing. They just want to enjoy however much life that they have left.

Code for this post can be found here.

Ryan P. Burge is a professor of practice at the Danforth Center on Religion and Politics at Washington University.

Let me start by stating a fact that seems incomprehensible to even type out, but it undergirds almost every single graph that I produce: we don’t actually know the religious composition of the United States with any real precision. I know that sounds bizarre. But the honest truth is that we are relying on a whole bunch of “best guesses” based on sound methodology and replication — and we have no way to know for certain if any of them are right.

The reason is simple: the United States made it illegal to ask about religious affiliation on the Census back in the 1970s. (This short piece from Pew explains the whole mess well.) That puts researchers like me in a real bind, because without a Census benchmark there’s no authoritative reference point to weight our surveys against. We know exactly what share of Americans have a four-year college degree because the Census tells us. We have no equivalent anchor for Catholics, or evangelicals, or the religiously unaffiliated.

To their great credit — and great expense — Pew has tried to fill this void with the National Public Opinion Reference Survey (NPORS), and it’s a genuine methodological achievement. But everyone working in this space knows that what we really need is a Census benchmark, and we don’t have one.

In a post from February, I pulled back the curtain on survey methodology — specifically the difference between probability-based polls and opt-in panels. I’m not going to rehash all of that here, so I’d strongly encourage you to go read it first. But in short: probability-based surveys like the Pew Religious Landscape Study randomly select respondents from the full population, while opt-in panels like the Cooperative Election Study draw from volunteers — and that distinction can produce meaningfully different pictures of American religion.

How meaningfully different? That’s exactly what today’s post is about. Let’s start by comparing the two surveys on the basics: religious attendance and religious affiliation across the full sample.

It doesn’t take long to find a meaningful difference between the two instruments. In the CES, 34% of respondents say they never attend religious services, compared to just 25% in the Pew data — a nine-point gap that would be genuinely alarming if it held up across the board. But zoom out a little and the picture gets more complicated. Pew actually records a higher share of “seldom” attenders than the CES — 25% versus 21%. Add those two bottom categories together and the gap shrinks considerably: 55% of CES respondents attend less than once a year compared to 50% in Pew. That’s a five-point difference, which is not nothing, but it’s hardly a chasm.

What I find most reassuring is the right side of the graph. The share of CES respondents who attend weekly or more? Exactly 25% — the same figure Pew produces. On the measure that matters most for congregational health and religious vitality, the two surveys are in complete agreement.

The non-religion picture tells a similar story. In the full CES sample, 34.5% identify as religiously unaffiliated compared to 29.4% in the Pew data — that same five-point gap showing up again. But the more interesting question isn’t the overall margin, it’s where that gap comes from. Here’s what it looks like when I break it down by birth decade.

It’s certainly the case that older Americans are much more likely to be recorded as non-religious in the CES compared to the Pew RLS. Among those born in the 1950s, the figures are 25% versus 17%. The gap is eight points among those born in the 1960s, and seven points for those born in the 1970s — 36% versus 29%. These are real, substantive differences that are impossible to simply wave away.

When we dig into the type of none, though, a clearer picture emerges. The primary driver of the divergence is the “nothing in particular” category — older CES respondents are considerably more likely to land there compared to the Pew sample. The atheist and agnostic gaps exist too, but they’re modest by comparison.

I’ve turned this over in my head many times and I can’t fully resolve it. My first instinct was social desirability bias — the idea that Pew respondents were more likely to be talking to a live interviewer and therefore less willing to disavow religion. But Pew’s own methodology indicates that 97% of RLS participants completed the survey through a web browser, just like CES respondents. That levels the playing field and takes my best explanation off the table.

The most I can say with confidence is that address-based sampling and opt-in panels are recruiting fundamentally different kinds of older Americans, and those differences show up in how they answer questions about religious identity. But I’d be overstating my certainty if I claimed to know exactly why. Sometimes the honest answer is that the data surfaces a question I can’t fully resolve.

