Friday, September 05, 2008

Vasopressin made me do it, baby -- NOT.

Excellent post over at Language Log about the so-called news from so-called research on male fidelity. If you haven't heard about it, they found the "gene for" male infidelity in relationships; except, as you will see on Language Log, not so much.

LL had one of the most sensible suggestions EVER for reporting on genetic studies. Read it. Read it NOW.

Essentially, like the post on infidelity, they point out that so-called science journalism should point out, when they find a "gene for,"
[a]) the proportion of the case subjects with the genomic variant in question; and ([b]) the proportion of the controls with the genomic variant in question.

If you know what "effect size" means, you'll recognize that [the BBC's reporting on the study] can be translated as

"If you pick a random man with the 334 allele of the AVPR1a gene and a random man with a different allele, the 334 guy will get a higher score from his partner for strength of relationship bond about 47% of the time, and a lower score about 53% of the time.
That is, the research on infidelity showed that guys with The Gene (334 allele of AVPR1a) will be perhaps 6% more likely to be be rated as having a worse relationship with their partner than average guys without the gene. 6% more likely. That's the state of "gene for" research and reporting these days, folks. Sigh. 6% more likely to get a lower score on one particular evaluation of relationships than people without the gene. (And this was the result from the 41 (out of 1,104) guys who were homozygotic -- had two copies of this gene -- for ONE of the ELEVEN different types of this gene (alleles) at ONE of THREE areas in the genetic makeup that the scientists examined.

Genetic determinism, my left detached earlobe.

No comments: