I'm reposting a couple of my comments from the Kyrie thread. I still believe that the main reason #1s perform better is because they are better teams. However, I will grant that #2s may have a slightly more difficult road because they might have to play a dominant #1 seed in the E8. Outside of the top 2 or 3 teams depending on the year (OSU, Duke, KU this year), I feel like there is no difference between the next 5-6 teams.
Here is my thought experiment:
If we flipped the #1s and #2s for the next five years (i.e., gave the four best teams #2 seeds and the next four best #1 seeds) and left #3-16 alone, do you think more #1s or #2s would make it to the Final Four? My bet would be on the #2 seeds (who are really the four best teams).
Also, another hypothetical: If we end up as a #2 with Notre Dame as a #1, would you prefer this to being a #1 with Kentucky as your #4 seed and Texas as your #2? UT and UK are both higher than ND in Pomeroy's ratings
I feel like the committee does such a bad job seeding teams that it doesn't matter if you get a #1 or a #2 as long as you avoid majorly underseeeded teams. And since the #1 seeds on average are the best four, they are the ones who will win.
I'm not going to use actual numbers, because it's hard to separate correlation and causation while analyzing them, unless we were able to run an analysis without seeds using power ratings.
Here is a quick read by John Gasaway:
http://www.basketballprospectus.com/unfiltered/?p=672Bear in mind that as long as the committee continues to use RPI as “just one tool” (a bit like using phrenology as just one tool), interest in who gets a 1-seed will exceed the importance of getting a 1-seed. Starting with the round of 32 the strength of your opponents is a crap shoot regardless of your seed, because the field’s been seeded using a system that thought New Mexico was a 3 last year. A 1-seed’s a really cool honor, but you may have an easier path as a 2.
There are two questions here. One is, regardless of anything else, do you want to be a #1 seed? The answer is an unsurprising but resounding yes. Since 1979, #1 seeds have won 18 national titles. All other seeds combined have won 14.
NCAA Record Book (see page 14-15)
The more interesting question (and the one I think people have in mind) is that, for a given good team, how much advantage is there in being a #1 seed versus a #2? This is trickier. Since 1985, #1 seeds have held a sizable advantage in winning percentage in each tournament round over #2s. #1 seeds are 91-13 in the second round (versus 67-33 for #2s), 75-16 in regional semis (versus 48-19), and 45-30 in regional finals (#2 seeds actually have a losing record, 23-25 in regional finals).
Bottom line, I think there is still an advantage controlling for team ability, but it's difficult to quantify without doing a full analysis. If anyone's interested, I've tracked down data with scores from every game played back to '85 or so (I found it through Google but can post a link tomorrow).