SOS has the following correlation to a team's post-season (2018-2019) winning percent:
(Data from Sagarin for all 353 teams: Correlation between each team's winning percent and SOS)
SOS 0.353010744
Weak. Why? Because most teams that have a strong schedule get beat. Chicago state had a fairly strong SOS but a horrible record. The strong SOS didn't make them a better team. For most teams, it doesn't make them win more games. But SOS is a positive when comparing teams with similar records to determine which team should get at-large bids. When you regress other metrics that have a stronger correlation to winning percent alongside SOS, the SOS coefficient typically becomes negative, which ends up punishing very good teams with strong SOS's. SOS can blow up an otherwise good model for predicting which team will win in the NCAA tournament.
And yet it is precisely why the 7th or 8th team from a P5 league gets in over the 2nd team from a mid-major league more times than not...
Whew! When I saw the post topic line, I immediately thought " Aw $#!£, we lost other transfer and the ship was going down." ;D
Thank you for looking at statistical analysis to address a question.
While I am not sure what method was used to calculate the correlation, my guess is that this is a Pearson correlation. If that is the case, a correlation (R) of .353 on a sample of 353 teams is highly significant. However, the inclusion of other variables in the prediction model makes a lot of sense.