>         To me, the implication of "the committee could easily see
> those teams were beating all the rest under consideration", coming
> right after a statement that pairise comparisons are used to determine
> the at large bids, is that the triage is done using PWR, not RPI.
 
I guess I just think you are taking everything too literally.  The
committee can easily pinpoint some lock teams via comparisons or RPI.
But what you must understand is, they don't use PWR per se .... they
didn't even know what it was.
 
I got the impression they look at RPI -- can tell that some teams are
clearly locks -- they can also be doing comparisons between this team to
that team, and start to get an idea that they are locks .... then
they'll start seeing teams that aren't clear cut, and that's when they
start comparing them to each other.
 
It winds up being the same thing as a direct application of PWR almost
any way you slice it.
 
>         Well, it's perfectly reasonable to compare all the possible
> at-large teams, and use the results of those comparisons to pick out
> the bubble and compare the individual teams.
 
Yes -- but they don't total wins ... a la PWR ... when approached with
that concept, which is what PWR is -- Marsh realized that it was very
close to what was done -- but not realy ... they're not that methodical
about it.
 
 
>         One of the reasons using the comparisons to identify the
> bubble makes more sense to me is that you can usually tell at a glance
> how big the bubble needs to be.  (Including the case where you have N
> at-large bids to give out and N teams win the comparisons with all the
> rest, when you don't need a bubble at all.)  If you're using an RPI
> cutoff, how many teams on either side should you include in the
> comparison?
 
I know that -- that's why I think they should use a direct application
of PWR to determine the teams ... but since they don't "TOTAL the
comparison wins" like PWR does -- that don't EXCPLICITLY do that -- then
how could they ever figure out bubble teams by their PWR win totals?
 
They don't -- I was told as much ....
 
Now -- whether they use RPI --- or just start comparing one team to
another and start to get a vague picture -- well, it's probably a little
of both.
 
You would think they would actually jot down all the teams this team was
winning comparisons .... and then, PRESTO, you'd have exactly what PWR
is ... but I was told that no such thing goes on.
 
So then -- how do they decide the bubble .... it's just looking at
comparisons and RPI and giving themselves a general idea of who's in the
mix ... so then they start getting an idea of who to do head-to-head
comparisons with.
 
You see what I mean .... I think you're just taking everything too
literally ... and the point I've been trying to make is that committee
isn't that literal.  And I wish they were.
 
 
>         I don't think either one of us is going to convince the other
> that one method is better.  I just wanted to challenge the statement
> that using the total PWR is *obviously* preferable, and that the
> committee couldn't help but do it that way if they really thought
> about it.  If you'll recognize my point about the pairwise comparison
> already factoring in performance against other teams, even if you
> don't agree with it, I'm satisfied.
 
I suppose I can see the point ... but I still disagree  :-)
 
 
AW
 
HOCKEY-L is for discussion of college ice hockey;  send information to
[log in to unmask], The College Hockey Information List.