The committee is officially provided RPI rankings and officially told to use RPI as a metric. Members are allowed to supply their own metrics, if they like, and individual members can present those as part of their discussion points for/against different teams, but RPI still rules the roost ---- at least until next season when reports indicate the NCAA may decide to insert other metrics into the official mix.
That said, most of "outliers" discussed in this thread are pretty easily explained with RPI. Michigan State, withstanding.
According to ESPN:
Michigan finished as RPI 30 -- which works out to an 8-seed, but close enough range to warrant the committee's 27 overall slot and a 7-seed.
Oklahoma State finished tied for RPI 39 --- which works out as a 10-seed
Wisconsin finished as RPI 32 --- which equates to an 8-seed
Minnesota finished as RPI 20 --- which equates to a 5-seed
You and I might be able to see the flaws in RPI as a metric, but that's the "advanced metric" the committee is tasked with using. And it appears that metric played a major factor in seeding this year.