Are children from non-religious homes more altruistic than those from religious homes? (UPDATE: Article retracted)

November 8, 2015 • 11:00 am

UPDATE: According to this link, this paper has been retracted. The reasons given are these:

In our paper, we reported cross-cultural differences in how the religious environment of a child negatively impacted their sharing, their judgments of the actions of others, and how their parents evaluated them. An error in this article, our incorrect inclusion of country of origin as a covariate in many analyses, was pointed out in a correspondence from Shariff, Willard, Muthukrishna, Kramer, and Henrich ( When we reanalyzed these data to correct this error, we found that country of origin, rather than religious affiliation, is the primary predictor of several of the outcomes. While our title finding that increased household religiousness predicts less sharing in children remains significant, we feel it necessary to explicitly correct the scientific record, and we are therefore retracting the article. We apologize to the scientific community for any inconvenience caused.
Thanks to Mark for pointing out the retraction.


I believe I’ve received more emails from readers about this study than about anything else in the last six years. Several dozen readers have directed me to a paper by Jean Decety et al. (Decety is at my university) just published online in Current Biology (reference below; free download).

I think one reason it’s been called to my attention so often is that the authors claim to show something counterintuitive: that children from nonbelieving homes are actually more altruistic than children from religious homes (mostly Christian and Muslim). That goes against the common belief—one often touted by religionists—that religion makes you more altruistic. But because the study plays into what we atheists would like to think, it behooves us to look at it especially carefully. The study has also received lots of press, much of it uncritical.

After having read the paper four times, I think the authors do show that, in a specific laboratory setting, children from homes that are agnostic or atheistic do indeed tend to be more generous on a specific task than children from religious homes. Further, the “religious” children tend to be more punitive and judgmental about such issues, though they’re less generous. Whether this is altruism, and whether it translates into general behavior in society, are issues that remain unresolved.

But I’m getting ahead of myself. Here are the methods and results:

  • The authors estimated the “altruism” of 1170 children from cities in six countries: Chicago (US), Toronto (Canada), Amman (Jordan), Cape Town (South Africa), Izmir and Istanbul (Turkey), and Guanzhou (China). Children also took animated computer tests to estimate their “moral sensivitivy”. The authors also surveyed the children’s parents to see what religion the household adhered to and to estimate “children’s sensitivity to injustice.”
  • The altruism was estimated by having the children play the “Dictator Game“, which in this case was conducted as follows (quotations marks enclose the description from the paper):

Children’s Dictator Game: This tabletop, modified version of the standard dictator game is designed to assess altruism/generosity in children and was run by trained research assistants. In this task, children were shown a set of 30 stickers and told to choose their 10 favorite. They were then told ‘these stickers are yours to keep.’  Children were instructed that the experimenter did not have the time to play this game with all of the children in the school, so not everyone would be able to receive stickers. Children were finally shown a set of envelopes and informed that they could give some of their stickers to another child who would not be able to play this game by putting them in one envelope and they could put the stickers they wanted to keep in the other envelope. Experimenters turned around during the child’s choice and children were instructed to inform the experimenter when they were finished. Altruism was calculated as the number of stickers shared out of 10.”

In other words, altruism was estimated as the number of stickers that kids gave away that they could have kept for themselves. This is presumably “altruism” because the kids are helping others at a cost to themselves, although of course this isn’t biological altruism, in which one incurs a reproductive cost—a loss in the number of genes one passes on. To ensure that kids weren’t just manifesting bias against outgroups and favoritism toward ingroups, the anonymous recipients were known to the subjects to be children from their same school and ethnic group. The authors say that this procedure, unlike methods of previous studies, is an “ecologically valid depiction of everyday mundane interpersonal harm that occur in schools”, and is a good index of “moral sensitivity.”

  • The main result, which has everyone excited, is that children from nonreligious homes gave away more stickers than children from the two other main groups surveyed: individuals from Christian or Muslim homes. Here’s the figure showing a highly significant difference between nonbelievers and religionists (I’ll characterize the children that way as shorthand, as we don’t know what they believed themselves), and no difference between Muslims and Christians.

Altruism Is Negatively Influenced by the Religiosity of Children’ Households Children from non-religious households (n = 323) are more altruistic with an anonymous beneficiary than children from religious families (n = 280 Christians; n = 510 Muslims). Bars represent SEs [standard errors of the mean].
One issue here is that although the differences are significant, the size of the “altruism effect” isn’t specified. In general, even in the supplemental information, the details and metrics of the paper are not well described—at least as far as I could see. How did they quantify “generosity”? Does the graph above show the number of stickers given away? If that’s the case, then the effect, while significant, is not huge: nonbelieving kids gave away, on average, half a sticker more than did Christian kids and one more sticker than did Muslim kids. How meaningful would that be?

