Doctor MJ wrote:I wouldn't say I don't use stats obviously, but I tend to be cautious about the conclusions I draw from them, and I think that's what Mufasa was getting at.
Ok, let us call it "numbers" instead of "stats", but both mean the same in the end. Stats are numbers, something which was a result of counting or calculating. Counting is just more simple than using some more sophisticated equations. And you draw constantly conclusions from those numbers. Heck, I would say EVERYBODY is at least biased by numbers in some sort of way.
And the group B Musafa described does not exist. We have seen multiple times discussions in which people claimed they understand "stats", but showed the inability to use them correctly. People don't like it, if facts are speaking against them. So, it seems natural to attack the stats instead of their own preconception.
Doctor MJ wrote:There are people who believe that they have the stats to say exactly what's going on (a la Berri), and then there are those of us using the stats as part of a broader analysis.
Berri is just one example, we can use Winston as an example from the +/- based stats. Neither of those are helpful in order to educate the broader fanbase in terms of advanced metrics. That might be a bigger issue, but most times the reluctance to use stats in not based on some objective measurements (we can show that WP48 isn't a good predictor or APM has big issues with overfitting and multicollinearity, which can only be solved with a huge sample size), but is based on the own preconceptions. People tend to use outliers as evidence that the stats are "wrong". But in a lot of cases the people just disagree with about 5 or 10% of the rankings. So, instead of using those stats as enlightment and finding an explanation for the difference between the "eye test" and those "stats", they just throw out that the "stats" are bogus, believing they own "eye test" more even though the bias is pretty clear.
Just recently my discussion with a pretty smart user went like this:
User: The stats is stupid, because I disagree with the results.
Me: Well, that is how it works.
User: Ok, looks solid, but I still disagree.
People are afraid to contradict their own preconceptions and they try to defend their existing position on players rather instead of re-evaluating the player and their own evaluation process. It is ignorance in the first place, it is overconfidence in their own abilties in the second place and it is the fear of being exposed as being not as smart as they want to look like.
mopper, I agree that there are people who understand the flaws of some metrics and are choosing not to use those specific metrics. But well, even PER can be used, if we put that into a certain context. Do I look for a scoring player with the high volume, I can certainly look at PER and choose the better scorer based on that. Well, if I look for a role player with great rebounding who is not taking shots away from better scoring options, I can use WP48 in order to pick one. Those stats have obviously flaws, but within a certain range they are still somewhat useful. Obviously, basketball is more than just scoring and more than just rebounding, that makes those two particular stats not good and that can be shown in predictions. But for all stats it is true, that they have to be looked at in a context. The stats are not intrinsic values of a player, they are results of an interaction of players within a 5 on 5 game.