What makes this even more puzzling is what happens at the other end of the age spectrum. Among the youngest Americans — those born in 1990 or later — the two surveys are essentially in agreement, never diverging by more than a couple of points. If the gap were simply a product of opt-in panels attracting a certain kind of respondent, you’d expect it to show up across all age groups. The fact that it’s concentrated among older Americans suggests something specific is happening with that cohort, not a universal feature of how these two instruments measure religion.

With that caveat on the table, let’s press on. I pulled together Pew’s estimates for the share of various religious groups with a four-year college degree and compared them to the same figures from the CES.

Read more

One of our all time favorite subjects for reality television was a huge family who lived in Northwest Arkansas named the Duggars. Their first iteration of their show was 14 Kids and Counting. A later iteration? 19 Kids and Counting. Of those 19 children, 13 are currently married and there are at least 40 grandchildren so far.

It’s a family that is not without controversy. Their oldest son, Joshua, had an account on the infamous cheating website Ashley Madison. He was subsequently convicted of possessing child pornography and is currently serving a 12.5 year prison sentence for the crime. Another Duggar child, Joseph, has also been arrested for molestation of a child under the age of 12. One of the daughters has accused their father of financial exploitation because he demanded that his adult children agree to be filmed for the television series without compensation. There’s a lot of drama there.

But what’s always in the background of any discussion about the Duggars is their religious beliefs. They were adherents to the Institute for Basic Life Principles – an extremely conservative evangelical organization. The IBLP, founded by Bill Gothard in the 1960s, teaches a strict hierarchy of authority in the home: children submit to parents, wives submit to husbands, and everyone submits to God. Gothard’s organization discouraged dating in favor of parent-guided courtship, promoted homeschooling through its own curriculum, and placed enormous emphasis on female modesty. (Gothard himself resigned from the ministry in 2014 after dozens of women accused him of sexual harassment.)

But the teaching that made the Duggars famous was the family’s embrace of what is often called the “Quiverfull“ ethos, drawn from Psalm 127 - children are arrows in the hand of a warrior, and “blessed is the man whose quiver is full of them.” No birth control of any kind. Every pregnancy is a blessing from God, and family size is left entirely in His hands. That’s how you get to nineteen kids.

And that’s the image that pops into most American minds when they see a family with a whole bunch of children at the grocery store or piling out of a fifteen-passenger van in the church parking lot. They assume that mom and dad are deeply devout people who are having all those kids for Jesus. It’s a tidy little stereotype. The only problem is that the data doesn’t really back it up.

Let’s start with this question, though: how many American families are large? I just looked at folks in the Cooperative Election Study who were at least 45 years old and had at least one child.

The only child family is not that widespread – less than a quarter of parents said that they had just one kid. The modal outcome, unsurprisingly, is having two children. Forty percent of households have a pair of kids and another 22% have had three children total. That means that 85% of all families have between one and three kids.

It’s pretty darn rare to have a family with at least five kids – that encompasses just 6% of all those households. Out of a total sample of 60,000, only 23,700 had ever had a child. Of that, just 113 had eight or more. That’s less than .5% of all parents. Even a family that is half the size of the Duggars is a statistical oddity.

And, of course, that is becoming even more of an outlier with each passing generation.

I did the same analysis across the three oldest generations in the data – Silent, Boomers, and Gen X – to see how much family sizes have changed over the last fifty years, and I was honestly a bit surprised at how consistent the results were across generations. For instance, having two kids was the most likely outcome for parents who were born in the 1930s, the 1950s, and the 1970s. In each case, that was about 40% of families. Large families are about as commonplace, too. Among Silent Generation folks, 20% had families of at least four kids; it was 15% of Boomers and Gen X.

