Saturday, October 02, 2010

Infidelity, Part 2.

A far better study looking at the correlates of infidelity is,

Sexual infidelity in a national survey of American women: Differences in prevalence and correlates as a function of method of assessment. MA Whisman, DK Snyder, Journal of Family Psychology,Vol. 21, No. 2, 147–154.

This was an interesting study based on the 1995 National Survey of Family Growth and involved 4884 women.

It was a study with a twist.

The authors recognised, from previous research, that the response people give at an interview may be influenced by by the method of data acquisition. Namely, people are less likely to give accurate answers with regard to socially questionable activity in a face to face interview compared to an anonymous technique.

For "sensitive" data, the National Survey of Family Growth employs two techniques:
a) A traditional face to face interview.
b) A-CASI method, where the person listens to questions on headphones privately and enters the data into a computer, eliminating any "interviewer" effect.

What the authors did was compare data from the two responses and measured the difference.

Respondents were asked if they had a "secondary" sex partner over the previous 12 months. The study controlled for race, age, education, previous sexual abuse, religiosity, cohabitation and divorce.

Face to face interview revealed an incidence of infidelity of 1.08%
A-CASI interview yielded an incidence of 6.13%, nearly a six times greater increase in infidelity.

In fact, the authors found that when comparing the two techniques, significant differences in response were found with regard to:

1) Educational achievement.
2) Lifetime sexual partners.
3) Premarital cohabitation.
4) Ethnicity.

No differences were observed with regard to:

1) Religion.
2) Remarriage
3) Race.
4) Age
5) Childhood sexual abuse.

To quote the authors;
Thus, the fact that infidelity has been assessed using different methods across different studies may help to explain the inconsistent findings across existing studies. For example, if the current data obtained from the two modes of interview had been reported in two separate studies, three predictors would have been significant in one study but not the other, and the magnitude of the effects for four of the predictors would have been significantly different between the two studies. As such, the present findings regarding significant differences in the magnitude of the association between predictors and infidelity underscore the importance of consistent use of assessment methods across studies if a replicable set of predictors of infidelity is to be identified
And what were the correlates?
(Based on the A-CASI method of data acquisition, table has been modified and data omitted per copyright compliance, any breach is unintentional.)

Religion was scored on a 5 point scale, from never going to church to going to church more than once a week. Education was scored in years with the mean being 13.2 years.

Each additional sexual partner increased the odds of infidelity by 7% while increasing years of education seem to decrease the risk by 10%. Very roughly speaking each addition partner negates the benefit of a year of education with regard to infidelity risk. Yet another study demonstrating the effect of promiscuity on relationship exclusivity/stability.