Yeah but correlation does not equal causation. In addition to what i said above, there is also the whole issue of multiple evaluations of the same data, ie multi-collinearity of variables, no reputable statistician will do that because it is not worth the paper it is printed on. When a drug company is evaluating a new drug, they have to go to the FDA apriori with a data analysis plan. They also have to outline the variables they will study for significance and what order they will analyze them in. This is all done before the trial even begins. So if they have submitted 10 variables and number 2 is not significant, they cannot even present the next 7 even if they turn out to be significant. This is because it is well understood that when you take a large database and just look for any correlation between factors a ton of things will come up as correlated by chance when in reality they are not, and vice versa.
Anyway, ive been down this rabbit hole with other posters and in the end it is: “yeah but” so it is not worth the time and effort. Like i said, show me multivariable analysis or C index analysis showing the run game is useless, then we can talk