3.05.2018

Deconstructing Viral Religion Graphics



One of the first books I read as a graduate student was Edward R. Tufte’s The Visual Display of Quantitative Information (from 1982; i.e., the social scientist’s coffee table book). Tufte argues that graphics reveal data and therefore should, “avoid distorting what the data have to say” (p. 13). In one section of this classic Tufte writes, “Each part of a graphic generates visual expectations about its other parts and, in the economy of graphical perception, these expectations often determine what the eye sees. Deception results from the incorrect extrapolation of visual expectations generated at one place on the graphic to other places” (p. 60).

In light of this advice, I think there are two recent graphics floating around the internet and social media could use a bit of deconstruction…

The Strangely Sliced Pie
A September 2017 report, “America’s Changing Religious Identity” from Public Religion Research Institute (PRRI) presents data that is said to reveal:

White evangelical Protestants, the single largest religious tradition, make up less than one in five (17%) Americans today. Compared to ten years ago, significantly fewer Americans identify as white mainline Protestant (13%) or white Catholic (11%). Mormons comprise two percent of the population. Fifteen percent of Americans are nonwhite Protestants, including black Protestants (8%), Hispanic Protestants (4%), and Asian, mixed-race, and other race Protestants (3%). Seven percent of the public is Hispanic Catholic. Non-Christian religious groups constitute less than one in ten Americans. Muslims, Buddhists, and Hindus are each roughly one percent of the population. Jewish Americans account for two percent of the public. No religious group is larger than those who are unaffiliated from religion [emphasis added]. Nearly one in four (24%) Americans are now religiously unaffiliated.


No religious group? Really? White Catholics? Asian, mixed-race, and other race Protestants? See what is done here? I could perhaps see a legitimate historical argument to specifically measure members of Black Protestant denominations (e.g., African Methodist Episcopal Church, National Baptist Convention, USA) but people who identify their race as Black of African American and their religion as Protestant or Christian are not necessarily members of any of these churches. It is true that pollsters often show results within religions for sub-groups (e.g., gender, generation, race and ethnicity, income) but they don't define these sub-groups as separate religious groups themselves (Look at PRRI's American Values Atlas and you'll see more of this odd segmentation).

The racial and ethnic division of Catholicism in the graphic is especially bizarre. Catholicism is a global faith and that diversity is represented well in the United States. CARA does in-pew surveys in parishes around the country. We’ve needed to translate these into 20 languages. In some cases we’ve distributed surveys in three or four languages at one parish. Pope Francis is the pope to “white Catholics,” “Hispanic Catholics,” and  “other nonwhite Catholics.” They all share the Mass, sacraments, Catechism, Canon Law, etc… Hispanic Catholics are not a different “religious group” than White Catholics (…to be a real stickler for details, in the 2016 General Social Survey, 55% of Hispanic or Latino Catholics self-identified their race as white. Hispanic or Latino is ethnicity).

Yet, even here there is an interesting religious distinction to make. I’ll let you in on a secret that isn’t often discussed or noticed in religion research. Some of the unaffiliated are likely….(wait for it, drum roll…) evangelical Christians! Many surveys ask what religion respondents affiliate with followed by a question about whether or not they are an evangelical or born again Christian. This second question was a quick rule of thumb used to classify mainline and evangelical Protestants or Christians (for more see “Measuring Evangelicalism: Consequences of Different Operationalization Strategies” by Conrad Hackett and D. Michael Lindsay. I don’t believe PRRI surveys ask the unaffiliated a question about evangelical identity). The figure below is from the General Social Survey (GSS) series. It shows the percentage of those who don’t have a religious affiliation who also say “yes” when asked, “Would you say you have been ‘born again’ or have had a ‘born again’ experience—that is, a turning point in your life when you committed yourself to Christ?”


Rather than the unaffiliated all being some universal “none” stereotype, some are perhaps more appropriately described as evangelical “dones” (For more about the concept of “dones” see Josh Packard and Ashleigh Hope’s Church Refugees, 2015). These are people who may have left behind the brick and mortar of traditional organized religions for more personal and local expressions of their Christianity. They don’t think of themselves as members of a religion as much as they might see themselves as believers within a specific local nondenominational church and/or community.

Then again, perhaps these respondents are just misunderstanding or incorrectly answering the evangelical question (…the GSS question itself could use some improvement)? We can compare the unaffiliated with the “unaffiliated evangelicals” to see if there are any differences on religion questions. As shown in the figure below, 94% of the unaffiliated who are not evangelicals attend religious services once a year or less often. By comparison, 67% of the unaffiliated evangelicals attend this often. While they do not match the attendance levels of the typical U.S. adult, there is clearly something different going on here.


