I saw an article that managed an interview with Bradley's internal pollster, and he was also of the opinion that there was no Bradley effect. He seemed to think that Mervin Field (he of the Field Poll) didn't account for the absentee ballots (which were new at the time) -- and then blew smoke instead of admitting a flaw in his model.

It's still not a good idea to trust the Field Poll. Among other things, they were totally wrong on Prop. 8 this year.