I didn't check out the three teams and I'm not familiar with all the inner workings of NET rankings but could it be because not all Q1 games are equal?
Playing #1 on the road or #30 at home are both Q1 games but for these three teams one of those would be a big upset if they won and one a big upset if they lost. I think that the rankings factor that in.
Happens in football too. People talk about "ranked wins" and it can be a good shorthand but it can also be misleading because there is a big difference between, #1 on the road and #25 at home and there is an even bigger difference between #26 on the road and #115 at home.
6 was Iowa, 10 was Kansas, and 21 was Purdue (hence why I was looking at it). I didn't download the team sheets for each, so it's quite possible this is the case...
It could be that Purdue was cleaning up on low-Q1 wins and losing to high-Q1 opponents while the other two were doing the opposite.
But that would also mean that the quality of Iowa and Kansas' losses would be worse too, because if they're beating the high-Q1 opponents they need low-Q1 losses to balance.
The only thing that would make sense, then, is if their schedules were much less favorable and they were playing more high-Q1 games than Purdue, whose schedule would be more dominated by low-Q1 games, giving them less chance to excel. But that would seem to be belied by one of the other teams being Iowa, who faces a pretty similar strength of schedule to Purdue--and the B1G being stronger than the B12 would seem that both teams have tougher slates than Kansas this year.
Or, of course, margin of victory might play a role. Again I haven't looked closely at that.
I find it interesting... KenPom has Purdue at 14 and Sagarin at 15, but Bart Torvik has them at 21, right where they are in the NET. Interestingly Torvik has Kansas at 23 and KenPom at 19, so in both rankings Purdue is ahead, while Sagarin has Kansas at 10. The Massey Composite as of Feb 28 (so not reflecting the Purdue win over Wisconsin) has Kansas at 13 and Purdue at 19.
Being a numbers geek, I find it interesting.
Obviously NET is a records-based ranking. KenPom and Torvik are efficiency-based. I'm not sure on Sagarin.