Fun with Research Studies: Something to Think About

When I was in graduate school, one of the classes I most expected to hate was statistics. But those two terms of statistics and research design turned out to be the most valuable courses I ever took. They allowed me to look critically at marketing claims for drugs ("Lipitor reduces heart disease by 33%!"), bodybuilding supplements (43% more muscle!), weight loss products (don't ask) and especially correlational studies ("saturated fat causes heart disease") that are deeply flawed and badly reported.

For example- does it really make sense that people taking some extra vitamin E are more likely to die? Or that people taking a multivitamin have a higher rate of prostate cancer? Come on.

Briefly, correlation studies show that two things are found together. But correlation studies do not demonstrate cause.

Two fun examples come to mind. Diabetes rates went up significantly during the Clinton presidency- the rate of diabetes and the Clinton years are significantly correlated. (Does this mean electing a Democrat will increase diabetes?) And in certain areas of Denmark the number of storks is correlated with the number of babies.

Both of these examples- and countless others- are 100 percent true.

So I read with amusement about a brand new study by World Bank economists showing that in houses with at least one smoker, a mobile-phone purchase led to a 32.6 percent drop in tobacco consumption for each adult, the equivalent of an entire pack a month! (The study was done in the Phillipines between 2003-2006)

You can imagine the field day that the press relations department of cell phone makers had with this one: "Stop smoking- buy a cell phone!"

I mention this because it provides an excellent object lesson: Read studies carefully. Just because two things are found together in a population does not mean that the presence of one causes the other. Dig a little deeper, and there's often another explanation.

Moral of the story: Think before unquestioningly accepting any sound bite on the news about the "findings" of a new study.

Sometimes there's really something there- but sometimes it's all smoke and mirrors.