Asterisk: You've done a lot of work on measuring religiosity in America. How do you break that down into quantities that you can make a data series about?

Lyman Stone: Religiosity is different from religious affiliation. Ask someone, “What’s your religion?” and you’ll get responses like Sikhism or Christianity or Lutheran Church–Missouri Synod. That’s your religion. 

Religiosity is essentially the answer to the question, “OK, but how much?” It gets at the intensity and specific mode of expression of someone’s beliefs, or the ways that religion matters to that person. And so to measure religiosity in a given society, you really need to know what the underlying religion is and what kind of practices might stem from it. So if we’re trying to understand religiosity among Christians, we might ask things like: “How often do you attend church services? How often do you pray? How often do you read the Bible?” For Jews, we might ask about abstinence from pork. For Hindus, we might ask about abstinence from alcohol, beef, or meat in general. For Sikhs, we might ask about shaving practices, the carrying of weapons, and things like that.

A: In your report on religiosity in America, “Promise and Peril,” you do a lot of work to break down religiosity and religious affiliation. I think of this as distinguishing between “Do you identify as a member of a group?” versus “Are you a member of a church?” versus “How often do you attend church?” 

L: Right. And what makes this hard is we really don’t have any measures of religion historically. In the past, most of the measures we can observe refer to the actual religiosity of a culture. We're looking at how people express their faith in terms of the naming of children or participating in rites and rituals that leave a historic record of some kind. For instance, if we go and dig up an archeological site, we can’t observe religion directly, but we can observe if there's a bunch of bones of ritual sacrifices or if there’s an altar that has an image of a god. We can observe religiosity, which can give us hints about religion.

At the same time, the separation of these things is kind of new. Religious diversity existed in the past, but it was more like a patchwork quilt of different communities. The idea of a community with different faiths living side by side, that everybody just goes to a different place on the weekend, is fairly new. If you consider the Ottoman Empire it might seem like they have a lot of religious diversity. There’s Christians, Sunnis, Shias, Druze, Alawis, Yazidis — all these different groups. But they’re not just all living side by side and going to church. They each have their towns. So even where you had diverse areas, you pretty much knew someone's affiliation if you just knew their address.

A: So, when you’re trying to measure something like trends in religiosity in America in the past, how do you do it? What kinds of things are you looking at? Obviously, you can't go back and interview people.

L: So there’s a large corpus of textual documents that qualitative historians have typically looked through to say, “This is how people thought about their religion.” I use those documents, but I mine a lot of that data quantitatively to look at various trends. I show that there have been changes in the frequency of certain kinds of religious words, for example, “trinity,” “sin,” and “sabbath.” Generally, over the course of the 19th century, by this metric American texts became less religious. And that’s true between genres as well. Even if you just look at fictional books, they got less religious. 

Another thing you can do is look at how people name their children. There are some names that are in the Bible or belong to Catholic saints and then there are other names that aren’t affiliated with those. This is especially evident if we look at rare religious names. A name like Hannah is not super interesting as a religious name because a lot of people name their kid Hannah. But no one names their child Hephzibah unless they mean something by it. 

Another thing you can do is look at membership in church bodies. Religious bodies have always had an interest in counting their people, so there are often historic, contemporaneous estimates. 

A: You would think church membership would be an extremely simple indicator, and you could ask questions like, “How many people are members of a church?” or, “What church are they a member of?” But, as you show, it ends up being very complicated and contested. 

I’d love to talk about some of the debates between the foundational sociologists of religion or religious historians trying to figure out how many Americans were members of the church in, for instance, the 1770s and 1780s.

L: If you take the traditional religious historiography view of 1780, America was a very religious place. But a wave of scholarship in the 1970s, ’80s, and ’90s, particularly that associated with Roger Finke and Rodney Stark, expands on it in an interesting way. 

Finke and Stark went through and cobbled together estimates of membership in religious denominations from two sources. One source was the denominations themselves. They went to denominations and said, “How many members do you have?” Those records have been collected for a very long time. Finke and Stark summed them all up.

