There has always been a spread between the various polls.. but I think it is getting much wider. How can every poll give the race to Kerry except Gallup?
http://www.pollingreport.com/wh04gen.htmI notice looking closer that part of the difference seems to be between polling "likely" voters vs "registered voters" registered voters seem to be significantly more likely not to have made up their minds.
Then when you get into the details it gets even more obvious.
http://www.pollingreport.com/BushJob.htmFOX News/Opinion Dynamics Poll. May 18-19, 2004. N=900 registered voters nationwide. MoE ± 3.
Approve Disapprove Don't Know.
48 43 9
...5 days earlier
Newsweek Poll conducted by Princeton Survey Research Associates. May 13-14, 2004. N=1,010 adults nationwide. MoE ± 3.
Approve Disapprove Don't know...
42 52 6
*****
These polls cannot statistically both be right. The divergence is way over the margin of error in fact it is three times the margin of error.
*****
Or try this one.. Bush Favourability
http://www.pollingreport.com/BushFav.htmNewsweek Poll conducted by Princeton Survey Research Associates. May 13-14, 2004. N=832 registered voters nationwide. MoE ± 4.
Approve Disapprove Don't know...
46 46 8
Nine days earlier
FOX News/Opinion Dynamics Poll. May 4-5, 2004. N=900 registered voters nationwide. MoE ± 3.
Approve Disapprove Don't know...
50 42 8
*****
This time the variation is closer to the margin of error but it is still over it, and notably it exceeds it in Bush's favour by the maximum possible margin.
*****