  • There are two results that will also appeal to us nonbelievers, but have received less interest. One is that, apparently from the computer games, nonbelieving children are less likely than religious children to judge harm to other people as being “mean” and being “deserving of harsher punishment.” In other words, children from nonbelieving homes are less judgmental—something that, we’d suppose—goes along with the stricter morality and the notion of divine punishment that accompanies religiosity. Here’s the plot of “meanness”, showing that such judgments are lower among nonreligious children than either Christian or Muslim children, but are also lower among Christian than among Muslim children:
Figure 3. Children from Religious Households Judge Interpersonal Harm More Severely Than Children from Non-religious Households Bars represent SEs.
  • Finally, children from nonreligious homes, as estimated by their parents, are less likely to be “empathic and sensitive to the plight of others”. Again, increased altruism goes along with increased judgmentalism. Here’s the plot for “sensitivity to injustice,” showing that, although nonreligious children were the least sensitive, they didn’t differ significantly from Muslim children, while Christian children were significantly more sensitive than both nonreligious and Muslim children:
Figure 4. Parents of Children from Christian Households View Their Children as More Sensitive to Injustices toward Others Bars represent SEs.
What’s the upshot?  Well, the authors did show that nonreligious children—again, this is shorthand for the tenor of their household—are more likely than religious children to give away stickers, though I can’t say the effect is huge. Whether this is a comparison of relative “altruism” is more questionable, as we’ll see in a minute.
Also, even if it is an index of altruism, what we see is short-term behavior of young children in a laboratory setting. It’s not clear whether this behavioral difference would carry over into adulthood, or into more real-world situations where one is actually making a real sacrifice by helping someone else. After all, these children are simply giving away stickers that they got without any effort. Does that mean the same thing as donating to charity money that you’ve earned by the sweat of your brow? Or saving a drowning person? Who knows?
The authors make a fairly strong conclusion about what their study means:
Overall, our findings cast light on the cultural input of religion on prosocial behavior and contradict the common-sense and popular assumption that children from religious households are more altruistic and kind toward others. More generally, they call into question whether religion is vital for moral development, supporting the idea that the secularization of moral discourse will not reduce human kindness—in fact, it will do just the opposite.
Will do just the opposite? Well, we don’t know that. And the notion that religion is vital for moral development has already been refuted by the high morality evinced in nonreligious countries like Sweden and Denmark. Sociologist and paleontologist Gregory Paul, who has published on the negative relationship between religiosity and social well-being among various countries, told me this in an email (reproduced with permission):
Whether or not the paper’s methods and immediate results are reasonably sound, the conclusion the authors come to is overdrawn. The paper finishes by firmly saying that secularization of moral discourse will increase human kindness. That’s a reasonable hypothesis, but it is not yet demonstrated science. The paper should have concluded by stating that the results indicate that fears that secularization will degrade altruism are contradicted by the study, and that secularization MAY instead increase ethical behavior. Had I reviewed the paper I would have insisted on that change.
Paul also made a very good point about whether this behavior is “altruistic,” at least among children from religious homes. It’s not clear whether these children already knew about divine reward and punishment (I suspect many of them did, since they were between ages 5 and 12), but is “altruism” among the faithful really altruism? Paul thinks not:
In any case the whole issue of assessing altruism has been problematic in the psychosociological literature. Studies keep claiming that religious people are more “altruistic.” But true altruism can only occcur if the person expects absolutely no reward or evasion of punishment, and really feels they are making a core sacrifice by committing an act of kindness. But of course the major religions are reward/punishment schemes in which members are calculating they are going to get something from a deity if they do something nice to another, so believers are inherently barred from being altruistic as long as they are seeking heaven or the like. Ergo all papers that conclude that the religious are altruistic are errant.
Finally, what about the increased judgmentalism among religious children? It goes along with the notion that religion promotes stricter judgment of right and wrong, but one could also make the case that nonreligious children’s lower sensitivity to injustice is a fault of atheism. But remember—that sensitivity was judged not in objective tests, but via parental assessment. There’s a strong impetus for biased judgment here.
Finally, there’s a bit of irony in this study that didn’t escape the attention of most readers. Here is the paper’s “acknowledgments” section:
Screen Shot 2015-11-07 at 12.12.35 PM
I’m not so sure that Templeton, dedicated to using science to buttress God and the “spiritual”, would like this result! And this isn’t the first time that Templeton funded a study that gave results inimical to their goals: another was the famous study of cardiac patients showing that intercessory prayer not only had no palpable curative effect, but could have made things worse! Such is science: it tells us what is real, not what we’d like to hear.

h/t: Bob, Matthew Cobb, Julian, Hugh et al.