Here’s the only difference that seems substantively significant to me: having just one child. Among the Silent Generation (born 1925–1945), only 15% of parents had a single child. It was 22% of Boomers and 27% of Generation X. And, although not all Millennials have gotten to the end of their childbearing years, when I just look at those born in the 1980s, 31% have only had one kid and 38% have had two kids. So the family size trend will just continue on.

Let’s dig into two possible explanations for this decline – politics and religion. I am going to start by just looking at the partisan distribution across family size. One caveat here: I’m only looking at people who have at least one child and we know that those on the left side of the political spectrum are less likely to have children.

Yeah, I’m not blown away by this result. I think it’s fair to say that Democratic parents are slightly more likely to only have one child compared to their Republican counterparts (26% vs. 20%). But I wouldn’t call that a huge difference. When it comes to having two children, the partisan gap is completely nonexistent (40% of both groups have two kids). There is some evidence here that Republicans tend to have slightly larger families but, again, the differences are not huge.

Among Republicans, 39% have a family with at least three children – which is also the same percentage for political Independents, by the way. For Democrats, it’s 34%. So, yeah, there’s a substantive difference there, but it’s not like Republicans are just outbreeding Democrats on a massive scale. The average GOP family is 2.46 kids compared to 2.29 for Democrats.

What about religion? Let me show you attendance two different ways. First, this is the family size breakdown by level of religious attendance.

For families who never attend religious services, 27% have had only one child, while 13% have at least four children. Among those who attend religious services more than once a week, 20% have just one child and 20% have at least four. That’s not what I would describe as a substantively large difference. The idea that those super-religious families have a whole brood of kids is just not even close to what the data shows. In fact, less than 10% of them have five or more kids. They are twice as likely to have one kid compared to five-plus.

The changes are also pretty subtle once you get to the “yearly” level of attendance. 60% of families at this attendance band are one to two children and 16% are four or more. Those numbers are within five percentage points of those most devout. In other words, the perception of non-religious families having one kid while the local Baptist church parking lot is full of Sprinter vans with five car seats is not borne out here.

Here’s that same data another way – how is the religious attendance of parents of four-plus children compared to families of any size?

I think it’s fair to say that large families are more likely to go to church, but not by that much. Among families who have had four or more children, 33% said that they go to church at least once a week. For families of any size, the weekly attendance rate is 29% – four points lower. A slim majority of all families report that they go to church less than once a year. Among large families, the seldom/never attendance rate is 46%.

You see what I’m saying now? The religious differences between big families and all families are pretty miniscule in the grand scheme of things. All things considered, I think that the average family of four is probably three to five percentage points more likely to be attending less than once a year compared to a family of six. Embracing an approach to religious renewal by encouraging fertility doesn’t seem like a strategy that will move the needle in a significant way.

What about comparing the religious affiliation of larger families versus families of any size? That’s what I did below.

In almost every dumbbell plot, the two point estimates are almost exactly on top of each other. For instance, Catholics made up 20% of the largest families. They were also 20% of families of any size. That was also true for people who claimed no religion in particular (~14% for both family sizes). In fact, the average gap was less than two percentage points across every religious tradition.

There were a couple of outliers worth flagging. Among families of all sizes, evangelicals were 29.5% of the sample. Among the largest families, they made up 32% of that subgroup. Evangelicals are overrepresented among families like the Duggars, but only by a handful of percentage points. For Latter-day Saints, those two figures were 1.3% vs. 3.6%. So, yes, there is empirical support for the idea that LDS do have larger families, but in pure numeric terms, only about 1 in 30 families with at least four kids are members of the Church of Jesus Christ of Latter-day Saints.

Outliers have an unbelievable hold on our imaginations. We see a family like the Duggars and think that they represent essential features of that entire phenomenon. When in reality, they are outliers even among outliers. Yes, religion does seem to have an impact on the decision about the number of kids a family will have, but it’s a very minor difference.

