What does it really prove?
Neither the six-cities study, nor any other statistical study, can show causation. Rather, it shows that two things occur at the same time -- that they are "correlated." And that offers ammunition to critics who say there is less here than meets the eye. "It's a correlational kind of analysis that didn't pay serious attention to any other pollution than particulates," says epidemiologist Suresh Moolgavkar of the University of Washington.

However, the study itself contradicts Moolgavkar: It turns out that Dockery and his colleagues examined "total suspended particulate matter, sulfur dioxide, ozone, and suspended sulfates" as well as PM-2.5, PM-10, and PM-15 and airborne acids. See "An Association Between Air Pollution and Mortality in Six U.S. Cities" in the bibliography

Nevertheless, Moolgavkar says "the results indicate that death rates are highest in Steubenville, and it's interesting to note that air pollution is also highest in Steubenville. I don't deny that it could be due to air pollution, but it could be that people are more sedentary in Steubenville," or numerous other factors, he says.

Nobody in the epidemiology business would be dumb enough to claim that correlation equals causation. (The Why Files covered some basics of epidemiology in a painless primer.) But C. Arden Pope III, a researcher at Brigham Young University, counters that the Six Cities study did not merely compare Steubenville to Portage. "The analysis was not done with two or three cities, but with all six. And they line almost in a straight line when you look at the dose-response (defined) curve." In other words, more particulate pollution level was found in cities with a higher death rate. Had any cities contradicted the hypothesis that disease is correlated with PM-2.5 pollution, "then the whole thing would have fallen apart," Pope adds.