Playing hunches? Or playing innumerate?

In case you don't know, I'm a grad student in epidemiology. My current class in my major area, Design and Analysis of Studies in Epidemiology, started off with discussions and readings on causation, bias, and confounding. And then I read this.

From the title and introduction, I was quite positive; it seemed the author was going to extol the virtues of evidence-based medicine, which (in my opinion) is irretrievably linked to epidemiology. It is the reason we want to apply statistics to health data. It is the best way (again, in my opinion) to choose public health plans that maximize the return on our dollar.

But people not drinking the evidence-based Kool-Aid seem to worry a lot about the ecological fallacy. That's what happens when group data is inappropriately applied to individuals. That's why people thought that it was wrong for a doctor not to order what he thought was an inappropriate test, although the person in question later developed the disease in question.

Too often, however, these people fall prey to the atomistic fallacy. That's what happens when individual data is inappropriately applied to groups. That's why the jury awarding a settlement in the above case was acting more with indignation than sense in second-guessing the doctor and his education.

What's my problem here? Well, I never heard of the atomistic fallacy, in exact or inexact terms, until I did the reading for my class last week. You may have noticed that I do quite a bit of reading on my field. Does that seem odd? I've heard lots about the ecologic fallacy, although never with that term attached, but never anything about an atomistic fallacy. I think that is because of a general atomistic leaning in science and medicine. I've been hearing about (and experiencing) the bias towards bench science since I became interested in epidemiology. What we do, playing with numbers, is not real science in the eyes of many of our colleagues. Of course they would caution against our findings, if they don't trust our methods. But why would they talk about the atomistic fallacy? Their source of funding is reliant on individual, atomistic findings having an effect on a grand level.

That's what most of this comes down to: funding. Researchers competing for funding need to make their research sound applicable. The dominant paradigm is atomistic, so ecological issues are held up for extra scrutiny. And evidence-based medicine, as opposed to "the art of medicine", is considered just as suspect.

Truth is, evidence-based medicine is as much of an art as the old-school anecdotal medicine. The problem is that the art is now based in math and statistics, not hunches. And people going to med (and vet) school don't generally like math and statistics. People in the general public don't generally like math and statistics. Nobody likes art in a medium they don't like.