Science Is Broken. Can Better Statistics Help?

This article was originally posted on RealClearScience.

Science is going through a bit of a crisis. Thanks largely to the pioneering work of Dr. John Ioannidis, scientists — particularly biomedical researchers — are grappling with the reality that much of the scientific literature cannot be replicated and is simply incorrect. The reasons are manifold, among them the tendency of top-notch journals to publish “flashy” research, publication bias, flaws in peer review and the inadequacy of statistical tests. It is this latter problem that new research in PNAS hopes to address. Continue reading