FWIW, I'm the analytics geek who tracks all kinds of raw and performance metrics for my 32-man league. Have for 2-3 years now.
I've swung multiple penalty sliders from 100 to 0 and seen absolutely zero impact on our #s (other than the penalty calls themselves). Most other gameplay sliders, sometimes even just moving them 2-3 points can result in a noticeable statistical effect.
So color me staunchly on the side of "penalty sliders = pure confirmation bias," at least until a single shred of evidence is posted otherwise. All I ever see is "trust me."