opebo for once we agree.
Muon-- it looks like they used the 01-03 data instead of the 00-03 data, but the results were almost exactly identical. The only difference was the switch between AL and MN... MN though is growing at a good rate and I think less likely to lose than AL, which is being passed by by GA and TN.
I ran the data myself with the base at 4/1/2000 and the estimates from 7/1/2003. That is technically a 3.25 year period. I then applied the official equal proportions method (it uses the geometric mean of the impact of current and next seat). I get the same as my earlier post except that Al loses instead of MN, so you are probably correct.