Normalizing the goal differential figure (i.e., scaling it in a range from -1 to 1) produced the following avg. goal differential results:The crazy thing about statistics is that they can be fairly easily manipulated to show a desired outcome. The original statistics are flawed. You cannot compare three unlike groups and get reliable data. You compared groups of 14, 38, and 17. The data was not normalized thus causing incorrect correlations that favored your point of view.
Dual vs DA Only +0.52
Dual vs Left ECNL +0.54
DA Only vs Dual (0.52)
DA Only vs Left ECNL (0.21)
Left ECNL vs Dual (0.54)
Left ECNL vs DA Only +0.21
I'm not sure I understand why normalization improves the analysis though, as goal differential is measured based on the same scale for each group.
In any event, I didn't set out to manipulate anything and I'm happy to compare the results for the various clubs in any way you think is fair. Just let me know what you'd like to see.
Last edited: