All you have to know to determine that the results of that poll are statistically biased is to see who conducted it.
The only question is what percentage of the Jewish vote will Bush actually get. An intuitive guess is somewhere in the 30s. Although that might not seem like much, if Bush had done that in Florida in 2000, there would have been no recount circus.
If the President picks up 10% of the Jewish vote from 2000 and five percent more of the black vote, it'll be a landslide. And I do think he is doing better among both of these constituencies.