Quote:
Originally Posted by Enoch Root
lol
thanks for proving my point for me, Jay.
|
Your point was that you could use numbers from football to do a numeric analysis from hockey. I didn't prove that point, because it was obviously wrong.
I thought you said you were done; but since you obviously aren't, here's a basic lesson for you from a professional statistician:
Quote:
Which brings us to today's topic: correlation. Ever since David Hume, correlation has held pride of place over causation due to the inability of inductive reasoning ever to establish causes. It establishes only a co-relation between two (or more) variables measured on the same unit. Thus, it requires:
measurements
on the same units
Neither is as simple as it sounds.
(Michael Flynn)
|
Your numbers from the NFL are not measurements; they are numbers on an arbitrary scale, derived by an unspecified method from measurements that you do not provide. No units are given.
For the NHL, you have provided no numbers, no measurements, and no units.
You then assert that the NFL and NHL numbers are correlated. You have established no correlation, because the
very minimum you need for a correlation is
two sets of measurements in the same units.
You then claim that this correlation is so very strong that you can simply take the NFL numbers as an accurate proxy for what the equivalent NHL numbers would be if they existed. You have absolutely no basis for making such a claim, and any statistician in the world would laugh at you for it.
I'll offer you a deal: You stop making fatuous arguments, and I'll stop calling you out for it. This has nothing at all with my needing to be right; this is about having an honest discussion. God forbid that any uninformed person should be misled into thinking that your argument represents a valid way of using statistics. Things are bad enough already.