Ok, I've brought this with you at length in the other thread and I'll do this one last time because I know you will try to change the subject because you don't have the answer.
Many of these things will depend on how you define it and how the person collecting the data defined it, because as you have admitted in the past, you have NO CONTROL of the data or HOW IT WAS ANALYZED, all you are doing is copying and pasting little bits and pieces of other people's work without any knowledge of whether it is accurate or not. You should know as a statistician that what data was collected and how it was collected and analyzed is CENTRAL to whether or not it is reliable. Yet you post statistics from data that you KNOW is incomplete and inaccurate.
Saying that the real data is too difficult to collect is not an explanation, that is an admission that the data you are posting is NOT RELIABLE and therefore should NOT be used to draw hard conclusions. Yet you post it repeatedly to try and shout down other voices as if you have accurate and reliable data, which you don't.
Garbage in garbage out
E.g. "Rushing success" was defined how? "Third down conversion had a good correlation" but a many of our 3rd downs are due to rushing and even our passing third downs are successful because the other team expects us to run so "rushing success" is inherently buried in "3rd down conversion ". This is patently obvious.
Anyway, go back to that thread and answer my questions I had brought up directly.
When data is inaccurately collected and without context and multiple correlations are run without corrrction, then flawed results are achieved. This not only a well known statistical issue but makes common sense.
Don't give us an isolated statistic that is convenient for you to make your point. This is EXACTLY why statistics have a bad name because people quote them out of context and parse bits and pieces without presenting the entire picture.
1. Show me where you KNOW data was collected in the context of situational running and other issues (e.g. running to run out the clock, running to get a first down, how often the running tendency of a team dictated the defense allowing the pass to be successful etc) as I had pointed out in that thread. What data SHOULD HAVE been collected, what WAS ACTUALLY collected and what COULD NOT be collected and entered and why not
2. show me what univariable models YOU RAN (not copied and pasted) and what was statistically significant in a multi variable fashion AFTER correction for confounding variables.
3. Show me where you ran other models like C statistic which are used to show the real significance of a measure in outcomes relative to other variables
4. Show me where any of this was published in a peer reviewed fashion so other experts could point out obvious flaws and limitations in the work
That has been my main problem with you. If you are a stats person you know this data has OBVIOUS limitations yet in your zeal to forward your belief system you ignore them and don't want to admit them. This tells me you have a conclusion looking for data to support it instead of following the data to let it lead you where it does as an open minded researcher would.
These are BASIC statistical norms and if you cannot answer them, then we will know you are simply copying and pasting other people's work and you need to stop trying to pretend you have any more information than any run of the mill blogger. I am doing this because other less sophisticated posters (like
@Idgit) assume what you post is gospel and run with it without realizing they are being made fools of.