Friday, January 7, 2011

95% of Statistics are Bogus (Including This One)

My own Statistics page still languishes in its unstarted state, but a new article in Science News manages to present, more or less in layman's terms, precisely the sorts of problems that I wanted to highlight for readers here. So in the meantime, it's a useful starting point.

Tom Siegfried explains very well that the scientific method and statistical analysis are not necessarily natural bed-partners, and that even the games we play to make statistics seem "reliable" can end up, in certain cases, actually making them more misleading. Most of this will be old news to people with stats training, but it's still a good refresher. There's also a section specifically on the problems inherent in clinical trials:
Statistical problems also afflict the “gold standard” for medical research, the randomized, controlled clinical trials that test drugs for their ability to cure or their power to harm. Such trials assign patients at random to receive either the substance being tested or a placebo, typically a sugar pill; random selection supposedly guarantees that patients’ personal characteristics won’t bias the choice of who gets the actual treatment. But in practice, selection biases may still occur, Vance Berger and Sherri Weinstein noted in 2004 in Controlled Clinical Trials. “Some of the benefits ascribed to randomization, for example that it eliminates all selection bias, can better be described as fantasy than reality,” they wrote.

No comments:

Post a Comment