Zelophehad's Daughters

How has interest in different churches changed during the coronavirus pandemic? This is a question that occurred to me while I was working on my last couple of posts where I looked at Google Trends data about the LDS Church in particular.

In this post, I’ve gathered Google Trends data on a bunch of different churches and I’ll show daily 2019 vs. 2020 comparisons for each one. I’m only making comparisons to Christian churches, and my list is pretty US-centric, both because I went with what I was most familiar with. Here’s the complete list, along with what Google Trends categorizes each as. Note that I went with what looked like a high-volume search term for each church or denomination, so for example for Methodists I chose the United Methodist Church, but for Baptists, which represents lots of different churches, even the biggest organizations (e.g., Southern Baptist Convention) had far lower volume, so I just went with “Baptists.” In addition to traditional denominations and fringe groups like us and the Jehovah’s Witnesses, I included a couple of big megachurches because I understand they’ve been growing a lot relative to the old mainline denominations. I’ve ordered them as fringe groups first, followed by mainstream ones sort of from high (more ritual) to low (less ritual) (based on nothing more than my sense of them), with the megachurches at the end.

As commenters on my first post pointed out, in the particular case of the LDS Church, there are reasons to doubt that this is that good a measure of interest in the Church. And what I found in my second post supported that, with far different results for a search term used by members of the Church versus by non-members. Given that difference, with all these other churches that I’m far less familiar with, who even knows what other complications I’m overlooking? There are probably a lot, but I think the data are fun to look at anyway if you just keep your truckload of salt handy.

Also, while I’m bringing up reasons to be leery of the Google Trends data, let me show you something about their scaling that makes me a little crazy. I noticed when I accidentally moved the end date of a time series one day forward or back that it changed the whole rest of the series, rather than just omitting the day in question. This suggests to me that the scaling to 100 that the help mentions isn’t all that Google Trends is doing. Rather, it’s aggregating a bunch of data and probably smoothing it together with some kind of model under the hood.

Here, let me show you what I’m talking about. The graph shows daily results for worldwide searches for “Mormon” from April 1 of this year through different ending days.

Read More

I posted a few days ago about change in interest in church (in the LDS Church in particular) during the coronavirus. I used Google Trends data on searches for the Church. Commenters on the post (and on Facebook) made some excellent points about how differences might be attributable to change in interest from Mormons versus from non-Mormons. In this post, I’m going to make a quick attempt to tease that apart. (Commenters also raised other good critiques, such as that a change in Google Trends data might not reflect a difference in organic interest so much as a structural change in how church is happening during the pandemic, that are beyond the scope of what I will look at in this post.)

My approach is simple. Rather than looking at Google Trends data for just the Church itself, I’ll look at data for two different Mormon-related terms, one that’s more often used by Mormons, and one that’s more often used by non-Mormons. For the term more often used by Mormons, I chose “Come Follow Me,” an excellent suggestion Ardis made on my last post. Given that it’s a pretty well-known statement Jesus made in the New Testament, I wouldn’t have guessed how much its usage would be dominated by Mormons, but it clearly is. Looking at the Google Trends data, here are reasons I think it’s mostly being searched by Mormons:

• Search volume is the highest for the US, and within the US, for Utah, followed by Idaho.
• There are weekly Sunday peaks that disappear on Conference weekend.
• The top five associated search terms Google Trends suggests are all Mormon related.

For the term more often used by non-Mormons, I’m using “Mormon,” which is ideal because President Nelson’s campaign to get us to stop saying it is clearly having an effect on people in the Church, but, not surprisingly, hasn’t been so quickly heeded by people outside the Church. As with “Come Follow Me,” the data looks consistent with the expectation of who’s using it, as the peaks are for Mormon-related news stories rather than for General Conference, when searches for the Church peak.

This graph shows daily results for “Come Follow Me” since the beginning of 2019.

Read More

I’ve seen the possibility floated in a number of places (e.g., in this W&T post) that when the Church resumes regular Sunday meetings, fewer people will go back than were attending regularly before the pandemic. I can certainly see the argument for this. It’s much easier to continuously keep up with a practice like weekly church attendance than to stop and start, particularly if there were already barriers making attendance difficult (distance, hearing offensive things, etc.).

I was wondering if there’s any way to forecast the degree to which this might happen by looking at what people are doing now, while not attending church. Of course this is an extremely difficult thing to measure. Not even individual wards, which typically have at least some sense of activity level from sacrament meeting attendance if nothing else, likely have much idea of what’s going on. But Google does. Google Trends tracks what people are searching for online. While this is obviously far from a perfect measure of how engage people still are with church while not attending, it might give us at least a little insight.

There are a few important things to know about Google Trends data:

• You can get results by countries, states/provinces, and even metro areas. I wasn’t interested in geographic differences here, so everything I’ll show comes from worldwide results.
• It doesn’t give absolute numbers–like the number of people who searched for something–but rather relative search numbers, comparing across search terms or across time for a single search term. It always gives results normed so the highest value is 100.
• It lets you search for words (literal searches people have performed) or topics, where related searches about the same thing are aggregated. I chose to search for the topic “Church of Jesus Christ of Latter-day Saints,” because I suspect Google is smart enough to aggregate searches like “Mormon church” and “LDS church,” and also because their help says that topics aggregate across languages. I also tried a few other LDS topics, but they gave a far smaller search volume, so for this analysis, I just stuck to that single topic.
• You can get results back to 2004 and up to a few days ago. The granularity of the data you get depends on how long a time frame you ask for. If it’s short enough, you get daily data. If it’s longer, you get weekly or monthly. I wanted daily data from January 2019. This is a long enough period that if I asked for it all at once, I got weekly data. My solution was to ask for daily data in a series of three-month periods. Unfortunately, this means that each query for a three-month period was normed to a different maximum value. In order to make the results from the different three-month periods comparable, I set them to overlap a little. By doing this, I got two results for the days that appeared in both periods, and I could use the ratio between the values given for those days in the two periods to re-norm one period so that it was comparable to the other. I used a series of re-norming steps like this to stitch together the entire daily series from January 2019 to May 2020.

The graph below shows relative search volume for the Church for the entire period.

There are two extremely clear patterns: First, there are big spikes at General Conference time. Second, there are weekly spikes every Sunday. Both of these are probably unsurprising.

Read More