Brad Wright is writing a series of posts on statistics and Christianity. He's currently focusing on variation in divorce rates across types of religious adherents. In the first data post (linked above) he uses General Social Survey data to show that folks who are actively religious are less likely to have ever been divorced than those whom have never been religious, or have stopped. His data are below:
Divorce rates by religious affiliation & attendance. General Social Survey, 2000, 2002, 2004 (N= 5,963)
58% Non-active Black Protestants
54% Non-active Evangelicals
51% No religious beliefs (e.g., atheists, agnostics)
48% Non-active other religions
47% Active Black Protestants
43% Average for all sample members
42% Non-active Mainline Protestants
41% Non-active Catholics
39% Jewish
38% Active other religions
34% Active Evangelicals
32% Active Mainline Protestants
23% Active Catholics
There's a causal time-order nit to pick with this analysis in that the GSS does not allow us to determine when a divorce occurred in the respondent's life course (e.g., before or after a religious conversion)... but the results are compelling none-the-less. These percentages diverge from the received wisdom that Evangelical Christians divorce at the same rate as everyone else (a received wisdom that I've cited in a Soc of Religion paper that was rejected 6 years ago and still sits in my file drawer begging for revision).
Now obviously this analysis is preliminary and tentative. But by posting this on his blog along with references to how he coded the data, Brad has invited folks to examine and comment. I believe that this is essentially what journals were designed to do before the days of publish or perish. They provided a forum for a scholar to work out an idea and let others comment (as they are). This is the way science should operate.
