Are larger families actually more religious?

In a society where artificial birth control is widely available and accepted, having a large family is increasingly seen as a choice taken primarily by people outside the mainstream, often for religious reasons.

But is that actually the case?

Sociologist Ryan Burge recently took a look at the question on his site, Graphs About Religion, and concluded: “It’s fair to say that large families are more likely to go to church, but not by that much.”

He found that 33% of families with four or more kids went to church at least weekly, while among families with any number of children, 29% went to church that often.

So is the association of large families with religiosity a myth? What about the stereotype of the full-size van pulling up to Mass with a small army of children?

The Pillar looks at the numbers.

Credit: Shutterstock.

Burge pulled his data from the 2024 Cooperative Election Survey, a large survey primarily for political analysis but which also asks demographic and religious questions.

The survey is very large, interviewing about 60,000 people every two years. This is helpful, as analyzing small portions of the population requires as large a sample size as possible.

To get the maximum power in our analysis, we used the most recent three CES surveys – 2024, 2022, and 2020 – and combined them into one dataset.

Our goal with this analysis is to look at the characteristics of large families, and in order to tell larger families from smaller ones, we need to look at the age at which people are unlikely to have more children. However, we also want people from roughly the same generation. If we mixed people born in 1980 with people born in 1940, we’d be thrown off by the fact that larger families were more common 40 years earlier, and also that the generation born in the 1940s was generally more religious than those born in the 1980s. This would artificially make larger families look more religious when we’d actually be seeing that older families were both larger and more religious.

So to get a consistent group of people to focus on in the CES survey, we looked at people who were aged 40-55 when surveyed. These were people who were likely to be done having children, but also were not yet too far from their childbearing years. Since the surveys were taken in 2020 through 2024, this means we are looking at people born between 1965 and 1984.

In those three survey years, there were a total of 38,300 people aged 40-55 who provided answers on their number of children, and the breakdown in the number of children they had is as follows:

One factor which seemed important in this analysis was to take marriage into account. When people envision the “big religious family,” what they tend to think of is a single couple who have had a large number of children together. However, there are other ways to get to a large family – there might be two people who each had children from prior relationships, or a single person who had multiple children with various people.

The CES survey does not ask questions that distinguish between children the respondent had with their current spouse and children from other relationships, but it does ask questions about marital status, so it is possible to break down the number of children people have by their current relationship status.

Unsurprisingly, people who are currently unmarried (never married, domestic partnered, divorced, or separated) are more likely to have no children than married couples. Among married couples, the most common number of children is two. When it comes to large numbers of children, 5.1% of married couples have five or more children, while 2.9% of those who are not married have that many.

Of the 1476 survey respondents who had five or more children, 65% were married or widowed, while 35% were currently unmarried.

Looking at people who regularly attend religious services, married people and unmarried people have very different trends.

Among married people aged 40-55 who have no children, 17% attend religious services at least once a week, a rate virtually identical to unmarried people with no children (16%).

However, according to the CES dataset, married people with children were more likely to attend religious services at least weekly the more children they had. Those with five or more children were the most likely, with 44% attending religious services once a week or more.

For unmarried people, however, the rate of religious attendance did not increase with the number of children. Among those with five or more children, 16% attended services at least once a week, the same as the rate among unmarried people with no children.

This tells us that going to church often is more common among married people with larger numbers of children. But the absolute number of families with large numbers of children is still small overall. Is it true that one could expect to see significantly more large families at church than in a secular context?

The answer is mixed. Among married people aged 40-55 who attend religious services at least weekly, parents with five or more children make up only 8% of the total. However, that is twice as many as among similarly aged married people who seldom or never attend religious services, among whom parents of five or more children make up just under 4%.

This is also true of people with three or four children. If we look at people who have three or more children, they make up 39% of married people aged 40-55 who attend religious services at least once a week, but only 23% of those who seldom or never attend.

What about differences between adherents to different religious traditions?

The groups at times called the “nones” (atheists, agnostics, and people describing themselves as “nothing in particular”) are the largest groups within our analysis sample. About a third – 32% – of married people between 40 and 55 in the sample were atheists, agnostics, or describe themselves as “nothing in particular.”

However, while these different “nones” were alike in almost never going to religious services, they are different in terms of the number of children they had.

Atheists were the people in our analysis group most likely to have no children at all, with 35% of married atheists aged 40-55 being childless. They were also some of the least likely to have five or more children, at just 0.86%.

Agnostics were somewhat less likely to have zero children – 30% – and more likely to have five or more – 2% – but still had significantly fewer children than the average.

People who describe themselves as “nothing in particular” closely resemble the average population, with about 5% having five or more children, 22% having no children, and the most common number of children being two.

When analyzing Catholics, there were significant differences based on the frequency with which Catholics reported going to Mass.

Catholics who go to church at least once a week were more likely than others to have larger families, with 8% having five or more children. Similarly, Catholics in our analysis group who attended Mass at least once a week were some of the least likely of any group we looked at to have no children.

But interestingly, Catholics who seldom or never attend Mass were the next most likely to have larger families, with 5% having five or more children. They were, however, more than twice as likely than others to have no children. Catholics who reported going to church monthly or yearly were somewhat less likely than the average person in our analysis to have five or more children.

This split, with the larger families among both the most frequently church-going Catholics and the least, suggests there may be a dynamic in which there are many ways of reaching large family status: one showing up in the larger families of the Catholics who do not often attend Mass, while another perhaps very different model showing up among the most frequent churchgoers. In the middle, we seem to have smaller families and churchgoing when convenient.