Then the other source they used is the U.S. census of religious bodies, which was conducted from 1850 until 1936. This was an institutional, not individual, census — they asked every church in America to answer a questionnaire. Ultimately, this practice was discontinued because of the politics of religion during the period around the 1930s and 1940s. As you’re aware, there were some things happening in Europe involving governments asking about peoples’ religions that made the U.S. think, “Maybe this is a bad idea.”
So we have this data. Some of those surveys ask how many people at this church worshiped on a Sunday, some of them ask about membership, and some of them just ask things like, “How many seats do you have in your church?” So the exact data vary, but you can use these things to cobble together an idea. 

Then we have these historic efforts to count churches in America. In the 1770s and 1780s, there aren’t that many cities in America. There also aren’t that many churches in America. So historians have gone through and identified almost every historic church in America and counted how many were there. 

Between all of this data the number that Finke and Stark use basically emerges from a kind of historic reconstruction of archival sources. They come up with a very low number — 17% of people in the U.S. are churched. It’s a very secular vision of the Revolutionary era in the U.S. 

There is a rival view, however, and there are various people who articulate this view, but the one I focus on is two authors, Patricia Bonomi and Peter Eisenstadt, and they take a very different approach. In this approach, they say, “OK, hold on, let’s say we accept this count of congregations. How many people are there per congregation?” And then they go through a small number of congregations in Virginia, which is a colony thought to be one of the more secular colonies. And they ask, “For these congregations that we have records on, how many families are actually associated with them?” And they find substantially more than Finke and Stark! They determine almost 60% of people are church-affiliated — which is still a lot lower than the traditional view that suggests it’s 100% a Christian society. They’re still pointing to a much more secular society than that, but it’s a much higher proportion than Finke and Stark.

Now, what’s going on here is kind of a dispute between technical affiliation and regular attendance. However, Finke and Stark are giving you something closer to how many people are probably attending on a Sunday whereas Bonomi and Eisenstadt are showing you the total number of warm bodies ever associated with this church. 

So I sat down and said, “OK, how can we deal with this?” I wanted to fix coverage errors and come up with a more complete estimate by updating them to account for more recent information and to more rigorously cover gaps. What I find in using this method is that I can stretch all the way back to the late 1600s. 

And using this method, I find very high rates of religious affiliation in the late 1600s and early 1700s. Then it declines over the 1700s until the Revolutionary generation, where I find a number between Finke and Stark and Bonomi and Eisenstadt. And then I find a rise in the modern period, which is what Finke and Stark show as well. 

A: So there are these different measures that seem to cut against each other. A higher percentage of Americans attend church today than they did in 1780 — depending on what estimate you use, somewhere around 40%, and possibly higher, compared to 20%. But of course, a much higher percentage of Americans in 1780 would have thought of themselves as Christians.

L: Yeah. So we've got a situation where today’s attendance is higher than in the past, but affiliation is lower. Nominal affiliation is a lot lower – around 75% now compared to over 90% in the past. You could argue about why that is. You can talk about changing social costs and benefits of nominal affiliation. We could get more into this, but as a measurement problem, it is really interesting. Measuring is difficult even when we know church membership rates or baby names or baptismal certificates. We can also look at things like crime rates and things of that nature if you want to use that as a proxy for clean living. We know that the 1780s were genuinely a time of inordinate questioning of traditional religion. You know, at this time, this same period sees the disestablishment of a lot of state churches. Then also, 20, 30, 40 years later, when you get the Second Great Awakening, those churches grow like wildfire. Well, where are they growing from if not this large pool of people who are not connected to anything?

At the same time, we don’t have any contemporary accounts of a person from that time thinking that they were living in a society where only about 20% of people were Christian. Nobody thought this was a society where people were all atheists. Certainly, the normative public speech pattern, and the way they talked about themselves, was in a Christian theological and moral way. This is why even in arguably the most secular political document created to that point in history, the U.S. Constitution, the preamble is still couched in this kind of Christian monotheistic way of talking about the world. So yes, it’s this very weird thing measurement-wise.

