"Profiles of the Godless": CFI's survey of nonbelievers

Two studies on nonbelievers were recently completed. A few days ago, the Center for Inquiry published “Profiles of the Godless: Results from a Survey of the Nonreligious.” They describe the survey, conducted by the Center with Associate Professor Luke Galen of Grand Valley State University, as “[t]he most comprehensive survey of the nonreligious to date, gathering data from nearly 6,000 respondents.” The results appear in a compact and readable summary in the latest issue of Free Inquiry magazine (click on the PDF link at the site above).

This questionnaire nicely complements the online survey conducted by Tom Arcaro at Atheist Nexus last fall, which he has reported on over the past several months. The Arcaro online survey analyzes the respondents demographically, draws from large samples from several countries, and focuses on the different pressures against coming out as an atheist, though it also touches upon self-identification. The CFI/Galen study, on the other hand, pays somewhat less attention to demographic differences, samples from a majority US population, and focuses on self-identification and psychological characteristics. Both have sampling issues but provide interesting insights. I’ll discuss the CFI survey in this post and the Arcaro poll shortly.

A strength of the CFI study is that it focuses on differences within the nonreligious community itself. The nonreligious segment of the population is growing and increasingly visible, Galen notes; however, “self-described religious believers constitute the vast majority of the American population, and so more attention has been paid by social scientists and survey researchers to distinctions such as religious denomination (say, evangelicals vs. mainline) or political leanings than to characteristics descriptive of a nonreligious orientation.” As CFI’s description puts it: “In general-population surveys, meaningful differences between distinct types of nonbelievers (atheists, humanists, agnostics, etc.) have often been neglected, with the broad category encompassing the nonreligious only included with the implication that they merely represent the ‘absence of religion’.” A more nuanced portrait of nonbelievers is quite welcome, as is a comparison with the religious, especially as nonbelievers are often dismissed with stereotypes, “e.g. that they are “‘angry loners’ or ‘asocial’.”

A pilot study was used to test the survey design prior to the larger survey of the nonreligious. The questionnaire was distributed to all of the subscribers of the Center For Inquiry’s Michigan branch e-mail and group newsletter as well as two good-sized local churches (responses n=333 and n=325 respectively), the latter in order to “provide some range on survey instrument items such as belief in God as well as to support the testing of hypotheses regarding characteristics distinguishing between religious and nonreligious individuals.”

The two samples of course differed in terms of beliefs, with almost all of the churchgoers reporting strong belief and the CFI-list sample exhibiting a range of beliefs, with about half self-described atheists and the rest divided among a variety of other labels. Despite some demographic differences, the two groups turned out to be quite similar in terms of “mental well-being.” One key dimension on which they did differ was “openness to experience,” with the nonreligious being “more intellectually oriented and unconventional.” “Even controlling for the large differences between religious and nonreligious individuals in regard to education, gender, marriage, and child-rearing,” Galen reports, “openness still was the strongest predictor of both lower religious belief and membership in CFI/Michigan as opposed to the churches.” In contrast, religious people were higher on “agreeableness” – “a quality of being amiable or nonconfrontational as opposed to skeptical of others.”

The larger study looked only at nonbelievers, sending an email request to CFI’s international membership, from which snowball samples emerged, with ultimately 5,831respondents, primarily in the US. The researchers were interested in demographic characteristics, psychological characteristics, and variation in self-labeling among the larger group. While Galen acknowledges that this is not a representative sample of nonbelievers but of more organizationally-active individuals, he notes that the demographic characteristics of respondents were consistent with those found in previous research: high education, mostly male, older, higher income (more about which in the later post).

When he talks about differences among nonbelievers, it’s initially a bit confusing. For example, he suggests that “meaningful differences between distinct types of nonbelievers (say, secular humanist vs. atheist) have been neglected.” This “vs.” is misleading in that these are, as Galen is clearly aware, not mutually exclusive or in opposition. Figure 1 at the bottom of page 42 is extremely confusing when you first see it, as what it describes is not explained for two more pages.

Here’s what they did:

Respondents were allowed to endorse multiple religious and philosophical views or labels (such as 'spiritual', 'agnostic', and 'humanistic'), but they were also asked to choose the single term that best described themselves. This self- identification term served as a basis for categorization. Despite the option of selecting among a dozen labels such as ‘deist’ or ‘polytheist’, the overwhelming majority of respondents were divided amongst four preferred labels: 57 percent atheist, 24 percent humanist, 10 percent agnostic, and 2 percent spiritual.

You can read the fuller description in the article, but the study found that while many selected non-atheist labels when allowed several choices, when forced to choose a “master label” people often jettisoned these other terms to master-label themselves as atheists, while the atheist label, chosen in the nonrestrictive phase by 77% of the respondents, was retained as a master-label by 57%:

Contrast the attrition from ‘spiritual’, ‘agnostic,’ and ‘humanist’ when reverting to a single label with the three-quarters who included ‘atheist’ as one of their multiple self-identifications; 57 percent of the latter retained ‘atheist’ as their as sole label. In other words, those respondents who included ‘atheist’ among other labels were most likely to end up retaining it when choosing one self-identification. Thus, atheist appears to be more of a ‘bridge-burning’ term; those who define themselves as atheists are less likely to shed that term or to dilute it with other labels. This indicates that although humanist is one of many hats that nonreligious individuals wear, when push comes to shove most of these individuals are ‘really’ atheists.

Ah! Now Fig. 1 becomes perfectly clear.

Galen continues:

This raises the question: are there individuals who are for all metaphysical intents and purposes either atheists or agnostics but do not label themselves as such? What characteristics distinguish those who otherwise metaphysically agree in unbelief but choose to describe themselves differently?

This is a bit more involved. Using the responses to philosophical questions, they created a category of “de facto atheists/agnostics.” Next they cross-referenced this with the master-labels, creating a category of “atheist/agnostic deniers” – those who were de facto atheists/agnostics but didn’t self-identify as such (using, e.g., “humanist” or “spiritual”).

They then compared the “deniers” and the “admitters,” finding some differences. The most marked, Galen suggests, was age:

Those nonbelievers who chose to self-label as ‘spiritual’ and ‘humanist’ were older (average of fifty-three and fifty-one, respectively) than those choosing ‘agnostic’ and ‘atheist’ (forty-nine and forty-seven, respectively). This would seem to indicate a cohort effect, such that the term atheist is becoming more common despite a shared de facto philosophical outlook with self-labeled humanists.

I’ll discuss the matter further in the later post, but I’ll say now that this raised an eyebrow. Galen doesn’t describe statistical significance, but these ages are not far apart, and it seems a real stretch to me to suggest that they represent any kind of cohort effect. While I do believe the atheist self-identification is expanding (and Arcaro has something to say about this), this doesn't appear sufficient to support that claim.

Galen then goes on to describe psychological differences across the differently self-identified groups: spiritual, agnostic, atheist, humanist. I won’t discuss these findings in detail, but he provides a nice chart on page 44 (again, before the material is covered :)). Most interestingly, they found that “mental well-being” among nonbelievers was the mirror image of that among the religious:

Those nonbelievers most confident in their nonbelief tended to be the most emotionally healthy, relative to the ‘fence sitters’ who reported more negative emotions. Similarly, life satisfaction was lower among the spirituals relative to the other three belief labels. Therefore, having uncertainty regarding one’s religious views appears to be associated with relatively greater emotional instability.

This is a useful study, and I hope to see many more like it in the (near) future. I was left, though, with several questions, some of which were addressed by Arcaro (and some, like multivariate analysis, not – at least yet). I’ll discuss a few of these in the upcoming post and more later on.