It's better than most polls. If we treat each group as 1/3rd of the electorate, then the 18-point Kerry lead among independents would give Kerry a 6-point lead. Bush's 86% support among Republicans vs. Kerry's 79% support among Dems would lead to a net gain of 4% for Bush (that's a 7-point difference in margin among 2/3rds of the electorate, that's over 4 points in Bush's direction).
So without any weighting of Dems over Republicans, his poll should show a 2-point Kerry lead. It shows a 4-point lead, which indicates a slight imbalance in favor of the Dems. He probably polled about 4% more Dems than Republicans.
This is what they're saying over at the Atlas Forum abuot this poll. The guy who goes by the user name "The Vorlon" is very VERY good at breaking down the internals of polls to say whether they are valid or not. But Zogby guards his internals so tight that all he could do was guestimate based on the info he had. Anyways here is what he said:
"http://www.zogby.com/news/ReadNews.dbm?ID=851
The self-discribed "worlds most accurate pollsters" has a new effort out for our consideration.
Bush's job approval went from -12 (44/56) to -5 (47/52) - a 7% improvement.
Zogby ask's his job approval question differently than other pollsters, so while the trend line is useful, the absolute value he finds is not directly comparable to other pollsters.
"Deserves to be re-elected" stays virtually the same at -10, versus -9 in his last poll.
Zogby finds that Kerry leads 49% to 31% among Independents.
Zogby guards his methodology in terms of weightings and other factors more tightly than Col Sanders guards his 11 different herbs & spices so I cannot offer any meaning comment on this poll.