David Myers

Head-Scratching Paradoxical Findings: Aggregate versus Individual Data Can Tell Very Different Stories

Blog Post created by David Myers Expert on Jul 18, 2016

Originally posted on June 4, 2015.


One curiosity of recent psychological science is what I’ve called the “religious engagement paradox”: The association between religious engagement and human flourishing is negative across places and positive across individuals. For example, in the most religious U.S. states people die sooner, commit more crime, divorce more, smoke more, and report lower emotional well-being than in the least religious states. Yet more religiously engaged individuals live longer, commit less crime, divorce less, smoke less, and are happier. (Don’t believe it? See here.)


Princeton economist Angus Deaton and psychologist Arthur Stone (2013) share my puzzlement (here): “Why might there be this sharp contradiction between religious people being happy and healthy, and religious places being anything but?” (One possible answer, as Ed Diener, Louis Tay, and I suggested, lies in the more impoverished life circumstances of people in highly religious countries and states.)


As I noted earlier, there also is a parallel “wealth and politics paradox”: In the U.S., low income states and high income individuals more often vote Republican:






Now we have a report of yet another paradox: In Europe, “More liberal countries and more conservative individuals have higher levels of SWB [subjective well-being].”


And another: People in highly religious states do more Google searches for sexually explicit content such as “gay sex,” as I was able to replicate using Google archives. So I couldn’t resist asking the lead researcher, Cara MacInnis at the University of Toronto, if it might nevertheless also be true that more religious individuals do less online searching for sexual content. Stay tuned, but MacInnis tells me that her latest data (paper forthcoming) do, indeed, seem to fit the religious engagement paradox pattern.


The repeated lesson: how we ask the question (comparing aggregate or individual data) can sharply change the answer. So beware: partisans on both sides can pick their data to make their point.