Dr. Elizabeth M. Whelan offers some very sensible advice in her TCS column:
Consumers must be constantly skeptical of studies linking variable A with disease B, whether the news is potentially good (as was the coffee/liver cancer link) or just plain scary (many people in the early 1980s made a concerted effort to wean themselves from their coffee habit after the coffee/pancreatic cancer headlines appeared).
Personally, I blame the MSM.
None of this is to say that the media is entirely blameless; but personally, I blame the university system in this country. Or, rather, the pathetic, watered-down training in writing endemic outside of the very top schools (and they're not entirely blameless, either). The impression that these epidemiological correlations mean absolute causation comes from bad writing, combined with not-much-better reading skills. I, for one, am sick to death of attempts to make complicated things simple for the sake of the purported laziness of the reading public. As bad a book as The Davinci Code is, it is not one to be tackled by a lazy reader. Neither, at younger ages, are the Harry Potter books; or, at any age, The Lord of the Rings. Perhaps laziness sets in only when there isn't anything really to say (although perhaps Brown's tome is a counterexample!).
Dr. Whelan makes her point accurately, elegantly, and concisely. Without going into technical details for a nontechnical audience, she correctly distinguishes between correlations drawn from epidemiological data and causation. (That she doesn't go into the difference between empirical and epidemiological is more than excusable in this context!) More importantly, she does so without misstating anything. That is good writing. And one can hardly blame the MSM for endemic bad writingjust look at statutes, business plans, and just about any other document produced in an "official" capacity!