Comparing evangelicals to atheists and agnostics – the widest religious contrast the data offers – and controlling for age, race, education, and sex, the gap is about 0.6 children. Real, robust, and the single biggest demographic divide in American family size. It’s also just over half a kid. So, yeah, religion matters here, but it’s not all that counts.

Code for this post can be found here.

Ryan P. Burge is a professor of practice at the Danforth Center on Religion and Politics at Washington University.

This post has been unlocked through a generous grant from the Lilly Endowment for the Association of Religion Data Archives (ARDA). The graphs you see here use data that is publicly available for download and analysis through link(s) provided in the text below.

One of the most common critiques I hear about survey research on religion is that it flattens something that is deeply personal and profoundly complex. Questions about belief certainly fall into this trap. The General Social Survey has questions about a respondent’s view of God and the Bible that only allow a handful of response options. A common response is, “I don’t really fit cleanly into any of those boxes.” And honestly? That criticism is right on target. Asking someone “do you believe in God?” and forcing them into just a couple of potential boxes misses an enormous amount of nuance. But that’s the nature of survey research — we trade depth for breadth.

That’s why I’m always happy to see a survey that goes one step further on a belief question. The Pew Religious Landscape Survey does just that and it’s easy to see how it’s broken down on the ARDA website. It first asks, “Do you believe in God or a universal spirit?” But then the very next question is, “How certain are you about this belief?” But here’s what I really appreciate about this setup — everyone gets the second question, no matter how they answered the first one.

Which means not only can we assess the certainty of believers, we can also get a good sense of how certain non-believers are in their position as well. That distinction matters — a lot. A faith held with ironclad certainty looks very different from one held with a kind of hopeful tentativeness. And as you’re about to see, where Americans land on that spectrum varies dramatically by generation, by ideology, and by how often they actually show up to worship.

First, I’ve got to just start by laying out how the sample falls across those eight possible response options.

A majority of the sample falls into the “I believe in God with absolute certainty” category — that’s 55% of all respondents. Clearly, over half of Americans have a deep well of religious belief. Another 22% are in the “I am fairly certain about my belief in God” camp. That means that nearly four in five respondents score quite high on this belief metric.

The share of Americans who are absolutely certain that God does not exist is a bit less than 6%, and then another 7% say that they are fairly certain that God (or a universal spirit) does not exist. Said another way, for every person who is pretty sure that God doesn’t exist, there are six who are fairly certain of the existence of a Higher Power. Non-believers are statistically rare.

I did want to explore how much that had shifted from one generation to the next, though. So this is the same analysis but I just divided the sample into the birth decade of the respondent.

It should come as little surprise that older Americans are more likely to espouse a certain belief in God. Among those born before 1970, almost two-thirds can be found in the top left of each heatmap, and 85% are in the two squares in the top left. But that begins to erode when we move down to younger birth cohorts. The real drop off occurs among people born in the 1980s, when less than half believe in God with absolute certainty. And it declines even further from there — 42% among those born in the 1990s and 39% for folks who were born in the 2000s.

This raises the question: if certain belief drops by ~25 points, where do all those folks end up? The answer is clearly not in the “I certainly don’t believe in God” option. Even among the youngest respondents, only 6% are found in the bottom left square and a total of 16% are fairly certain or completely certain that God doesn’t exist. There’s a noticeable shift in the share of younger cohorts who fall into the “I’m fairly certain God exists” box. Which means that they aren’t sliding all the way to the other side of the spectrum. Instead, they are just moving one level down.

One thing that could be driving the differences in the above heatmaps is politics — younger folks are more likely to be ideologically liberal and older folks tend to lean conservative. Could that possibly have an impact on these results?

One rock solid finding from my analysis is that ideologically conservative folks are more likely to hold a certain belief in God compared to moderates and liberals. Among older conservatives, about 80% hold to a certain belief. It does erode a bit among younger conservatives, but even among the youngest adults, it’s still two-thirds of them who say, “I believe in God and I’m absolutely certain of that.”

Moderates are consistently 20 points lower than that. Among the oldest moderates, 60% are in the certain belief camp and it dips down to 40% among the youngest moderates. It’s interesting to me how the 1980s birth decade stands out as the pivotal one. For conservatives, certain belief drops by ten points while it’s 8 points among moderates.

Liberals are much lower than conservatives. The gap is consistently 35-40 percentage points when comparing one cohort to another. Among the youngest ideological liberals, only ~20% of them say that God exists and that they are certain of this fact. That’s 45 points lower than conservatives from the same age group.

But one thing to also note here – the drop in certain belief is the same size for both liberals and conservatives. It’s 15 percentage points when comparing the oldest respondents to the youngest. So, certain beliefs are eroding in a pretty consistent way from one generation to the next. It’s just that the baseline is so much lower for liberals.

I wanted to go one layer deeper on this and really try to figure out where liberals ended up on the belief continuum. I visualized that below based on the birth decade of the respondents.

What is striking to me is that while there are a significant number of liberals on the right side of this heatmap, it’s nowhere close to a majority of them. For instance, among liberals who were born in the 1980s, 30% of them said that they were fairly certain or absolutely certain that God doesn’t exist. In contrast, 29% were certain of the existence of God. So, it’s just not empirically true that younger liberals are atheists — the data clearly says that’s not true for 70% of them.

Instead, a whole bunch of liberals can be found in the “I’m fairly certain God exists” camp. What’s interesting to me is that among those born in the 1990s, the share who are in the fairly certain camp is about the same as the absolutely certain camp, and it even reverses among the youngest liberals with 27% being fairly certain of God’s existence and 21% being completely certain.

To this point, I’ve only focused on looking at the belief question by ideology and age of respondent. There are two other dimensions of religiosity that should be explored: belonging and behavior. The most conventional measure of behavior is religious attendance. I wanted to focus specifically on folks who attend church less than once a year.

What I really wanted to sort out for us is this question: is the decline of religious behavior related to a decline in religious belief? And I think the answer is a qualified “yes.” For older Americans who don’t go to church that much, belief is still really robust. In many older cohorts, 70% of low attendees are in the top two categories of belief in God. Meanwhile, only ~20% are in the bottom two buckets. So, older Americans may not show up to church often, but there’s still a robust belief system there.

For younger respondents who seldom or never attend religious services, the level of belief is undoubtedly much lower. Among those born in 1990 or later who go to church less than once a year, 47-48% say that they are absolutely or fairly certain of God’s existence. That’s at least 20 points lower than their grandparents’ generation. However, look at the columns on the far right — even among young adults who rarely attend religious services, less than 30% of them say that they are absolutely or fairly certain that God does not exist.

If we were looking for a subgroup where atheist views would be widespread, it’s the bottom right corner of this heatmap, and even there, it’s still less than one in three.

Speaking of atheists, I wanted to just visualize how the three types of non-religious Americans (atheists, agnostics, and nothing in particular) fall on this belief continuum.

Among people who say that their present religion is atheist, 53% indicate that they are absolutely certain that God doesn’t exist and another 31% say that they are fairly certain that there is no universal spirit. Of course, very few atheists land on the left side of this graph. But that’s not the case for agnostics. In fact, 45% of them fall above the midpoint on this analysis by saying “yes” to the first question about the existence of God or a universal spirit. For comparison, only 40% fall into the bottom two boxes.

It couldn’t be more apparent, however, that “nothing in particular” folks have a much more robust belief in God than their atheist or agnostic counterparts. Among these nothing in particulars, 54% say that they are fairly certain or absolutely certain of God’s existence. Just 8% said that they were sure God didn’t exist and 13% were fairly certain of God’s non-existence.

