On Friday, Gallup released a really fascinating study of religiosity in the United States called Provo-Orem, Utah Is Most Religious U.S. Metro Area. It’s a huge survey showing the percentages of religious and non-religious people living in different areas of the United States. There’s a pretty wide spectrum, with just 17% of people saying they’re “very religious” in Burlington, VT and Boulder, CO but the other extreme is Provo-Orem, UT with 77% very religious people. I live in Utah, and Provo is about a 30 minute drive from my house so the study caught my attention.
I wanted to know more about this. What kinds of lives do people live in these different places? A good way of finding out is to compare per capita income in religious places vs. non-religious places. Surely there would be some trends in the data.
The Gallup study lists statistics for US metropolitan statistical areas, which include cities and their surrounding areas, not just the cities themselves. From the US Census Bureau I found a very detailed report called Personal Income by Selected Large Metropolitan Areas, which is report #683 from the Income, Expenditures, Poverty, & Wealth statistics page. It’s a huge table showing overall and average personal income figures for each US metro. I created a chart combining the census data with Gallup’s survey (download: us_religious_trends_by_metro) to compare incomes for each statistical area based on religious prevalence. I added one calculation to the data, a figure I called “believer ratio” which is simply the number of highly religious people for every non-religious person for each metro. The income data is from 2009 because it was the most recent information available.
Here’s what the data shows.
It’s clear that per capita income decreases as the percent of highly religious people increases. In short, religious places are poorer on average. Now, I think it’s important to use caution when drawing conclusions here. Perhaps poorer, humbler people are more inspired to be religious, or perhaps religion causes people to be poorer. Maybe it’s something else entirely, but causation is not evident.
That orange blip on the far right is Provo-Orem, UT. This metro is lies well outside the main group because of its high percentage of highly religious people (almost 80%), but it falls neatly along the trend line because it’s also predictably poor. Locally there are a lot of explanations for this, because Provo-Orem is a college area with poor students and local families have lots of kids. Even so, the Provo-Orem area appears, proportionally, to be exactly as we would expect it to be: religious and poor.
Non-religious percentages have the opposite effect. In areas where non-religious people predominate, per capita incomes are higher. In fact the R-squared figures are almost identical, showing the same income correlations for prevalence of non-religious people vs. prevalence of highly religious people. Provo-Orem, UT is the higher of the two dots on the lower-left and McAllen-Edinburg-Mission, TX is below it.
This is a scatterplot showing per capita incomes vs. the “believer ratio” I calculated. Again, the believer ratio is the number of highly religious people for every non-religious person in a given metropolitan area. There’s an obvious clump of metros on the left side of the graph, showing that cities usually range from about 1:1 to around 3:1 highly religious to non-religious people.
In this case the Provo-Orem statistical area is second from the right, and Jackson, MS is on the far right. The trend is as we would expect from the previous graphs; higher proportions of highly religious people correspond to lower per capita incomes. It’s worth noting, however, that the correlation here is a little weaker than the correlations we saw previously, meaning that believer ratio isn’t quite as reliable as a predictor of per capita income although there is a clear trend.
I was also curious about whether larger metropolitan areas (big cities) tend to have higher or lower percentages of highly religious or non-religious people compared to smaller metropolitan areas (towns and small cities). My completely uneducated guess was that big cities would tend to have larger percentages of non-religious people. Here’s the data.
There are some pretty weak correlations here. The first two graphs, showing the number of highly religious people and non-religious people compared to the populations of metropolitan areas, indicate that bigger cities do tend to have more non-religious people, but only just. This means that my belief that big cities were much more non-religious was not really true. Percent of people who are highly religious or non-religious isn’t a very good indicator of population size.
The third graph, which compares population to believer ratio, also shows a weak correlation but it reveals something interesting: large metropolitan areas are much more likely to have a balanced believer ratio of 1:1. In fact, the three largest metropolitan areas of New York, Los Angeles and Chicago are nearly balanced. Other large metropolitan areas with greater than five million people show Philadelphia, Miami and Washington, D.C. hovering close to 1:1 while Dallas, Houston and Atlanta hover close to 2:1. From there, the more unbalanced metros are far smaller.
This got me wondering about the believer ratio. What happens when we simply graph highly religious percentages against non-religious percentages? It looks like this.
As you would expect, there is a very strong correlation here. As the number of highly religious people in an area goes up, there is a very predictable drop in non-religious people. There’s a maximum here too, since we can’t have more than 50/50 highly religious people to non-religious people, and the same goes for 60/40 or 70/30 and so on. We can imagine the maximum limit as a line that goes through these points. Here’s a very interesting graph showing the maximum line.
Points that sit close to the maximum line can be considered closer to maximum polarization. The main group of metros is clumped along a predictable line showing that while cities may be more or less religious, they retain very consistent proportions of highly religious people and non-religious people. Yet there are a few points that deviate noticeably from the main pack, especially the one to the far right. That’s Provo-Orem, UT.
It turns out that the metropolitan areas with the greatest religious polarization are the three located in Utah. These Utah metros–Provo-Orem, Ogden-Clearfield and Salt Lake City–are located in the red shaded area I like to call the Utah Zone. Every other area lies outside this zone. For those of us who live in Utah, this might confirm what you already suspected. Even the Salt Lake City metro, which doesn’t have an unusual percentage of highly religious or non-religious people, does have an unusual mix of the two. Ogden-Clearfield is a bit more unusual, and Provo-Orem is the most unusual of all. People living in Provo-Orem, far more than any other place in the United States, are more likely to fall to one religious extreme or the other. As the graph’s subtitle suggests, living in Utah can really suck because it is the most religiously divided place in the nation.