Agree2Disagree
Source
In this article, we have analyzed the effect of religion on crime as reported in previous empirical studies.
We examined data from 60 studies, and we found that religion had a statisitically significant, moderately sized effect on crime
of about r= -.12. Since Hirschi and Stark's (1969) finding of religious nondeterrance, many sociologists
have questioned whether religion has any effect on crime. Our findings give confidence that religion does
indeed have some deterrent effect.
Now I've provided my case study. Can you provide yours?
That study has a pile of problems I'd have to look at more to take seriously:
* It's a meta-analysis of other studies
* The collection of studies have different standards of what qualifies as religious and what doesn't
* Studies were focusing on vastly different things ... for example some studies focused on alcohol use by minors and marijuana whilst others discuss
murder and robbery. What if all the people who didn't rob or murder were smoking pot?
* Some of the studies are significantly larger than others. Larger sample sizes mean a smaller effect. To quote Wells and Rankin: "the most
substantial correlations are produced by the smallest, least reliable, and least representative studies". The author of the study states that the
negative effect of the sample size might be due to the file drawer problem. 'Journal editors ... are simply more likely to believe and publish null
findings derived from larger samples'. I'm not sure that makes sense ... because if anything the file drawer problem would imply there are more null
finding studies that haven't been published, it wouldn't support the study.
There is also very little discussion of cultural factors in relation to sample size. The author makes a point that white people are less effected by
religion and tries to tie this to African American church practice but I'm not sure that applies here. Given that white people are 70% ish of the
population of America, you would expect to find a wider variation in demographics. African Americans are disproportionately poor in the United States,
and therefore it doesn't take a big leap to point out that more African Americans are likely to be actively attending and having ties to the local
church. More African Americans attending church = the 'deterrence' factor is higher but only because more African Americans from various
demographics are attending. The author acknowledges this but tries to tie it into moral teaching when in actual fact given most people aren't
criminals a generic rise in church attendance would seem significant in a small population.
It's not the worst study ever I don't think ... it's honest about a lot of its problems, but I think it needs a fair bit of academic
interpretation. Studies for publication do have a tendency to play down their problems (especially meta-analysis) and this one seems to fall into that
category. I especially find that explaining away of large sample sizes having close to null effects as 'hand waving'.
Peer review for a lot of these journals is really just two academics doing a blind review.
Especially in social science areas they can be a
little lax on empirical evidence; really they're just confirming that the author of the journal has something relevant to add to the discussion ...
they're not endorsing the findings 100%. The author actually acknowledges eight articles that disagree with their findings soooooo ...
Sorry for the rambling, I just find persons put too much faith in these types of articles and, at the end of the day, the author points out that the
'type' of religion didn't change their findings ... so it may be there needs to be a secular equivalent service made available to the community at
large for gatherings and the discussing of moral questions; that's if all the problems I mentioned in this study aren't enough to deem it
inconsequential.
I personally wouldn't have the bones to put pot smoking and under age drinking in the same category as robbing and murder in a meta study about
morality and crime frankly.