I’m a Church growth amateur; occasionally I enjoy dropping by Matt Martinich’s blog to look at the latest temple predictions, and I’ll often skim through headlines about the latest data point on Church growth and what it means. However, for some time now I’ve been suspicious that we’re reading way too much into the natural jitters in the data. Church growth was 1.2%, now it’s 1.5%, what does that mean!? What gets credit?
In data science this is what we call “overfitting.” Sometimes there’s a random blip in the data that we interpret as meaningful change when in reality, it’s just a random blip. If we’re oversensitive to variations we can read trends into the data that aren’t there, and we’re definitely at a high risk for this when we only look at the data once a year when the latest numbers are announced at spring General Conference.
In reality things like the growth of a relatively developed religion follow slow-moving, long-term patterns. To use a metaphor, some social phenomenon are like a large boat that requires a long time to turn left or right, and short of some nuclear level event like the President of the Church abdicating and declaring the Church a fraud, it is likely that growth trends won’t drastically swing in a real, meaningful sense on a dime.
And even events that might be considered “nuclear level” don’t turn out to be. For example, the Jehovah’s Witness hierarchy predicted that the second coming would come in 1975, and put a lot of real energy into preparations. While my understanding of that event is that the hierarchy always technically maintained some plausible deniability in case it didn’t happen, the fact is that it was a huge deal when it didn’t, yet the aftermath of the “failed prophecy” event is barely noticeable in their growth rates.
So what are the overall trends for the Church of Jesus Christ of Latter-day Saints? To figure this out we need to ask questions about our measurements.
For some time now there’s been chatter about how Church membership numbers are notoriously inaccurate because of people who are on the rolls but do not identify as Latter-day Saints, or who the Church has completely lost track of. If you’ve ever gone through the ward list A to Z in a non-US area, you’ll find that these people constitute a significant portion of the Church membership.
Fair enough, but what about ward and stake growth? Wards and stakes require active members, not just members on the rolls, so they are going to be a more accurate indicator of church growth as a percentage than raw numbers of members. For a while now I’ve assumed that ward and stake growth tells a significantly different story because of baseball baptisms and the like, but I had not actually seen a data visualization comparing the different measures, so I made one (hat tip to Clint Kimball for compiling the data from the Church Almanac and other sources).
The data do indeed show bumpiness, but within overall trends. Here I took two different timeframes (1900-2021 and 1950-2021) and applied a variety of trendlines and smoothing approaches (technically speaking, a 6th order polynomial and a 10-year moving average) to try to cut through the noise and uncover the real trends in Church growth post-1900.
As seen, when we cut through the noise, my earlier preconceptions notwithstanding, measuring stake growth and ward growth does not tell a terribly different story than raw, on-the-rolls, membership growth, which is that church growth was low during the Great Depression, picked up during World War II, saw its apex during the late 1970s/early 1980s, and has since declined rather steadily until now (with a possible plateau from 2010 on for ward and stake growth). For the wonks, when ward and stake growth is compared they correlate to a .59 level, with membership growth and stake growth correlating to a .51 level, so the two aren’t synonymous, but they are similar. (I don’t know how much the history of changing requirements for stake formation plays into all of this).
There are a few points that rise to the surface when the different timeframes are considered. For example, the half century trendline puts more weight on the abnormally low 2020 year, which is presumably why stake growth saw a downward decline that was not picked up when the whole century is considered (which actually shows a spike in stake formation near the very end). The next few years will tell us whether 2020 was just a fluke year and stake formation rates are actually trending upwards (likely, given COVID), or whether it portends a real change.
With the exception of the COVID-influenced 2020 growth decline, this suggests that the chatter one hears about a recent sharp decline in Church growth rates (whether due to the Internet or the particular social hobby horse of the interlocutor) isn’t really born out by the data (if anything the decline may have stalled out in the 2010s for ward and stake growth). Church growth rates are basically following a pattern that they’ve been following for several decades, and the story is basically more of the same whether we’re talking about new wards, members, or stakes.