Atheism might be more common than assumed...but it's complicated

The study, "How many atheists are there?," appears in the journal Social Psychological and Personality Science.

Based on the results, "Just as there are, and have historically been, closeted gay men and lesbians out there, there are probably lots and lots of closeted atheists out there, who don't even identify themselves as such in anonymous polls," says lead author Will Gervais (University of Kentucky).

The authors surveyed 4000 people across two studies of 2000 people each in nationally representative samples. They used an indirect sampling method, which gives participants a list of things to look at and then record how many of the statements, but not which specific statements, are true for them. This technique lets researchers infer overall prevalence of things people might not want to admit, and has previously been used to estimate the prevalence of various undesirable or criminal behaviors.

"Within our sample, one in three atheists in our online survey did not disclose their lack of belief, highlighting the level of stigma associated with lack of belief," says coauthor Mazine Najle (University of Kentucky).

Gervais says he was "surprised by just how far the indirect measurement diverged from established polls of religious disbelief. Our best estimate is more than double what Gallup telephone polls estimate." Gervais and Najle expected a much more modest gap between self-reports and indirect measurements.

The findings also have potential impact for science. For decades, researchers have been developing and testing theories about how religion works, but they focus on belief being prevalent and atheism a rare occurrence.

"If it turns out that atheism isn't all that rare, it might challenge a lot of prevailing theories," says Gervais, "Basically, it'd mean that those of us who study the natural foundations of religion need to question our bedrock theories and assumptions."

It's important to note that their results weren't crystal clear and perfect. Their estimates are "pretty noisy," says Gervais, "It's 26% as a best guess, but it could be quite a bit lower...or higher. Could be 20%, could be 35%. We also had some findings that made us question the validity of the very task we employed."

The authors appreciate that even with the "noise" their research was accepted for publication.

"Social psychology has been taking a beating lately for producing results that are too good to be true, and it's a real feather in the cap for this journal that they didn't pressure us at all to put out the "perfect" paper, rather than just transparently communicate what we found, including the good, the bad, and the ugly," summarizes Gervais.