Thursday, January 8, 2009

Non-peer-reviewed research

I'm disappointed that former director of the NIH
Bernadine Healy is citing non-peer-reviewed research. A director of the NIH should know better.

Specifically, she cites of Robert Rector and colleagues at the Heritage Foundation. I read the paper when it came out and found two things notable.

First, research papers are always named with a descriptive title that at best suggest their conclusions to maintain objectivity. The title of their paper was a headline: "Teens who take virginity pledgers have markedly improved life outcomes."

Second, it was extremely thin on both psychological theory and statistics, without which no one can be sure that they didn't just seek out factors that yield the results they want without correction. Given that the paper had a headline rather than a title, the idea that the conclusions preceded the study is not particularly far-fetched.

The advantage of matching on the other hand ensures that you can't search for results: you get a match and once you settle on that being the right match, then you find out what the results are.

No comments: