Friday 3 March 2017

Lies, Damned Lies, and Suicide Statistics

     Ages ago, when I submitted my scientific studies for publication, they were in turn submitted to a couple of anonymous reviewers who, I presume, had undertaken studies in a similar field, and they made recommendations - not all appreciated - on how my data could be presented, and their possible implications. Likewise, when I was asked to review someone else's paper, I also made recommendations - which I hope were appreciated. That is how peer review works. So how could a prestigious journal publish a paper whose data make no sense whatsoever?

    Take, for instance, the recent paper by Raifman, Moscoe, Austin, and McConnell, entitled
Difference-in-Differences Analysis of the Association Between State Same-Sex Marriage Policies and Adolescent Suicide Attempts,
 which was widely touted as proving that the introduction of same-sex "marriage" (SSM) in America reduced adolescent suicide by 7% overall, and 14% among homosexuals and lesbians. Right away, an alarm bell rang. How could even a large reduction among a very small minority result in such a large overall reduction? It was time to examine the document in detail.