Map of life expectancy at birth from Global Education Project.

Wednesday, July 20, 2005

Likely to confuse ?

Maybe even more likely to bore you, it's dirty work, but somebody has to do it.

In public health, there is a high probability that we are speaking in terms of probabilities. One way we talk about probabilities is in terms of risk. What is your annual or lifetime risk of breast cancer? How is your risk different depending on your reproductive history, your body mass index, your age? Of course, more than likely, we're actually talking about the probability that you will be diagnosed with the disease, not the probability that you have it. That probability depends on how hard you, or your doctor, look for it, as well as whether or not you have it.

But very often, we are dealing with what I might call meta-probabilities. Much of our information does not come from counting up facts about 100% of the population. Instead, it comes from samples that we believe are "representative" of some population. The probabilities we derive from those samples are only estimates, but we try to be very precise about what we mean by estimation. We talk about confidence intervals -- the probability that an actual risk lies between an upper and lower boundary. But is it meaningful to talk about the probability of a probability? Now this starts to get philosophically complicated. Can this sort of finding properly be called "knowledge"?

When the hairhatted weather android on TV says that "the probability of precipitation is 50%," is what he says meaningful? It might rain and it might not, but I already knew that. What if I said that when the weather forecaster says the probability of rain is 50%, there's a 50% probability that it's actually more than that, but there's a 67% probability that it's less than 56%? Does that mean something? I'm not sure, but we make analagous statements in public health all the time. The bottom line is still the same, it might rain and it might not. But we still watch the weather forecast, and depending on what the weatherdroid says, we do or do not carry an umbrella.

In public health, we hope that people will do the same -- eat what we advise, get certain contaminants out of the air, water and food, get the shots, take the pills, get the tests, exercise, have the surgery, or not, whatever it may be. But we tend to do a really lousy job of explaining ourselves.

So be warned. I'm going to try to explain things better. Be prepared for some serious wonkery. It it's boring, skip it and come back in a couple of days. If you like it, your probably need to get a life, but welcome to the club.

No comments: