Essentially what is being said here is that win % and recruiting rankings, while related, are not everything. There are more variables effecting win % than just the recruits - which is obvious when you say it out loud.
So you analyze the same correlation with multiple variables (SOS, recruiting, etc...).
I would assume that if you took Win % of current coaching staff, SOS, and recruiting rankings and used those variables to predict win %, you would have a very high correlation.
This isn't rocket science, it just looks more complicated than it really is. Thus the saying, lies, damn lies, and statistics.