This page will present my observations on the misuse of statistics in scientific papers and reports about them in the media.
Many scientific findings reported in the media turn out to be wrong. A recent op-ed column in the New York Times by a philosopher places the blame for this situation on science writers. While it's true that science journalism has its deficiencies, in this case the philosopher's allegation is baloney. The fault lies not in science writers, but in science itself. Basically, the problem is that statistical methods for testing hypotheses are flawed. I discussed these issues in detail in a lecture at the University of Tennessee in 2009, available here.
The FDA says it will reconsider restrictions on the diabetes drug Avandia. About time. The 2007 paper linking Avandia to heart attacks was seriously flawed, as I described on pages 18-19 of my Hill lecture and in a 2010 Science Newsessay.