Behavior is a higher bar of comparison than belief. Perhaps the more interesting differences between the unaffiliated and the unaffiliated evangelical respondents are to be found in their beliefs about the Bible or God. The unaffiliated evangelicals indicate beliefs about the Bible that are similar to the typical U.S. adult with only 28% believing this is an “ancient book of fables” (compared to 22% of U.S. adults). About two-thirds of the unaffiliated who are not evangelicals believes the Bible is this “book of fables.”


Seventy-four percent of unaffiliated evangelicals believe in God and an additional 15% believe in a higher power. Just 7% are agnostic and 4% atheist. By comparison only 27% of the unaffiliated who are not evangelicals believe in God (33% believe in a higher power, 40% are atheist or agnostic). Unaffiliated evangelicals look very similar to the typical U.S. adult in terms of belief in God. The unaffiliated who are not evangelicals do not look similar at all in this regard.


While some researchers and the media more generally have taken to equating the unaffiliated or “nones” as a distinct group lacking in religiosity there are more interesting realities in the data. I’m not sure if this sub-group(s) is really the “slice” of the religion pie that it is portrayed to be by the PRRI graphic. It is by no means larger than any “religious group” in the United States.

If you wanted to make a pie out of race, ethnicity, religion, and evangelical identity using the GSS it would perhaps most accurately look like what is shown below (I’ve restrained myself from trying to subdivide the small slices for those identifying as Jewish, Muslim, Hindu, Buddhist, or some other religion by race and ethnicity… I will show these distinctions in another figure that follows where these can be more fully seen):


“No religious group is larger than those who are” white evangelicals at 21%? But again, I’m just not sure “white evangelicals” is a distinct religious group. Despite the apparent lack of race and ethnicity for people without a religious affiliation in the original PRRI graph, you can see there is diversity here. There are white nones, black nones, Hispanic nones, and other race and ethnicity nones. The figure below shows this diversity in a comparative context:


It seems to me the most direct measure of religion is the figure below, which accounts for evangelical identity among the unaffiliated (as in the GSS figures above):


The original pie chart is just one representation of PRRI’s use of race and ethnicity to subdivide some religious groups. They also have another popular graphic in social media that adds generation into the mix, as shown below:


Removing race and ethnicity for the measurement of religious groups and accounting for the unaffiliated who self-identify as evangelicals, perhaps a more accurate representation of religion by generation appears as such:


The Seas of Disbelief?
Another popular graphic on social media is “Faithland.” This graphic was created by Alex Egoshin, an environmental scientist, using religious “adherent” numbers from the 2010 U.S. Religion Census. These data are not actually from a census in any traditional sense. Researchers asked congregations from 236 faith groups in the United States about their members and worshipers and then derive an estimate of adherents to different religious groups. The Religion Census defines an “adherent” as such: “The adherent figure is meant to be the most complete count of people affiliated with a congregation, and the most comparable count of people across all participating groups. Adherents may include all those with an affiliation to a congregation (children, members, and attendees who are not members).” Practically speaking, a Catholic adherent “is roughly equivalent to those who are known in some way to each parish or mission.” These adherents are tallied and subtracted from the U.S. Census population to also produce an estimate of the “unclaimed.”

The Faithland graphic shows land where adherents make up 50% or more of the Census population. The water areas are places where 51% or more of the population in an area are “unclaimed” by congregations.


The graphic displays the data in a compelling manner. At the same time, there are two things that I think the general public, who are sharing and viewing the map, are generally unaware of. Knowing these might alter their interpretations.

First, adherents is not equivalent to “religious people.” A majority of self-identified Catholic adults, 57%, attends religious services only a few times a year or less often. None of these Catholics are likely to be considered “adherents” by the Religion Census. The same can be said for 25% of evangelical Christians, 65% of all other Christians, and 76% of people of other faiths. The graphic does not display the components of religiosity that are independent of a church or temple’s membership and/or attendance numbers. In other words, those seas are not necessarily full of agnostics and atheists. There are a lot of people there who have a religious affiliations but who are not “claimed” by any congregation. There is a lot more of this map that should be dry land if one were simply measuring “faith.”

Second, I’m not sure how areas where no one lives are treated in the mapping. The size of these areas in the country are larger than one may think (see below... more commentary here). The Religion Census data are reported at the county-level. However, the map is not displaying counties and some sort of smoothing is being used that viewers are likely unaware of.