It should. Firstly, the Spurs aren't a bad rebounding team -- they are a bad offensive rebounding team. They were third in the league in defensive rebounds. The Bucks were third-to-last. So Splitter was competing with better defensive rebounders like Duncan and Leonard, when Sanders didn't have anyone comparable. Then on offense, since the Spurs don't go for those boards, Splitter's rebounding rate is at a significant disadvantage. Not that this means Sanders isn't a better rebounder (he is), but it does explain some of the difference.
Regardless, that's why I said it was just an example. The other percentages are a big factor here in addition the other ratios brought forth by the formula.
http://www.basketball-reference.com/about/per.html
As far as offensive rating goes, part that you quoted said that you have to look at ORtg in relation to USG%. Splitter's USG% was 18.2 last year. Sander's was 16.9. That means that the difference between their offensive impacts is actually slightly greater than the ORtgs suggest. Splitter was more productive in a bigger role. I didn't nitpick. Splitter is a significantly better offensive player than Sanders is, and all advanced stats will bear that out.