The problem with studies on the social effects of religion…

One of the many newspaper columns published over the weekend was this one, in The Miami Herald, on the alleged beneficial effects of religion. Most of it was drawn from the work of Mike McCullough, a psychologist at the University of Miami.

McCullough’s research suggests that religious people of all faiths, by sizable margins, do better in school, live longer, have more satisfying marriages and are generally happier than their nonbelieving peers.

Yes well, that’s all true enough, but does it justify the claim that religions causes all these effects? In fact, the evidence is surprisingly weak. Take, for example, McCullough’s recent study, quoted in the article:

In the Journal of Drug Issues, he reported that in neighborhoods plagued by alcoholism, church attendance helps more than Alcoholics Anonymous.

In fact, what they showed was that women who say that they go to Church once a week (or more) also say that they rarely binge drink. This doesn’t really prove that Church attendance reduces binge drinking.

But this study does very nicely exemplify all of the problems that bedevil research into the social effects of religion (all of which McCullough acknowledges):

What people say they do and what they actually do are different things.
Most studies into the effects of religion rely on self-reported behaviour. The subjects fill in a questionnaire reporting how often they go to church, and (in this case) how often they get drunk.

But we know that how people want to see themselves strongly influences what they put in these sorts of surveys. How many people are going to put down that they both go to Church regularly and get drunk regularly, when they know that the two together are highly socially unacceptable.

Correlation is not causation.
Religious people are different to non-religious. Hard drinkers are likely to avoid going to Church for a variety of reasons. Almost all studies are purely cross-sectional. That is, they look at people at a single point in time. But that really proves nothing, especially if you don’t control for personality type.

Even longitudinal studies are suspect. Suppose it works like this. A drunkard decides to turn her life around. They start going to go to the church (because, in the US, that’s the premier source of support networks), and with the help of their new friends, they start to turn their life around. Is this really a story of religion causing temperance?

There’s no such thing as ‘religion’.
OK, this will probably come as a surprise to many. But religion is a nebulous concept, and the reason is that it’s actually a label applied to a bunch of different things – most notably participating in ritual activities, and a variety of supernatural beliefs.

Mixing the two up should be verboten. And yet that is exactly what McCullough does in this study. He begins by theorizing why religion should reduce binge drinking – because it contravenes religious beliefs in a variety of ways. And then goes on to look at how church attendance, not religious belief, is linked to less drinking. There’s lots of reasons why people go to church – and only some of them have anything to do with religious beliefs.

So is there any practical way of untangling this sticky mess? I think there is, and there is a limited amount of data out there that’s highly suggestive.

Firstly, if you want to make the arguments that McCullough is making, that adopting a religion helps people who would otherwise get into trouble, then you really need to do an interventional study.

This means getting a group of people and giving half of them religious instruction, and the other half some secular alternative – like engagement in a support group. Amazingly, given the amount of money spent on religion, and the widespread belief that it is effective, these sorts of studies are almost never done!

However, there are two recent examples. A study that looked to see whether virginity pledges were effective. And a study that looked at whether spiritual guidance helps drug addicts. Neither showed any benefit.

Secondly, you could do research in populations where the religious are in the minority. That would help sort out whether it’s just a socialization effect. In other words, people who want to conform and have the willpower to participate in wider society will turn to religious groups in a religious society, and non-religious groups in a secular one.

This also is an under-researched area (there just isn’t the interest in non-religious countries). But one recent study, in Scottish teens who were mostly non religious, found that religiosity did not affect sexual behaviour.

And finally, you can get a reality check on whether encouraging religion is really a useful way to focus society’s energy by looking at non-religious countries. Across a wide range of outcomes, less religious countries are happier, healthier and more secure than religious ones.

If you want to make the world a better place, then worrying about religion is not the place to start.

_____________________________________________________________________________________

Terrence D. Hill, & Michael E. McCullough (2008). Religious Involvement and the Intoxication Trajectories of Low Income Urban Women Journal of Drug Issues, 38 (3)