Friday, March 13, 2009

Do abstinence pledges work?

This article presents a nice analysis of a statistic about abstinence pledging. The study in question uses a matched-case design, in which they get people who are similar in as many ways possible except for the key difference under study. Then, the two groups are observed to see if they act differently over time.

Turns out that people who are identical in regards to their sexual attitudes, behaviors, values, etc... but differ only in making an abstinence pledge end up having similar sexual behaviors.

While the authors make various criticisms of the study, I wonder if there is another criticism to be noted. Specifically, perhaps the study controls for the causal mechanism, thus suppressing the effect of a pledge. If pledges work, it might be by changing values and attitudes. The study in question, however, makes these similar for both groups--hence no difference.

Here's an everyday example of how this might work. Eating chocolate can cause weight gain by increasing the number of calories a person eats. However, if a study compares people who consume the exact same number of calories, then those who eat chocolate will probably not gain any more weight than those they are compared to. The conclusion: Eating chocolate does not cause weight gain (and I die a happy sociologist). In reality, the researcher just controlled for the causal mechanism.

Going back to the abstinence study, it may be that young people who make an abstinence pledge differ in their behavior precisely because it changes their values and attitudes.

Thanks Andy!

No comments: