What Did American Religion Look Like Before Modern Surveys Began?

Using the General Social Survey to Estimate Religion in the Early 20th Century.

Here’s the thing. We don’t really have any good religion data before 1972. That’s when the General Social Survey first started asking people questions. Sure, there’s the American National Election Study which goes back to 1948, but I just absolutely despise that dataset. If someone wants to know more about that, I would be happy to write up a thousand words about everything it does poorly. And I would only scratch the surface.

So, for all intents and purposes, the GSS is the only tool I have when looking at long term historical trends.

I woke up this morning with this weird realization. I can actually use the GSS to get a glimpse of what religion looked like before the General Social Survey fielded its first questionnaire in 1972. How can I do that? Well, the GSS has been asking folks about the religion in which they were raised since 1973.

Here’s what the means, practically speaking. In the 1975 wave, they asked a sixty-year-old what religion they were raised in, which means that if I look at the responses to those questions among a bunch of people born in the 1910s, I can get a pretty good approximation of what religion looked like during that time period. Which means I can create a timeline of American religion dating back at least a hundred years.

First, I want to show you just how that is possible. I plotted the birth years of every single person who took the GSS between 1973 and 2018. The colors are decades of birth. This makes it crystal clear what I am talking about. There are tons of people in the GSS who were born in the 1950s. But there are also really robust sample sizes for people born in every decade dating back to about 1900.

I have to mention this caveat here before we get to the good part. People do not have perfect memories of everything from their childhood. I think we can all agree to that. But I do think there’s ample reason to believe that folks are going to be pretty accurate when talking about the religion in which they were raised. It’s not like it was a one-time event. So, this is not a perfect approximation, but I think it’s pretty darn close.

So, let's get to it. This first graph is the broadest look at American religion. I broke folks into four groups: Protestants, Catholics, No Religion, and Everyone Else. Obviously, that’s not fair to the non-Christian faiths, but honestly none of them are large enough on their own to actually shoot a trend line. Way too noisy. So this is pretty simplified, but there’s a lot to learn here.

Let’s just state the obvious - Protestants absolutely dominated American religion around the turn of the century. About two-thirds of all folks born around 1900 were raised in a Protestant household. Catholics were a very distinct minority back in those days, with less than 20% of households being Catholic in the early 20th Century. The nones were also basically non-existent. About 2% of folks born during this time period said that they were raised without religion.

The trend lines are certainly interesting, as well. Protestants have gone from being around 65% of all households to just under 40% among those born in the 1990s. Still the plurality, but most certainly not the majority. At the same time, the Catholic share has exploded from 18% in 1900 to 32% in the 1990s. Nearly doubling!

The nones have also experienced an increase, but it has been incredibly gradual. From 1900 to 1960, the share raised without religion went from basically zero to about 5%. Then the line does begin to bend upwards a bit. In the most recent data, about 15% of folks are raised in a non-religious household. It’s interesting to see that the Catholic numbers in 1900 are just slightly higher than the nones share in the 1990s.