Decety, J. et al. 2015. The negative association between religiousness and children’s altruism across the world. Current Biology 25:1-5.

50 thoughts on “Are children from non-religious homes more altruistic than those from religious homes? (UPDATE: Article retracted)

  1. Thanks for the careful analysis. I didn’t catch the subtleties you delve into in my reading of the popular press coverage.

    …hmmm…thinking a bit…might it be time for some sort of meta-study to see whether repeated scientific refutations of Templeton fantasies has changed the attitudes of those in the Templeton Foundation, or at least their funding choices? How much backfiring can they withstand before executing the same sort of strategic retreat as BioLogos?


    1. I doubt whether much doubt is cast among the Templeton insiders. I think there strategy must be, they know that some part of the research they sponsor will give the wrnog answer, but it’s easy to ignore those and highlight in their newsletter the ones they like. Throw sh it on the wall and see what sticks. So, methinks they dismiss failures without much thought.

  2. Indeed the conclusion was stated far too strongly. “May have the opposite effect” would have been more credible. Also I wish studies like this would stop using the word “altruism” and replace it with words like generosity, sharing, compassion, etc. As we know, most behaviors that are considered “altruistic” are not technically altruism by the scientific definition.

    And finally I’d like to point out another important difference between these children. The children from the secular homes are the only ones with freedom of religion. And one can not state that too strongly.

    1. Those were pretty much my thoughts when I saw this study. I especially don’t like the way altruism is defined either.

      I also don’t like that the paper calls it a “common-sense assumption” that religious children will be more altruistic. I don’t see why that’s common sense. If children are brought up with the idea that there’s an in group and an out group, I would assume they would be more altruistic to the ins than the outs. I would assume children brought up without those assumptions to be more generally altruistic.

      I thought it was a really interesting paper though, and I’m pleased with how widely it’s being reported. It might give those who assume moral superiority based on their religiosity some food for thought. The attitude towards atheists, though improving, needs help in many countries.

      1. “The attitude towards atheists, though improving, needs help in many countries”

        You can say that again…
        It’s good to keep in mind that even though atheists are pretty widely disrespected in the U.S. and other western quarters, it’s downright hell in others parts of the globe.

  3. I’m so sick of psychology experiments that test just a handful of subjects in just one way, rely on self (or parental) assessments, and draw huge sweeping conclusions from them. Sure, it’s nice that their results favor secularism, but such grandiose extrapolations should set off anyone’s bullshit alarm.

    (And when did we stop having to clearly label all axes in a graph or chart? Since they apparently figured out a way to quantify “Generosity,” “Judgments of meanness,” and “Children’s sensitivity to injustice,” they should at least define their units. (Agree with PCC that the first one was probably simply “numbers of stickers given away.”)

      1. To the criticisms above:

        I think (at least PCC”s quotes) the paper’s conclusion says “…… “supports”….. (the conclusion readers of WEIT like)”. It doesn’t say “shows” or “proves” or anything like that.

        The part about parents’ opinions of their children in this study, I think was made to contrast what is thought about children by their parents and what the data actually seem to show. That parents (and by extension most people —“the common sense view” as the paper states) view religious people (their children) as less judgemental etc. is contrasted with the actual data – limited as it may be.

        1. You have misparsed the statement. “supporting” does not apply to the final claim, which is stated as “in fact”.

  4. I’m so glad you wrote this because I was too lazy to do so. I thought the paper was weak – especially with the value you mention with the stickers but also that there seemed to be a lot of assumptions (that children behaved this way outside of the artificial test environment, that the correlations were accurate).

    I like the Gregory Paul’s better conclusion thatthe results indicate that fears that secularization will degrade altruism are contradicted by the study.

  5. I’d like to add that there was also a predictor that was actually even larger than religiosity: maternal education. Well, the authors called it “socioeconomic status” but the one variable they used was degree of educational attainment the mother had. The more educated the mother, the more prosocial the child. They basically ignored this effect in the write-up, but I find it just as interesting.

