New Video on the Spitzer Study

The "Spitzer study" is the one and only peer-reviewed piece of research that can be interpreted to suggest that therapy can change a person's sexual orientation from gay to straight. Spitzer set out to find people who said they had changed, narrowed down his list, and asked a smaller number of them some questions over the phone, concluding that some of them had actually changed.

As far as I know, this is a completely unique form of social science research. The two classic research forms are the survey and the experiment, and variations on those. In the survey, you select a sample of respondents to represent the population that you want to make inferences about. Through careful sampling and weighting, and careful attention to details of the measurement situation, it is possible to make accurate estimates of a population value through inference from your sample.

The Spitzer study is not a survey. It did not attempt to estimate the number of people who can change, it only went through a large number of people to see if any of them were convincing.

In an experiment, subjects are randomly assigned to receive one treatment or another. Because differences between groups are random, and because random variation is well understood as a probabilistic phenomenon, it is possible to analytically assess whether the experimental manipulation caused a change in a measured outcome.

The Spitzer study is not an experiment. He didn't manipulate anything, there was no independent variable, and no attempt to discover causation.

Spitzer appears to have come up with this method himself. His reasoning seems to have been that, if sexual orientation can change through therapy, then he should be able to find someone who this has happened to. And after interviewing thousands of people, he found a few who convinced him that they had actually changed. This is a nice reasonable way to satisfy your curiosity about something, but I'm not really sure how it got past a journal's reviewers.

Dan Gonzales at ExGay Watch has produced a new video discussing the Spitzer study. You can see it at Google Video. It's about twelve minutes long, which is too long to go on YouTube in one piece. The video walks you through the study, discussing how it was done and what it really means.

There are two terms that are mentioned in the video, which I'd like to underline. These are internal validity and external validity.

Social scientists work with many kinds of issues in scientific validity, and these two are considered very important.

Internal validity asks the question, were the effects observed in an experiment actually caused by the manipulation? Because Spitzer's study is not an experiment, the question of internal validity for this study would be: is the researcher measuring what he thinks he is measuring? It is a clear and important question in this study: was the "change in sexual orientation" that Spitzer reported really a change in sexual orientation, or was there some other reason the respondent would try to convince the researcher that they had changed? Let's just say that self-report is very unreliable, and there are lots of reasons his respondents might be motivated to exaggerate their "change."

External validity has to do with the generalizability of a finding. This is where both surveys and experiments have well-studied statistical properties that let you place, for instance, margins of error around population estimates, or confidence intervals, or do significance testing. Spitzer's study doesn't allow any of that. He simply went out to try to find if there was anybody who could convince him that they had changed their sexual orientation.

This video is nicely produced. It is a non-technical discussion of the study and some of its issues, just to get the topic on the table. Go ahead, pour yourself a cup of coffee and click on the link.