Golson is right and you are wrong. You are doing a uni variable analysis assuming past experience as a predictor for success. It’s incorrect. The fact is most coaches are unsuccessful and only a few are, and there are several factors that go into that. The other fact is that if you have past experience and you ascend to a better job it’s BECAUSE you have had success so you are already dealing with a limited and biased sample size. None of that matters to you because you have a narrative and you are backing into the analysis.
I am actually now inputting a spreadsheet which I will input into a database and run a few prediction algorithms. I'll start with simple enough regressions...dependent variable = winning %, independent variables = total years coaching, years head coaching experience, wins, losses, ties, age, start year, end year.
I will add classification as well. Maybe there is a correlation based on era, age, even education if that is added, within some groups...successful coaches clustering based on some factors? We can even throw in some factor analysis, eliminating factors that could be added to a model.
I'll do it for all ND coaches all time. If somebody wants to give data, we can add more. It will be a fun exercise. I do this kind of work anyways, maybe I can demo it for my customer? I'll do this in python using the usual freely available functions. I can post it here if anybody wants a crack.
Let's see what we can cook up. Again, statistics are statistics...let's see.