    1. Indeed…if nothing else, it suggests that maternal education is correlated with lower rates of childhood religiosity…which one would suspect…and likely is well suspected by those throwing acid in the faces of girls going to schools….


    2. Good catch! That would be consistent with similar research, and with the dysfunctional society/religiosity correlation. Here religion acts much as a symptom and/or mediator for ‘sick’ societies, not a root cause as Jerry notes.

    3. At the risk of responding without reading the paper (no I’m not reading the paper), your point is the most important here. Obviously religiosity wasn’t randomized among individuals. So there will be many confounders that will influence the effect size and any inference from this (a P-value) unless measured. Some of these confounders will be unmeasured, so the effect size estimate will be biased by some unknown amount and will have a (likely, much) larger error. So the best that can be said from this is that at least it doesn’t support the idea that children in religious families are more generous.

  6. Thanks for this, I just found the paper myself!

    So I just read the Science magazine site’s comments on an article over the study so this was a refreshing change. The religious was clutching their pearls, so sure that it was something wrong with the study that they didn’t bother to read it before criticizing sundry bits. (E.g. claims that there was insufficient control, but then the experiment found a significant signal however weak or meaningless. Such claims were mostly that it was cross-cultural and cross-religion, ‘too much mohammedanism’. Et cetera.)

    So nothing new that not the existence of Scandinavia doesn’t already tells us, and a weak study at best. But a nice irritant for the religious and especially Templeton.

  7. Taking the numbers from the article, the effect from that comparison shown in the first figure is pretty moderate:
    Cohen’s d = .35, Pearson’s r = .17

  8. Thank you for dissecting this paper. I saw the paper on quite a lot of news websites, but none of them had a critical look at it.

  9. This is not new. In his 1989 book You Know What They Say…The Truth About Popular Beliefs, author Alfie Kohn tackled the notion that Religious People Are More Altruistic” (AZ999.K64, Page 129). Among other things, he cites a 1950s study of Episcopalians that showed no correlation between church involvements and charity. Another test from the 1970s showed that, when given the opportunity to cheat on a test, the only group in which the majority did not cheat were those who self-identified as atheist.

    Here are Mr. Kohn’s sources:

    Churchgoers’ intolerance: G. W. Allport and J. M. Ross, “Personal
    Religious Orientation and Prejudice,” Journal of Personality & Social Psychology (5:432, 1967).

    Episcopa­lians: C. Y. Glock, B. B. Ringer, and E. R. Babbie, To Comfort and to
    Challenge (Berkeley: University of California Press, 1967, 182-83).

    Col­lege males: R. W. Friedrichs, “Alter versus Ego,” American Sociological
    Review (25:496-508, 1960).

    The 1965 interviews: V. B. Cline and J. M.
    RIChards, Jr., “A Factor-Analytic Study of Religious Belief and Behav­iors,’)PSP (I :577: ! 965).

    Biblical literalists: L. V. Annis, “Emergency Helping and Religious Behavior,” Psychological Reports (39:151-58, 1976).

    Volunteering and cheating: R. E. Smith, G. Wheeler, and E. Diener, “Faith without Works,”Journal of Applied Social Psychology (5:320-30,.1975).

    Neighborhood involvement: S. Georgianna, “Is a Religious Neighborhood a Good Neighborhood?” Humboldt Journal of Social Rela­tions (11: 1-16, 1984).

    Rescuers: S. P. Oliner and P. M. Oliner, The Altruis­tic Personality (New York: Free Press, 1988, 156).

  10. Another issue with this study. They used about 200 children each from the countries: USA, Canada, South Africa, Jordan, Turkey and China, and then they divided them into Christians, Muslims and non-religious (and other, minor, categories).

    One would guess that the Christians came mostly from the US, Canada and South Africa, the Muslims mostly from Jordan and Turkey, and the non-religious mostly from China, plus some from Canada (and a few from the other countries).

    Thus the religiosity correlates hugely with other cultural differences, which surely makes it completely impossible to separate religion from other cultural factors.

    Maybe all this shows is that Chinese children are more generous?

    1. Yes exactly. This is the biggest problem here, by far, to my eye.

      They lump together a bunch of kids from radically different societies: if you’ll excuse the caricature, some of them regard anyone further than a second cousin as basically sub-human, while some wish to extend human rights to turkeys and tuna. Also some of these societies execute people for atheism, while some it’s almost the default.

