Friday, November 05, 2010

The relationship between religiosity and infidelity differs by religion

Using General Social Survey data, I calculated the relationship between attending religious services and cheating on your spouse for people of various religions. (The sample is whites only, except in the case of Hindus).  

Logistic Regression Coefficients

Protestant -.12*
Catholic -.15*
Jewish -.21*
Buddhist .22
Hindu .26
Muslim -.10
No religion .04

*p < .05

More attendance of religious services predicts remaining faithful for the Abrahamic religions only, and the effect for Muslims is not statistically significant. Buddhists, Hindus, and people with no religion who are active in religious activities are just as likely to cheat as those who never attend. Being a religious Jew is the strongest predictor of fidelity.  


  1. Hi,

    I'm wondering, how likely are these groups to cheat? Wouldn't we have to know that in order to get a sense of the effect that religion has on the propensity to cheat?

    For example, non-religious Jews are probably very liberal. And aren't very liberal people more likely to cheat?

    There may be a huge difference between liberal athiest Jews and religious Jews.

    Maybe Protestants tend to be more monogamous, even with no religion, so that you find that religion has a much more negligible effect.


  2. Is there a particular reason you selected logistic regression as your model of choice? Was it because you wanted to fit model parameters that were easy to interpret from an economic/intuitive standpoint?

  3. A good point by AS above.

    Also, there is more than one way to measure religiosity. It might be too clumsy to use "attendance" (not clear what this even means -- ever? more than once a year? more than once a month?)

    Maybe it's better to use self-identification in religiosity: Very, Somewhat, Little, Not at all. And how each correlates to overall cheating rate. Following these results, we'd expect likelihood-to-cheat to get higher as religiosity gets lower.

  4. Jason: My thinking was very simple: dichotomous dependent variable; continuous predictor; let's use logistic. (I'm partial to it, too.)

  5. Wanderer: As far as I know, the GSS does not ask about how important religion is to the respondent. Attendance is actually pretty good since answers range across nine categories from "never" to "more than once a week." It might be more valid than other measures since it relies on objective behavior instead of subjective evaluations.

  6. Readers who are feeling theologically insecure over these results should consider the fact that religious Jews are almost always part of an insular community, where infidelity is extremely risky.

  7. Ron: That makes sense.

    I'd wonder what the "threshold" is -- i.e., at which of those nine categories people start to be more likely to cheat than the mean.


Allelic variants found only in populations of African ancestry predict kidney disease and preeclampsia in blacks

Study Link Black women in the United States and Africa are at an increased risk for preeclampsia. Allelic variants in the gene for apolip...