Summary of findings

(Post seven of a series on Christian divorce rates)

In the past several posts, I have presented data from various surveys regarding divorce rates among Christians compared to others. Here I would like to summarize these findings plus offer some qualifications about how much they tell us.

1) Active Christians vs. non-active Christians. In each of the studies, those Christians, whether Protestant or Catholic, who attended church about once a week or more had substantially lower divorce rates than those Christians who did attended church less frequently.

2) Christians vs. members of other faiths. It's not clear that there are meaningful differences in divorce rates between Christians and members of other religions. Of all the issues analyzed here, this is the most ambiguous because of the inconsistent nature of the "other religion" data. The composition of "other religions" varied a lot across sample, and a clearer statement on this comparison would need to examine other religions individually.

3) Christians vs. people of no faith. Christians as a group, but especially those who were active, had substantially lower divorce rates than individuals professing no religious belief (e.g., atheism, agnosticism). Sometimes as much as half the rate.

4) Black versus white Christians. Black Protestants, or, in the case of the GSS data, individuals attending predominately black denominations, have much higher divorce rates than white Protestants or members of other religions. It's unclear if this is a simple race effect (i.e., blacks vs. whites regardless of religion) or a race by religion effect (the effect of religion varies by race).

5) Protestants versus Catholics. In some data sets active Catholics had somewhat lower divorce rates than protestants, in other data sets they had comparable rates. No consistent difference emerged across data sets.

6) Frequency of attendance. Divorce rates for Christians drop considerably when going from period attendance to weekly attendance. I.e., the functional form does not appear to be linear, rather it's a threshold effect at about one week.

Thoughts?

Next: Qualification of findings