This is in direct contradiction of another Zogby poll.
I see your confusion!! However, both the Zogby site and the AJC site reflect their numbers more than once each. So, I really doubt there's any typo or error. So, what's going on??
Well, while we can't know for sure, I'd bet that the 11 selective Southern states for AJC is different than the definition in Zogby. Two possible example of this...
AJC may not include Florida... a very high population, but evenly split non-Southernesque state. Also, in fact, it looks for sure (based on interviewees in the article) like they did include VA. So, KY was probably also included. That means only one of the following "borderline" states was included... FL, TX, or MO. Most likely, it was MO. But, to my second possibility... what if it was TX? That would also give Bush a huge boost, but could be in Zogby's "West" definition. Still, though TX is unlikely... MO is the most probably candidate. That means FL was NOT in the AJC numbers. Based on the breakdown on Zogby's page, it MUST be in his "South" section.
To go a little overboard here with this analysis... The total population of AJC's PRESUMED 11 states is about 45 million. Florida is close to 10 million. 45 million at 52%-37% Bush, averaged with 10 million at 45%-45% even comes out to..... Bush 50.7% Kerry 38.5%. While that does fall short of the 47%-43% split on Zogby's page, I've been very elementary here. For one thing, the average age of the Florida population is such that probably a much greater percentage of the population votes (in 2000 almost 60% of the population... not 60% of registered voters, but 60% of the population!!... voted). That could tighten things up considerably more. Plus, if Zogby had Kerry in the lead slightly in Florida (possible... since some polls show that), that'd tighten it even further.
Point is... THAT is probably the reason. AJC's 11 states includes VA, so it likely includes KY and MO. That would require it to exclude FL. Zogby's numbers on his page include FL.