Do Religious People Give More, Or Just Lie More?

PZ Myers responds to an article by Jonathan Haidt which starts out great, but segues into a rant against "New Atheists". It is interesting to read both articles as well as comments. One of the many points of contention was this:

Haidt says that "surveys have long shown that religious believers in the United States are happier, healthier, longer-lived, and more generous to charity and to each other than are secular people,"

PZ offers plausible, but speculative, counterpoints before pointing out that such results, even if true, are irrelevant to the issue of whether religion is true. He also correctly points at the inconsistency of justifying Christianity (which is the de facto dominant American religion) on hedonistic and utilitarian grounds.

Still, I was curious about those surveys, particularly because Haidt accused none other than Daniel Dennett, the most cautious of the "New Atheists", of willfully ignoring evidence and of thinking morally under the guise of thinking scientifically:

I have italicized the two sections that show ordinary moral thinking rather than scientific thinking. The first is Dennett's claim not just that there is no evidence, but that there is certainly no evidence, when in fact surveys have shown for decades that religious practice is a strong predictor of charitable giving. Arthur Brooks recently analyzed these data (in Who Really Cares) and concluded that the enormous generosity of religious believers is not just recycled to religious charities.

Thankfully, a commenter critical of PZ provided a link to Brooks' own summary of that study. Another commenter linked to a mild criticism of Brooks on The Volokh Conspiracy. But there is a lot more to criticize!

Brooks' definition of "secular", based entirely on infrequent religious service attendance, includes 26% of the total sample. The most optimistic numbers I've seen for nonreligious in any poll are around 14%, which means that about half (and possibly more) of the so-called "secular" were in fact religious people who rarely or never attend services. Those probably differ significantly from the nonreligious, so which subgroup is driving the results?

ReligiousTolerance.org (RT) has an analysis of false self-reporting of attendance cites sources that estimate that, although about 40% Americans report that they attend services weekly, only 20% actually do. Brooks counts 33% of his sample among the "religious" based on weekly attendance; it would thus appear that a third of those are only included because they lied.

Even more interestingly, the RT page cites Barna Group data that, while 17% report giving 10% or more of their income to their church, only 3% actually do. So we know from previous studies that self-reported giving is exaggerated. The usual explanation is that people say they do what they think they should do, not what they actually do. If the religious have a stronger sense that they should give more than they do, they will exaggerate their giving more.

I think that's enough to invalidate any conclusions from Brooks' study, but, as a general rule, whenever you see a policy paper, don't forget to check the sources and the context. This was published by the Hoover Institution, a conservative think tank. It is true that Hoover is more credible than Heritage and AEI - not everyone at Hoover is a hack - but keep in mind that Dinesh D'Souza is a fellow there. As a further clue to author's bias, note that Brooks approvingly quotes Robert Bork and Irving Kristol. I must say I would have more confidence in honesty and objectivity of a Vatican study on the same topic.

In light of such glaring problems with Brooks' study, I wonder how Haidt can honestly write

These data are complex and perhaps they can be spun the other way, but at the moment it appears that Dennett is wrong in his reading of the literature.

He was addressing Dennett's statement that "[c]ertainly no reliable survey has yet been done that shows" that "as a group atheists and agnostics are [less] respectful of the law, [less] sensitive to the needs of others, or [less] ethical than religious people." Well, I have proven above that Brooks' study is not reliable, so Haidt is the one not thinking scientifically when he uses it as evidence against Dennett.