from Amazon
Monday, January 09, 2012
What is evidence? Colin v Neyer
Rob and Colin are talking about evidence.
It sounds to me that Rob is REALLY talking about observational bias, not evidence. I tried finding a good definition for evidence. Kind of hard. At the minimum, it requires data or information or something that otherwise exists. It also requires an event to have occurred, or that relates to an entity’s property or behaviour.
So, evidence would require some sort of association of information to property.
What is the evidence that Edgar is more likely to have been a PED user than Jeter? That there were more PED users during Edgar’s time than Jeter’s? Well, that is an inference based on information. But, then we’re not talking about Edgar and Jeter specifically, but rather them as representative of a population. Rob is asking us to think of those two specifically.
We can list the 17,000 players in MLB history, and list their odds of having used PED from 0.00001 to .999999, and using nothing but Bayes and the Mitchell Report. Just because you use evidence and you apply Bayes doesn’t mean you have used evidence to learn anything about Edgar and Jeter SPECIFICALLY.
So, a third requirement in evidence would I think be that you can associate the information and event more directly to the specific entity being targetted.
Otherwise, we’re getting into an impractical philosophical discussion with no hope of having a resolution. That’s what drinking at a bar at 2 AM is for.