Nonreligious children aren’t more generous after all

New analysis sheds a different light on a widely publicized study.

A study finding that children with a Christian or Muslim upbringing were less altruistic than their non-religious counterparts made waves in 2015. However, new analysis of the data reveals that this is not actually the case. In the study, developmental psychologists looked at five- to twelve-year-olds in Canada, China, Jordan, Turkey, South Africa, and the United States. The children were given stickers and provided the opportunity to share them with peers who were not present. The number of stickers shared provided a quantifiable indicator of each child’s altruism.

After interpreting the resulting data, the study’s authors concluded that children from religious families were less generous. However, new analysis shows that the original study failed to adequately control for the children’s nationality. After correcting this error, the authors of a newly published paper found no relationship between religiosity and generosity. We speak with first author Azim Shariff to learn more.

ResearchGate: What first led you to reexamine this data?

Azim Shariff: I had received several inquiries from journalists asking me to comment on the study since I’ve also done quite a bit of research on the topic of religion and prosocial behavior. After a couple of these inquiries, my colleagues and I started looking more closely at the original paper’s analysis section and noticed some oddities and omissions. So I asked the original author about it. He eventually sent me the original dataset so that we could run our reanalysis.

RG: What were you looking for?

Shariff: The key question is whether the differences in children’s altruism were due to their parents' religion or to some other associated variable. Imagine you found that the more fiber someone ate, the more risk-averse they were. Is there a meaningful connection here? Does eating a lot of oat bran and figs make people less risky? Or is the relationship more likely explained away be some other variable—like age? Maybe people start eating more fiber as they age, and maybe they also engage in fewer risky behaviors as they get older. Once you account for the age explanation, there may be no remaining relationship between fiber and risk-taking.

Now, the original authors of the religion article collected data from six quite different countries. Places like Jordan and Turkey had essentially entirely Muslim families in the study. On the other hand, places like China were predominantly non-religious. So, were the altruism differences due to religious differences or to differences between country? Or other unmeasured variables related to country, such as number of siblings? Since other studies had found that kids from different countries behave differently on the exact altruism task that was used in this paper, it was important to control for country. The original authors had intended to, but unfortunately made a basic statistical error in doing so. So we just ran their intended analyses with that error fixed.

RG: What did you find?

Shariff: This corrected analysis yielded results quite different from the ones the authors highlighted. Once you properly control for the family’s country of origin, the clear differences disappear. So, it wasn’t the case that the children of atheistic parents were any more generous than the children of religious parents, nor were those children any less punitive than the children of religious parents. Finally, the religious parents did not report their children being any more empathic or sensitive to injustice. Most of these findings were instead attributable to the country of origin, with children in Turkey and South Africa being less generous.

We did find that there was a small negative correlation between religiosity and generosity, but that was mainly about the children of the very religious being slightly less generous than those of the moderately religious, rather than a difference between a religious and non-religious families.

RG: Did you find other factors that do correlate with children’s altruism?

Shariff:  Yes, in addition to children in Turkey and South Africa being a bit less generous than those from the other countries, older children were more generous than younger ones. This is typical of these kinds of studies.

RG: Do you have recommendations for ways parents can raise more altruistic children?

Shariff:  Oh, no, I probably shouldn’t do that. What was most notable about these results was ultimately, what they don’t show. So if I were to hazard something, it would be to tell parents not to worry about needing to raise your kid religiously or non-religiously for them to be altruistic. At least on this particular sticker-sharing task, it isn’t being religious or being non-religious that is determining whether or not they share.

RG: Why do you think there was such public interest in the original study when it was released?

Shariff:  Part of it is that it was a “man bites dog” story. People have been told for ages that being religious is necessary for being moral. In fact, a poll from 2014 showed that 53 percent of Americans think that believing in God is a necessary precondition for being a good person. This finding ran counter to that. That conventional wisdom is also something that is really frustrating to non-religious people who go about their day being good people, and nonetheless get maligned as being immoral heathens. This frustration is compounded when they see so many religious people acting immorally—showing not only that religiosity isn’t a necessary condition for morality, it’s not a sufficient one either. As a result, there is a nice sense of validation—and (dare I say) smug satisfaction—when seeing scientific support for the contrary position: that the non-religious may actually be the more moral ones. People are attracted to news that makes them feel good, and since the non-religious tend to be overrepresented among science consumers, Decety et al.’s paper got a lot of play.

At the same time, there is a tendency to be less critical of information that one wants to be true, which may have contributed to a lot of non-religious people being a bit less skeptical of this particular piece of evidence than they otherwise would have been. Of course, it was also likely met with elevated skepticism from the religious. Being aware of and suppressing one’s own biases is an extremely difficult task—if it’s fully possible at all. Luckily, part of the scientific process is to distribute this task over a community of scholars who can check each others' work.

Featured image courtesy of Ben Grey.