In a new study, the only group participants found as untrustworthy as nonbelievers were rapists.
One of the main issues in studies of behavior and stereotype is the desire to avoid telegraphing to the subject what he or she is “supposed” to do. But in explicitly selecting “religious” people from the USA via an internet survey company, Gervais – a foreigner – may have been unconsciously prompting them to act more in accordance with common stereotypes of religious Americans.
Furthermore, the way in which the questions were asked is definitely not as it was explained on Jezebel. Jezebel would have the reader believe that the test subjects, for no particular reason, homed in on atheism to explain immoral or selfish behavior. But that isn’t exactly what happened; read the study setup and you may find that the participants were, well, set up:
Across subjects, we manipulated the target groups to which the man might belong by asking participants whether they thought it more probable that the man was a teacher or a teacher and (a) a Christian, (b) a Muslim, (c) a rapist, and (d) an atheist. In this way, we evaluated the degree to which people find an untrustworthy description to be representative of atheists, relative to a majority religious ingroup (Christians), a religious outgroup (Muslims), and an unambiguously distrusted group (rapists).
Let us presume that the average person, knowing both positive and negative examples of Christians, is willing to consider being religious a positive, negative, or neutral matter. Being asked how “strongly religious” they are by the very terms of the study, Christians, who consider being Christian at least neutral, are primed to consider being atheist not a neutral matter, but a negative matter: it clashes with their avowed identity, and they feel an obligation to act out that identity. (Studies have shown that reminding people of cultural, ethnic, or other stereotypes produces a similar reversion to type, even if that stereotype is seen as negative.)
Further, we already have one positive piece of information – the character is a teacher, generally a fairly selfless role in society. There must be some negative element to counterbalance this given positive, or the result is essentially, “this man is a jerk for no explainable reason.” It’s sort of a game of “which of these is not like the other?” Being a rapist is categorically not like any of the others. Being an atheist is not like being a religious believer, and so also not like the others (or at least, more dissimilar than being a Christian and being a Muslim.) The pressure on the respondent is to pick something to account for the behavior. It’s at least equally valid to look at this study and say that the respondents are surprisingly free of bias against Muslims as it is to say that they are astonishingly biased against atheists.
As an aside, I question whether rapists truly constitute a proper “unambiguously distrusted group:” while we all consciously say we don’t accept rape and abhor it, as a population this is simply not true. Feminist studies assert that men often have mental blind spots that cause them to effectively downplay the likelihood or severity of it. According to the more cynical of these sociological studies, this is what enables men to bond with other men and work together on matters that do not touch their personal histories. In other words, it’s more that men are wired to mute their outrage over rape than that America morally equates atheism with violent crime. This is a hidden cognitive bias that, if present but not accounted for, could easily skew interpretation of the results.
When deciding whether it is more probable that Linda is a bank teller, or that Linda is a bank teller and a feminist, most participants incorrectly choose the latter option—that is, they commit the conjunction fallacy— because they heuristically judge that the description sounds representative of a feminist, even though logic dictates this option is necessarily less probable. People only commit the conjunction fallacy when the target’s description (single, outspoken, and liberal) is deemed representative of the target’s potential group membership (feminist).
Well, certainly. According to models of how our minds make sense of the world, we wish to tell tales that account for the given facts. If making Linda a feminist makes the whole story more plausible and easier to remember, then it is unsurprising that we are willing to believe it. If “atheist” implies “not dedicated to a fixed moral code” (falsely, in many cases, but widely believed) then it may be invoked to make a story of opportunistic actions easier to rationalize and believe. Also, the conjunction fallacy is easy to commit where there is no penalty for guessing wrong; I would be interested to see an economic study where the participants are asked if they are willing to bet $20 on the bank teller being a feminist with no supporting evidence, or similarly on the selfish man being an atheist with no supporting evidence.
A far better study design would be to take reactions from thousands of people, and only mine the data later for the relevant reactions. This avoids the issue of prompting people to behave like stereotypical fundamentalist Christians. It would also be best to even out the response options so that rapist and atheist are not two options that might be perceived as too strongly dissimilar to the other options, or rather, to eliminate options which are too similar. If testing distrust, why use “rapist” and not “used car salesman” or “telemarketer”? Finer gradations would more precisely pinpoint the level of bias the public has against atheists without necessarily throwing everything to the apparent level of violent crime.