Sunday, December 2, 2012

Never Let Significance Get in the Way of a Good Story???

Via Seamus Ross, great example of misrepresenting research and the statistical significance debate, in this article by Robin McKie for The Guardian. Here's a short excerpt:

They are mainstay stories of tabloid newspapers and women's magazines, linking common foods from burnt toast to low-fat salad dressing to cancer. But now US scientists have warned that many reports connecting familiar ingredients with increased cancer risk have little statistical significance and should be treated with caution. 
"When we examined the reports, we found many had borderline or no statistical significance," said Dr Jonathan Schoenfeld of the Harvard School of Public Health in Boston. 
In a paper in the American Journal of Clinical Nutrition, Schoenfeld and his co-author, John Ioannidis of Stanford University, say trials have repeatedly failed to find effects for observational studies which had initially linked various foods to cancer. Nevertheless these initial studies have often triggered public debates "rife with emotional and sensational rhetoric that can subject the general public to increased anxiety and contradictory advice".