A: Zooming out a little bit — you’ve got this pattern, these two peaks and troughs where religiosity is quite high in the 17th century, declines into the mid-18th century, rises again, peaks in the mid-20th century, and then starts to fall again. I’m curious about what phenomenon you’re measuring here. What pattern in American life do you think this trend is reflecting? 

L: The people who moved to the colonies weren’t random; disproportionately they were people who were not happy with the way things were in Britain. That tended to be religious people. 

Even though it was legally mandatory to attend Anglican church in 1600s England, probably only 10% to 20% of people did. In general, a lot of people, a lot of religious dissidents, didn’t like that. They felt that this system was both repressive and weirdly secular. Like, “How is it that on the one hand, we’re forced to go to church and also all these guys are just at the pub gambling all day Sunday? I want to go to my Baptist church and be devout.”

It wasn’t just the Puritans. The Americas are appealing for Quakers, Baptists, Puritans — all these groups. The people coming were people who, even if they didn't move for religion, had an opportunity to build a society in their own image and took it. They built religious societies. 

That didn’t last. Because over time more and more people came. Immigration became less selective. On descent from the norm in England, they got migrants from Germany and Ireland and these other places. The society got bigger and more complex and it became harder to affect social control through traditional means. What you see over the course of the 18th century is that the system of social control broke down. Then the First Great Awakening is an attempt by the established religions, by and large, to rebrand themselves and achieve social control. My data shows that it fails — and that they keep declining.

A: Until the Second Great Awakening.

L: The Second Great Awakening in the late 18th century and early 19th century, however, involves new denominations. Methodists, Baptists, Mormons, Adventists, Jehovah’s Witnesses — all these groups get launched, and they’re very successful. They succeed in a kind of religious entrepreneurship. This is where I agree with Finke and Stark when they advance what’s called the “supply-side model,” basically the notion that religious entrepreneurship creates a form of religion that appeals to what Americans want at that period. 

So the old system of social control breaks down. But as Americans settle in the country and find that they want a form of sort of voluntaristic and communal order, religion steps into that void. And so the rise of religiosity in America is hand in hand with the rise of prohibitionism. The same goes for the rise of moral activism and the rise of abolitionism, because settlers are confronting this society that is clearly — and you can see from the murder rates — constantly on the edge of chaos. And they want to find a source of order. These emergent religious movements are very credible sources of order.

A: I’d like to pivot to some of your other work as a demographer. One of the areas you’ve focused on recently is fertility preferences, and there you look at a few different metrics. Can you describe what they are?

L: There are a variety of metrics when it comes to fertility, but probably the two easiest ones to look at are how many kids you want to have, and then how many kids you do have. In almost every Western country, both of these are in pretty rapid decline. 

On average, when asked how many kids they want — fertility desires — people say they want two, maybe three. Some say one, and there are a small percentage of people who want more than three, but not that many. 

Now, regarding how many kids one wants, you might think that’s just a flight of fancy, but it’s generally predictive of behavior in longitudinal data. If I ask you when you're 15, “How many kids do you want to have?” The people who say “three” do in fact have more children than the people who say “one.” And importantly, when people end their life with more or fewer children than they desired, they tend to be less happy than people who hit their goals.

A: You also look at fertility intention as separate from both desires and outcomes. Can you explain the difference? 

L: Intentions are not the answer to the question, “How many children do you want to have?” but, “How many kids do you plan to have?” or, “How many kids do you intend to have?” Or even, “How many kids do you expect to have?” 

In the vast majority of cases, intentions are equal to or lower than desires. (There are a few people who intend to have kids they don’t want — the reason for that is often that their spouse or partner wants that kid, and they want their spouse or partner to be happy.)

But in general, intentions tend to be a compromise with reality — and reality tends to push that number lower than desires. Why? It’s rare that you discover that you're more fertile than you expected. Most people assume that they can have babies until they discover they can’t. 

Having a married partner makes you approximately four times as likely to have a child in a given year.  But if you find your partner later in life, as more people do now, you’re more likely to be exposed to age-related infertility. And actual fertility tends to fall below desires and intentions because of life course factors. By the time you get to the point where you’re to start having children, and you’re shaping your intentions as you go, you may realize you don’t have as much time to have children as you thought, or you realize that children are more expensive than you expected.  So with intentions, you tend to revise downward. 

Then you have what actually happens — fertility outcomes —  and what actually happens tends to be below intentions as well. Because even though you intend to get what you want, life happens: divorce, fertility issues, etc.

The terminology on this is sometimes frustrating. What do you say about someone who meets their reproductive desires? Were they successful? In some sense, yes. However, that sometimes makes it sound like you’re passing judgment. Like if they tried harder, they would have been able to achieve their intentions. 

A: You recently published a report on fertility desires in Canada, and I’d like to dive into some of that data. You surveyed women between 18 and 44 years old. Let’s say you ask a woman at the lower end of that range how many children she wants. Fast-forward to when she’s older and looking back. You can view her not having as many children as she said she wanted at an earlier point in her life as a failure, or as saying life intervened, or you can say that she changed her mind. But could you also look at it as a difference between stated and lived preferences? 

L: This gets really complicated. Our surveys were cross-sectional, so we observed just a snapshot. It might be that a lot of those women who say they have the number of kids they wanted actually wanted more, but they’ve changed their desires to fit what they had because it's uncomfortable saying you didn’t meet your desires. 

On the other hand, it might be that some women who say they have more than they wanted actually did have the number they said when they were 15, but then their life went badly and they now feel like their life would have been better if they’d had more or fewer. That’s another kind of rationalization. 

We can’t control for those biases in our study. However, there are also longitudinal studies that show that, yes, rationalization and bias occur, but however you measure it, people on average tend to have fewer children than they say they desired whether you measure it using their preferences early in life or late in life.

In the cross-sectional data, we can clearly see that under- or overshooting, or not hitting your target, is associated with less happiness. Unfortunately, the longitudinal data sets that we have that do this don’t have great questions about happiness. Still, in the kind of oblique way we can get at it, it seems like in the longitudinal data that under- or overshooting is associated with happiness costs. But the data is not great yet because, unfortunately, subjective well-being is not as widely surveyed as I think it should be.

Ultimately, if we’re thinking about how we make policies, happiness is fleeting but fulfilling people’s freely expressed desires is straightforward. We should be designing policies that meet the reasonable desires that people express. We should not be trying to target their happiness that they don’t even know how to target. Like, what will make you happy? Can you perfectly target that for yourself?

If people say they want to get married, banning marriage is a bad idea. Think about this in the context of same-sex marriage. It doesn't matter if getting married will actually make gay people happy or not. The fact that they said they wanted it meant that not allowing it was seen as an encumbrance of their freedom and ipso facto a bad thing. More broadly, the whole point of a free society is if you want it, you can get it.

A: Let’s put some actual numbers on this. In your sample in Canada, you stratified life satisfaction across three groups. The unhappiest group was women who had more children than they considered ideal — their self-reported life satisfaction was 5.4 out of 10. Women who had fewer children than their stated desires scored 6.8, and women who achieved their ideal were at 7.2. I’m wondering if you can elaborate on that gap, because the difference between 6.8 and 7.2 is much smaller than 5.4 and 6.8. 

L: Having more children than desired has a much bigger effect on happiness. The reason for that is that excess children come with a price tag, whereas missing children don’t. Having an excess child could mean, for instance, you’re suffering financially for the rest of your life. It has follow-on effects that a missing child doesn’t. Arguably, if we had a social welfare state that compensated for the cost to parents of raising children, we could eliminate some of that.

The second thing is that yes, excess children still mean having a thing happen that you didn’t want to happen to you; humans have loss and harm aversion. Having a bad thing happen to you will always hurt you more than not having a good thing happen to you. 

But missing children are so much more common. We did a back-of-the-envelope calculation where we multiplied the prevalence of missing children by the welfare cost and excess children by the welfare cost, and the total welfare loss from missing children was a whole lot more. We didn't publish an exact number because that kind of utility math is gross and weird. 

