City-Level Data in the Census of Religious Bodies

Lincoln Mullen

When the Census Bureau published its reports on the Censuses of Religious Bodies, it reported the data at different levels of aggregation. One of the most interesting types of data they reported is the number of congregations and members for denominations in major U.S. cities. The Bureau defined a major cities as one with a population about 25,000. Think Port Huron, Fort Smith, Revere, and other similarly sized cities and above.

Our team has been hard at work transcribing this city-level data from the published reports for 1906, 1916, 1926, and 1936. We plan on releasing both some maps of this data as well as the datasets that we have transcribed so that other scholars can use them.

For today, though, I want to give an overview of the data for congregations in cities counted in the 1926 Census of Religious Bodies and give an example of two ways that the data can be visualized.

First, how many cities were counted? For 1926, 292 cities made the cut, ranging in size from Kingston, NY, which was just barely over the cutoff, to New York City. All but one of those cities met the criterion in both the 1920 and the 1930 population censuses, so for most cities it is possible to make reasonable guess of their population in 1926 by interpolating between the two population censuses. In other words, we have data for a fairly wide swathe of American cities. And we have data for far more cities than have typically been studied by scholars of American religion. It is fair to say that Bangor, Maine, or Bellingham, Washington, are unlikely to receive major scholarly attention. And so while we know a great deal about religion in New York, Chicago, and Los Angeles, this data can at least give us a glimpse into cities less studied.

Second, how many congregations and people were counted? We’ve earlier discussed the limitations of how the Census Bureau counted religious groups in general, and paid specific attention to how they counted Jews and Black Baptists, so we won’t go over that ground again. Even with those limitations, the 1926 census counted a total of 232,154 congregations. Of those, 34,116 were in the major cities. The Census Bureau also enumerated a total membership in the country of 54.6 million people, of whom a membership of 26.6 million were in major cities. In other words, 14.7% of the congregations enumerated were in major cities, but 48.7% of the total membership. This is not surprising, given that urban congregations had on average a larger membership. But it does give a sense of the scale for the divide between congregations in rural places (which the Census Bureau defined as having a population of fewer than 2,500 people), towns and minor cities, and major cities.

Having a sense of what is in the data, we can ask some basic questions about the data by visualizing it. For now, we will pass over the most obvious question about mapping different denominations in space … but only because we have a more complicated visualization soon to be released which will let you see that approach. Instead, we can ask two questions that relate to religious diversity in American cities. While the metric for calculating this kind of diversity is imperfect and rough—some might even say, crude—it does offer a way of looking at the variety of all major U.S. cities, rather than one or two at a time.

We can observe that the number of distinct denominations that the Census Bureau counted within a city is a rough metric of religious diversity. I emphasize that the metric is rough for a number of reasons. Most important is all the kinds of people the Census Bureau did not count, most notably Muslims. Then too, the concept of a denomination is a slippery one, and we might be better off, for instance, aggregating all the kinds of Methodists into a single category, or perhaps highlighting the distinction between predominantly Black and predominantly White denominations. While all those considerations are important, for now we will use the most straightforward measure.

Denominational diversity in American cities, 1926

Figure 1. This plot shows the number of distinct denominations, as counted by the Census Bureau, versus the population of the city. It thus shows a rough idea of the number of types of religious options available in any given city. Population in 1926 has been estimated from the 1920 and 1930 population censuses using linear interpolation, and it is plotted on a logarithmic scale.

As we see in this plot of the number of denominations relative to the population of the city (figure 1), the number of denominations scales rather predictably with the size of the city. This is not surprising, and boils down to a common sense observation that the more people are in a place, the more kinds of people there are likely to be in that place. But we can note that within that broad trend, there is some surprising variation. Of three cities all having the same population of about 100,000 people, why would Somerville, Massachusetts, have many fewer denominations than Canton, Ohio, or Tacoma, Washington? Likewise, why did Jersey City have fewer denominations than its relatively near neighbor Newark? There is a great deal more to be learned following these clues.

Percentage of city population counted as members of congregations in major American cities, 1926

Figure 2. This plot shows the percentage of a city’s population that was counted as members in the religious congregations enumerated in the 1926 Census of Religious Bodies. This chart thus shows the interaction between the percentage of the population that the Census Bureau enumerated and the percentage of the city that claimed a formal membership in a religious group, though it is difficult to disambiguate between these two considerations. As in figure 1, population in 1926 has been estimated from the 1920 and 1930 population censuses using linear interpolation, and it is plotted on a logarithmic scale. In some instances the population must have grown faster at the beginning of the decade than at the end, and so the number of congregation members enumerated is greater than the interpolated population data; in those instances, the percentage of membership has been capped at 100%.

We can look at the data a different way and ask what portion of the population of a city was counted as belonging to a religious group. Here again we can see a huge amount of variation between cities (figure 2). Undoubtedly some of that variation is due to the way the data was counted, including the ways that data was gathered in particular places and the religious groups that the Census Bureau chose to exclude (or simply failed to see). Nevertheless, the range of variation is intriguing. San Francisco, Baltimore, and Pittsburgh were relatively close in size, and for that matter had a history of a significant Catholic population, yet the proportion of their population that was counted as members by the Religious Bodies census was quite different. Earlier we saw that Chicago was home to more denominations than even New York, but now we see that it had a lower proportion of its population counted as belonging to religious bodies.

In the examples provided by both plots we see some of the possibilities for using the cities data. We will pick up some of these threads in future posts, visualizations, and the release of the data itself.