HOWTO: report science in a lay publication

Say you are a journalist and you get a breathless call from a university PR office or a researcher touting a world-shattering scientific breakthrough. For instance, the discovery of a new "genetic link" to violence and delinquency. Hot damn! Well, you had better be the first to break this exciting story, lest you miss the prime opportunity of your career. The nature/nurture debate has been settled once and for all, right?

Not even close. The study I referenced above is an egregious example of what my former boss (an HHMI investigator at Yale who found the first gene and 20 subsequent others for hypertension) would call "genetics amateur hour." For one thing, the article was published in the "American Sociological Review," a SOCIOLOGY journal with a pathetic impact factor of 2.38. Anyone with even a modicum of genetics training would see that the study in question was a prime "fishing expedition" where the researchers looked at several of their favorite "candidate genes" hoping to find any relationship at all. Perhaps the most telling aspect of the study: one of their so called delinquency-causing genetic variants, "seemed to set off a young man if he did not have regular meals with his family."

It's called the Bonferroni correction, people. If you test a large number of hypotheses, it is certain that many tests will be statistically significant merely by chance. The fact that the study's authors broke down their cohort into "eats regular meals with family vs. does not eat regular meals with family" tells me that this is truly JV work. How many other tests did they try and not report?

In the spirit of giving constructive criticism, here are my guidelines for how to report science in the media:

1) If the work in question is not going to be published in a major journal, such as Nature, Science, Cell, PNAS, PLoS, or top speciality journals (e.g., "baby" Nature or Cell journals like Nature Genetics or Nature Medicine), don't bother reporting on it.

* A caveat to this point is if the work is not yet published and is being presented at a major scientific meeting, such as the American Society of Nephrology, FASEB Experimental Biology, etc.

2) Ask experts in the field. If you don't know who an expert is, find out. Call departmental chairs at other universities. Get a second, third, and fourth opinion on the work.

3) For God's sake, learn the difference between a gene, a mutation, and a phenotype. I will probably post on this again when a good example comes up.