/sci/ - Science & Math

SO - how is this relevant?

Well, sometimes you make very precise guesses about exceedingly unlikely events, and you will get bad results if your policy is "The test is very precise, so we will act as if it is also correct."

Let us say that I am a medical practitioner examining women for breast cancer. I am highly skilled, which in this case means that given any case before me, I will predict the correct answer 90% of the time and the wrong answer only 10% of the time.

But breast cancer is not so very common - among all the people who come in to get screened for it, only 1% actually has it.

Now I tell you that you (Or your daughter, if male) have breast cancer. Do you? Should I schedule you for a masectomy right now and disfigure you, or should we maybe run a different test, get a second opinion?

Pic related.