Results for Evangelical Protestants were very similar to those among Catholics, with those who go to church most often being the least likely to have no children and more likely to have five or more. Among Evangelicals, the numbers for those who go to church monthly were almost the same as those who go weekly or more, while those who reported going less than once a month had fewer children.

Mainline Protestants had a different pattern, with no association between more frequent church attendance and the rate at which they had families of five or more.

This very different pattern among Mainline Protestants suggests that whatever connection there is between active participation in Catholic or Evangelical faith and having larger families does not exist within the mainline denominations.

Two final groups represent a small enough percentage of the U.S. population that their results have some uncertainty, but they are nonetheless interesting because they are such outliers.

Among Mormons who attend religious services at least once a week, we saw the highest rates of people with five or more children (and the lowest rates of those with zero) of any group within our analysis. Twenty-seven percent of married people aged 40-55 had five or more children, while only 2.8% had zero. Among Mormons who reported never or seldom attending religious services, the number with five or more children was below the general average, just 1.3%, though they did have an unusually high percentage of people with four children.

Another interesting result was among Jews. Forty-five percent of Jewish respondents described themselves as seldom or never attending religious services, and among these the level of childbearing was similar to atheists, with only 0.7% of married people aged 40-55 having five or more children, and 31% having none at all.

However, among those Jews who did attend religious services weekly or more, 16% had five or more children, a rate about twice that of Catholics who attended religious services at a similar frequency.

We did not have a sufficient number of people in the survey dataset who were Muslim, Hindu, or Orthodox to provide results for those groups.

Among the most difficult questions to answer from survey data is what the people who responded to this study looked like in terms of practice. We have a response telling us which Catholics say they go to church at least once a week, but which people go to confession frequently? Which ones believe in the Real Presence? Which ones believe that the Church’s teaching on artificial contraception is authoritative? The answers to these questions are unknowable from a generalized survey like this.

One small window that is available into these families regards demographic questions the survey does ask.

Some commentators suggest that the emphasis on theology (including John Paul II’s Theology of the Body) is something which appeals more to educated Catholics than others, so we looked at Catholics with at least a four-year college degree to see if the difference between those who attended Mass at least once a week and those who attended least would be greater than for Catholics in general.

However, the opposite was true. While for the broader Catholic population, 8% of those who go to church at least once a week have five or more children, among Catholics with at least a four-year college degree, that number was a little lower: 7%.

Another occasional line of commentary suggests that having a larger family is a luxury status symbol in which only higher earning families can indulge. So we broke down the rates at which Catholics in the data had different numbers of children according to whether they had an annual family income above or below $100k.

The less affluent Catholics actually turned out to be more likely to have large families than those making over $100k.

They were also more likely to have no children, while married Catholics making over $100k who go to church at least once a week were more likely to have two or three children.

To get a look at more details about the marriage histories of people in relation to their number of children, it is necessary to turn to another dataset, the National Survey of Family Growth (NSFG) from the U.S. CDC. This is a smaller survey, so it can be harder to zero in on some smaller groups (such as specific religious traditions), but it does allow us to understand more about the marriage and family history of the respondents.

One limitation with the NSFG, which creates some differences in how we had to analyze data, is that the survey is designed to look at growing families. As such, it’s primarily a survey of people still in their childbearing years. Since we’re trying to look at people with larger families, that’s a group which is easier to identify as people near the end of those years. In the CES data we were able to do that easily by looking at people aged 40-55.

The NSFG has changed over the years. From 2011 to 2014 it surveyed adults up to 44 years old. From 2015 to 2019, it surveyed adults up to 49 years old. Due to the sizes of the surveys in different years, the largest consistent analysis group we could get was looking at men and women aged 35-44 from 2011 to 2019. Clearly, people who are 35 years old might go on to have more children, but we wanted to have at least a 10-year age range, and 35 is at least old enough that our analysis group should be near the maximum number of children they will have. However, we did also look at the smaller group of respondents who were aged 35-49 from just 2014 to 2019 in order to confirm our findings, and we found both analysis groups to be directionally similar.

Because the NSFG is designed to look at family structure and history, we were able to make what turned out to be a very important distinction. When we looked at large families in the CES survey, we could identify people who were married and had a certain number of children, but we did not have data on whether they had those children with their current spouse, or if the children came from prior relationships. The NSFG allows us to separate those two groups.

For married people with 0-3 children, 65% to 75% of the time, the respondent had had all of their children with their current spouse, and that current spouse had entered the marriage with no children from previous relationships.

However, for married people with four children, this dropped to 56%. And among married people with 5 or more children, in only 39% of cases had the respondent had all of their children with their current spouse, while the spouse also had no outside children.

This is a different model of family than people may initially imagine when they picture six kids piling out of the van to go to Mass. In the wider U.S. population represented by this survey, people who have five or more children are more likely to have gotten to that point because they have had multiple relationships which resulted in children over the course of their lives.

However, when we looked at how often people attended religious services at least weekly, the rates diverged dramatically between those who were in marriages where both spouses had only had children with each other, and those where one or both spouses had children from previous relationships.

Using the CES data, we were able to find that there were different rates of religious service attendance between unmarried and married people with large families. With this data, we are able to further separate married people into those who had all their children with their spouse (who also had children exclusively with them) and married people who had children from prior relationships and/or whose spouse had children from prior relationships.

The church attendance rates for married people where both spouses have only had children with the other spouse is much higher than any group we have analyzed up to this point.

We narrowed the analysis to look specifically at Catholics and found a similar pattern.

In addition, with this level of relationship detail it becomes clear that regardless of the type of family, Catholics with more children are more likely to attend Mass frequently than those with fewer.