I think that the overarching conclusion that I can derive from this data is that it’s pretty rare to find an individual who says that they have a great deal of confidence in their belief that God doesn’t exist. In the full sample, that’s just 13% of folks. Even among the youngest adults who never or seldom attend religious services, it’s less than 30% of them. The only group where this belief reaches majority status is among atheists. For agnostics, it stands at 40%.

Americans may be 25 points less likely to identify with a faith tradition today compared to 30 years ago, and over 40% say that they never attend religious services. But almost 80% of Americans agree that God or a universal spirit exists.

The United States may no longer be a country of people widely engaging in religious behavior or religious belonging. But it certainly has a whole bunch of religious believers.

Code for this post can be found here.

Ryan P. Burge is a professor of practice at the Danforth Center on Religion and Politics at Washington University.

There are a couple of books in the sociology of religion that always stick in my head, too. American Evangelicalism: Embattled and Thriving is one I will never forget. Another was released a few years ago that keeps coming back to mind — Religion Is Raced: Understanding American Religion in the Twenty-First Century. It’s an edited volume, and I bet you can guess the subject matter just from the title.

Religion in the United States “works” differently for Asian Americans compared to African Americans. The average white American doesn’t think about religion the same way, or practice their faith in the same patterns, as Hispanic Americans. Race is a central component of how religion operates in the American context. But despite those varied expressions, I wanted to try to answer, with as much data as I can muster, a deceptively simple question: which racial group is the most religious?

To do that, I’m going to use the Pew Religious Landscape Study, because it has a huge sample size (~37,000) and asks a ton of questions about the various manifestations of religion, which are primarily along the dimensions of behavior, belief, and belonging. For more on this, I will always link to this post: What Does it Mean to Be Religious?

Let’s start with the religious composition of each racial group.

Huge differences emerge right away. Which group is the most likely to be Protestant? That’s Black respondents, by nearly twenty-five percentage points. The most likely to be Catholic? Easily Hispanics at 42% — again, about twenty-five points higher than any other group.

Asian Americans are far and away the most likely to identify with another faith tradition, a category that includes Jews, Hindus, Muslims, Buddhists, and the like. In the rest of the population, those groups rarely combine for more than 10% of the sample. What about the nones? The “All Others” category leads the pack at 37%, followed closely by Asian Americans at 34%. I also need to point out that Asian Americans are the only racial group that is not majority Christian, too.

What about religious importance? This is where a clear pattern starts to emerge.

Black respondents are, by a long shot, the most likely to say that religion is very important to them — 58%, compared to 30-40% for every other racial group. In fact, only 18% of Black respondents say religion is not too important or not at all important to them. That’s ten points lower than any other group.

By contrast, 22% of white respondents say religion is not at all important, the same share as the “All Others” category. Among Hispanic respondents it’s only 13%, and among Asian respondents it’s 18%. Looking at this graph in totality, it’s fair to say that African Americans are the most religious, Hispanic Americans are next in line, and then it’s basically a tie between white Americans, Asian Americans, and everyone else.

When I analyzed a question about belief in God, the same pattern came up again.

Black respondents are undoubtedly the most likely to believe in God without a doubt — 75% of them. The share of African Americans who say they don’t believe in God at all is tiny, about five percent. You get a clear sense from this graph that African Americans, on belief metrics, are just in a different world than every other racial group in the United States.

Then the next question becomes: who is the least likely to believe? The answer is Asian Americans. Only 37% believe in God with absolute certainty — fifteen points lower than white Americans, who come in second-to-last on this measure. At the same time, 25% of Asian respondents say they don’t believe in God at all, easily the highest of any group.

On religious importance, African Americans are clearly in the lead. On belief in God, it’s African Americans again — but for the other groups, it’s hard to settle on a rank ordering. Maybe questions about religious behavior can break the tie.

When it comes to religious service attendance, Black Americans are the most likely to report weekly attendance at 30% — six or seven points higher than white, Hispanic, and Asian respondents. In total, 61% of African Americans say they attend church at least yearly. That’s about five points higher than Hispanic respondents, who come in second.