      Then lo and behold, there’s a correlation between generosity to strangers and religion? Amazing.

    2. I think that geographic (nation, population) effects were statistically controlled in the “ANCOVA.” The different national samples have been statistically adjusted to eliminate bias due to geographic region.
      I would like to have seen results for Buddhism (a religion without a perversely demanding god) and for different sects of Christianity, Judaism, and Muslimism. Of course, like Jerry says, the variances are enormous and there is no basis for generalizing to individual behavior from mean results. In a sample of 40 kids per group, you would not be able to detect between-group differences.
      On the other hand, the results do show, and this is important, that living in a house where god is neither discussed nor praised does not seem make people, at least kids, less generous or more judgmental towards the acts of others. To many people this will be a revelation.

  11. that presumes that such a study is possible to even craft and that the conductors are actually moral

    which is a dubious assertion

    morality is a cultural construct and given the history of religion, there is no morality there

  12. the anonymous recipients were known to the subjects to be children from their same school and ethnic group


    So if I understand this right, the white kids were told, “Put some stickers in this envelope if you want to share them with the other white kids in your school” – and similar for other racial categories?

    In which case any child with a functional moral system would have refused to distribute any stickers in a racist manner. At least, I hope any child of mine would. Far better to hang on to the stickers and share them out oneself after the fact.

    1. No, it doesn’t say that. It doesn’t say that they mentioned race to the kids. It just says that they picked racially homogeneous classes.

      1. Hmm, OK, the supplement says they recruited kids from “ethnically / socially homogeneous schools”. That does raise the question of how they managed to find such schools in multicultural regions like Chicago and Cape Town, and how representative the kids in such schools are of the general population. But at least (I hope) it’s not as bad as it originally looked.

  13. This finding is consistent with studies that find non-religious people donate a larger fraction of their incomes to charity than do the religious (after subtracting out donations by the religious to their churches, most of which is self-serving rather than charitable.)

  14. To me it makes intuitive sense – when people are accepting of a stringent natural order, it’s easy to be more accepting of injustice and inequity.

    On a practical level, this doesn’t say much about how those behaviours translate into adulthood, nor how they work on an individual level.

    At best, such findings are good for shattering the myth one needs religion to be moral, but there was already plenty of other evidence of that anyway. Perhaps it’s taking away the myth that non-religious upbringing is a parasite on religious morality by showing that note religion isn’t correlated with better behaviour.

  15. A very interesting and balanced analysis. The other problem of the sample is they it uses categories that are so broad they become unhelpful. A child raised in a mainline Christian home will have little in common in terms religious ethics to a child raised in a strict fundamentalist home. To lump them together as “Christian” is equivalent to lumping Bernie Sanders and Ted Cruz into a category that seeks to know what men as a whole think on a particular subject. Religion is a notoriously unquantifiable category and studies such as this should take care to note that.

    1. | The other problem of the sample is they it uses categories that are so broad they become unhelpful.

      Good point. I was also wondering how old the children were.

      When I first read about the results of the study yesterday, I first was pleased and then became skeptical since it seemed like a large conclusion to make.

  16. I’ve always had an issue with studies like this that I have never seen addressed. I suspect that the relevant variable in behavior is usually not whether people are religious v. non-religious but whether people were raised with some sort of moral framework or were not. If that is true, then since, as a matter of sociological fact, most people raised with a moral framework happen to have been raised in a religious moral framework, the religious should do considerably better than the non-religious because the non-religious class should include a bunch of moral slackers who don’t believe in much of anything and have no real moral framework. So even if the non-religious folks with non-religious moral frameworks are as good as or better than religious folks with religious moral frameworks, shouldn’t the slackers with neither religious nor secular moral frameworks drag down the numbers for the non-religious?

    1. If that is true, then since, as a matter of sociological fact, most people raised with a moral framework happen to have been raised in a religious moral framework…

      Have you ever looked at what passes for a religious “moral framework?” Let’s take Catholicism: homophobia, tribalism, idol worship, and misogyny for starters. IMO a great deal of most “religious moral frameworks” is demonstrably immoral.

  17. I’ve misplaced the source, but I read somewhere that one important variable wasn’t controlled for: number of siblings. The idea is that religious families have more kids, and the kids will have learned to share first their siblings, so they’ll naturally not share with not siblings, even though, proportionally, they might be sharing as much. It seems like a possibility to me.

  18. Not a big surprise. I suspected that the article was a little over-exuberant and premature in its conclusions.

Leave a Reply