A: Unless you're an effective altruist.

L: I’m happy to go on record saying that I think the repugnant conclusion is misunderstood.

There are a lot of reasons we didn’t publish an exact estimate on this. But, certainly, if you’re talking an order of magnitude, these are comparable harms to society. This doesn’t mean we should ignore the problem of excess fertility. We shouldn’t. It is good that society provides women with free or low-cost contraception in virtually every country.

A: When you looked across income levels, the gap between desires and intentions is consistent. Even in households earning above $200,000, women don’t meet their fertility desires. It brings to me the question of whether anyone achieves their fertility intentions. And if so, who?

L: The binding constraint is not that you didn’t want it hard enough, and it’s not even that you didn't have enough money. The reality is that there is a limited number of days in your life on which you can conceive a child and they are running out with every second. That is true for the rich and the poor. Yes, rich people can use IVF, but it doesn’t work that well. Life course factors influence how long it takes you to meet a partner, how long it takes you to finish school, how long it takes you to feel comfortable in your marriage, your house. Time, demands of your job, etc. — these are the big barriers that people face. Rich people do have more kids, but it's not just about money.

A: How does the calculus change, if it changes at all, when men are asked? 

L: In low-income countries, men tend to have higher fertility desires than women. That’s typically because in those countries, you tend to have more gender-specialized views of work and family. In traditional societies, you tend to have a stronger emphasis on women's status, prestige, and worth as being derived through home and children. Men tend to hold that view more than women. So, in lower-income societies, there tends to be a gap between men and women. 

In lower-income societies, men also tend to report fewer children than women. Men often do not acknowledge their biological children born out of a marital union, which means in lower-income societies, you really cannot measure the fertility gap among men because they’re giving much higher responses on desires when they’re not necessarily the ones bearing the cost. 

In higher-income societies, you still get a problem of men not reporting paternity. In the U.S., about 11% of babies do not have an acknowledged father. In most of the rest of the developed world, it ranges from 2% to 6%. This means that you underestimate fertility for men. But in most high-income countries, men and women have virtually identical desired fertility. 

A: How do you tell that you are measuring preferences as distinct from intention? That seems quite tricky. I'm curious, in terms of survey design and demography, how you can isolate those things.

L: There are a couple of ways. One is you can experiment with a bunch of different words. At different points in the survey, you can ask them how many kids would be ideal for you to have and how many kids do you want to have. Then, at different points of the survey, you can ask them, how many kids do you plan to have and how many kids do you intend to have. If the plan and the intention cluster close together and the ideal and the want cluster close together, then you've identified two concepts. However, if all of them cluster together, what are you really measuring? 

It’s a fuzzy concept but it is highly predictive behavior. Fertility intentions are more predictive of behavior than fertility preferences for the obvious reason that they are intentions to take action. Fertility preferences are predictive of changes in intentions longitudinally.

A: When you are studying fertility, you brush up against sacred cows: Is this a backdoor argument to keep women out of the workplace? Is this conservative denial of climate change? Those sorts of things. I'm wondering what you think about the politicization of fertility and family planning and how you navigate that in your work.

L: Politics is becoming more and more predictive of fertility. The gap in fertility behaviors along ideological lines is growing over time, and that is because the question of sex and the body is becoming more and more of a dividing line in our society. And that’s essentially because of biomedical advances. As we spend more time in our bodies because we live longer and as we can make our bodies do more interesting things using pharmaceuticals and surgery and all these things, the body becomes a bigger locus of political debate. That’s going to continue to be true and it’s going to become more so. Ultimately, these are going to become more contentious.

Population is fundamental to every political question. This is why I think fertility preferences are really important because while there are many things we disagree about in society — climate change, women’s place in whatever domain or sphere you want to talk about — what most of us tend to agree with is that if somebody says they want something and that something is fairly reasonable, especially if that something creates major benefits for other people in society — and having a child creates major benefits for other people in society — we at least shouldn’t oppose, and it might be worthwhile supporting it.