At the bottom end of the scale, 27% of white respondents say they never attend. It’s 29% among “All Others,” and 22% for both Hispanic and Asian respondents. Those numbers are close enough that it’s hard to say definitively which group is the least likely to darken a church door.

Prayer frequency follows the same script. African Americans are clearly in the lead — nearly two-thirds of them pray every single day. Among Hispanic respondents, 48% pray daily and another 17% pray at least weekly, which means about two-thirds are praying on a weekly basis. That’s roughly ten points higher than white respondents and eighteen points higher than Asian respondents.

So on religious behavior, the picture is: African Americans, and then everyone else in a muddled pack.

But what happens when I combine all four of these items into a single composite measure? I constructed the scale to run from 0 to 10, with 0 indicating the least religious and 10 the most religious. This additive index brings some real clarity to the question.

For anyone paying attention to the prior analysis, it should be no surprise that African Americans score the highest — 7.2 out of 10. The share of African Americans scoring at least a 9 on this index was 28%, which is eight points higher than white Americans and ten points higher than Hispanic respondents. No matter how I measure it, African Americans lead on the question of which racial group is the most religious.

The comparison between the white histogram and the Hispanic histogram is the most statistically interesting one in this graph. Hispanics post the higher mean on this index — 6.1 versus 5.5 for white respondents. Yet look at the far right edge of each panel: there are clearly more white respondents maxing out the scale than Hispanic ones.

So why is the Hispanic mean about half a point higher? It’s the middle of the distribution. Nearly 71% of Hispanics score at least a 5 on this index, compared to just 59% of white respondents. Meanwhile, 26% of white respondents land below a 3, compared to only 16% of Hispanics. Said another way: white America contains a smaller group that is incredibly religious and a much larger bloc that isn’t religious at all. Hispanic religiosity doesn’t have that hollowed-out bottom — the typical Hispanic respondent sits comfortably on the religious side of the scale.

And when it comes to the question of which racial group is the least religious, I think this graph makes it clear: Asian Americans, with a mean of just 5.1. Only 52% of Asian respondents score at least a 5 out of 10, which is seven points lower than any other group. And here’s a startling data nugget: for every Black American who scores near the bottom of this index, there are more than four who score a nine or higher. Among white Americans, the bottom group is actually larger than the top one

But what happens when I throw age into the mix? Maybe the religiosity gap is closing among young Americans?

One macro-level trend that’s hard to deny: each successive birth cohort is less religious than the one before it, and that’s true across all racial groups. Among the oldest African Americans, the mean score is about 8 out of 10. Among the youngest, it’s only 6.

However, the gap between African Americans and everyone else persists across nearly every birth cohort in this data. There’s one aberration among people born in the 1970s, where Hispanic and Black respondents post essentially the same level of religiosity. In every other cohort, Black respondents score noticeably higher than anyone else.

I also have to point out the white respondents, especially those born in 1980 or later. Among these younger white adults, the average score is 4 to 4.5 on a ten-point scale — about a point lower than Hispanics from the same birth cohorts and roughly two points lower than African Americans born in the same decades.

But the overall rank ordering holds. African Americans are easily the most religious, followed by Hispanic Americans. In the full sample, white Americans were about half a point more religious than Asian Americans, but I’m not sure that’s going to hold up in future generations.

So there you have it. African Americans are significantly more religious than anyone else, whether you measure it through behavior, belief, or belonging. It holds up among older respondents and younger ones, too. That’s the clear thread running through all this data.

But it’s also hard to deny this: a young African American — while notably more religious than their peers — is still significantly less religious than the average Baby Boomer.

Secularism is coming for us all, and I wonder if there will be a new edited volume someday: Non-Religion Is Raced.

Code for this post can be found here.

Ryan P. Burge is a professor of practice at the Danforth Center on Religion and Politics at Washington University.