I don't know if religion makes kids less altruistic. But I do know that this study doesn't do much to answer the question, and The Guardian needs better science reporting.
I’d like to get a little wonky, in my reflections on this paper. But here’s the TL;DR: the conclusions drawn by the study and the media's representation of those conclusions are way broader than the research warrants. A better summary of this paper's core finding would be, "Don't stereotype kids and their likelihood to share stickers with unknown children based on their family's religiosity."
Here’s what this study actually found: kids from non-religious households within a non-random and non-representative sample shared about 4 stickers with an unknown stranger, while kids from religious households from the same non-random and non-representative sample shared about 3 stickers with an unknown stranger. (The gap is actually a bit smaller than that.) The standard deviations were about 2.5 for all groups...which means that within each of these groups, variation between individual kids was a lot wider than the gap between the groups. There were lots of kids from religious households who were a lot more likely to give stickers to random schoolmates than the average. There were also lots of non-religious kids who were more likely to do the same. And the same holds for those who didn't give as many stickers. In other words, the study isn’t actually telling us about “religious” or “non-religious” kids more broadly.
Given that, is the paper actually telling us how altruistic these particular kids were? I don’t think so. For example, we don't know what the kids planned to do with their stickers later. Did they plan to keep them, or maybe share them with siblings or parents? Also, we don’t know how much these stickers were of value to the various children. For kids in lower-income countries (such as the Islamic countries in the study), were these stickers more appealing to them because stickers and toys are harder to come by? So I don’t think this is really a study of “altruism” either.
What's more, the amount of this correspondence that a regression analysis attributes to "religion" is pretty small (r=-.15) and is about equal to differences based on "nationality." (r=.1). (That means that religion and/or nationality correlate with about 15% or 10% of the variation in the data...in other words, the data fits very loosely.) It's possible that the "religion" variable is explained away by the "nationality" variable, and I'd bet that these two variables are highly correlated to each other. These r-values mean you should definitely not be confident in a prediction of “altruism” for any particular kid, based on the religion of their household. What you’ve got is a very fuzzy cloud of data points, with a very small tilt.
Also I put "nationality" in quotes, because the "countries" studied are Chicago (USA), Toronto (Canada), Amman (Jordan), Izmir and Istanbul (Turkey), Cape Town (South Africa), and Guangzhou (China). Obviously, you can't generalize to these nations based on samples from some cities in those nations. And this raises the further question: where did the researchers sample within those cities? They can't even generalize to the populations of the cities, since the experiments took place within particular schools. Would I understand people in Chicago if I picked a school and talked to 100 kids from that school? Even worse, depending on their sampling methods and sizes these results might not even generalize to the _schools_ in which they occurred...let alone the cities or nations! (The methodology section of the paper doesn’t tell us much about their sampling methods, so I’d assume that they weren’t random.) The generalization to "religion" is also extremely broad, in a similar way. Were the South African Christian kids from churches that traditionally embraced apartheid, for example? In Chicago, did they hit a patch of Jehova's Witnesses, or nominal Catholics, or devout Catholics, or kids from JPUSA? These are all very different groups! The upshot is that this study isn't actually designed to draw inferences about "altruism," "nationality" or "religion." It is designed to statistically characterize a non-random, non-represenative sample of kids, based on their likelihood to share stickers with strangers from their schools. When the researchers did that, they found extremely small correlations, and tons of variation.
I think this nicely illustrates some of the weaknesses of this kind of experimental design for social psychology research. It's easy in these cases to operationalize variables in ways that make findings seem far more sweeping than they are. (This also relates to the fact that tons of psychology research doesn't yield replicable results. I think you can see a lot of ways that efforts to replicate this study with different sampling methods could easily yield very different results and interpretations.) In its very limited ‘literature reivew’, the study links to a critique of survey research, which makes sense. There is quite a bit of survey research that has found that religion and giving correspond well to each other. Still, the critiques of this survey research have some merit: surveys of self-reported giving probably are subject to self-reporting effects, and studies that include documentation of actual giving are clearly stronger than studies that only rely on self-reporting. However, experimental designs of the sort in this study have a lot of shortcomings as well, including weird variable operationalization (“altruism”) and overgeneralization. This is why scientists routinely complain about the way individual research studies are sensationalized in the media. Thinking scientifically, this study very clearly falls in the "needs more research" column before I would even consider its findings on "altruism," "religion," and "nationality" in children to be corroborated. The word "experiment" often sounds impressive, and more reliable than things like survey research. But it is really easy for experiments, especially in fields like social psychology, to be wildly misleading. This is a nice example of how that happens.
Having said that, I do think a more basic inference from the paper is probably true: I don’t think religiosity has a terribly strong relationship with sharing among children. At the very least, the relationship is probably a lot weaker than we tend to think. People are naturally moral, and we find that people share and are hospitable all around the world, regardless of religious background or lack thereof. Also, individual variation is often greater than group variation, although human moral psychology and other cognitive biases can easily blind us to this fact. Unfortunately, the way this study was written and has been reported, it is likely to contribute to predictably irrational stereotypes. That’s really too bad, because if the exact same research had been written up and reported differently, it could have helped counter those kinds of stereotypes. In that case, the inferences would have been more warranted by the evidence in the study.
In other words, the study helps us see the point of "The Good Samaritan" more clearly: don't focus on peoples' group membership. Remember that even your enemies (meaning people from groups that are engaged in violent conflict with your group) can be generous. In that, you should let them present a challenging moral example to you, when you encounter opportunities to help strangers in need.
No comments:
Post a Comment