Doctors and computer scientists

I do admire doctors. Cardio-thoracic surgeons are able to take a scalpel and actually fix a human heart. This is an unimaginable thing for me.

Somehow, the data analysis doesn't seem unimaginable for doctors. They don't hesitate to take a table of values, add, divide, calculate a mean value and discuss a p-value. A doctor can think that he knows, what's an analysis about. When I'm doing an analysis, a doctor can look over my shoulder and say: “Well, this makes sense.”

I guess that looking over a surgeon's shoulder I would say the same. But it does not mean that I'd be able to perform a surgery. If anybody would see me taking a surgeon's knife and approaching a patient, I would be stopped immediately.

You might say that it's a matter of mistake cost. A mistake during analysis doesn't kill. A mistake during surgery does. But is it really like this with the analysis? A wrong analysis can miss an important effect, which can result in mistreating a large number of patients, so this (possibly) little error would be repeated over and over… So what's worse, one big mistake or thousands of small ones?

Doctors, before creating scores and calculate mean values, please go to an analyst and have a little talk. It will help everybody.