No, actually the poll shows Kerry at +10.
Just figured I'd toss an "Educational Feature" into the mix...
The thread on Kerry +10 in Michigan is an example of an
unweighted poll.
Purely as an educational example I "reweighted" the raw data in this poll using the 2% GOP party ID advantage suggested by the PEW data..
NOTE - I DO NOT SUGGEST IN ANY WAY THAT THIS REWEIGHTED RESULT IS THE "RIGHT" ONE, OR IS MORE REPRESENTATIVE OF REALITY THANT THE SUSA POLL.
THIS IS "PURELY" AN EDUCATIONAL EXAMPLE.
(ok.. that is enough disclaimers)
Raw Results
Party ID
GOP DEM IND
Bush 147 10 60
Kerry 9 182 78
Other 7 13 17
Raw Results as Percentages
GOP DEM IND
Bush 90.18% 4.88% 38.71%
Kerry 5.52% 88.78% 50.32%
Other 4.29% 6.34% 10.97%
x 0.38 0.36 0.26 (Assumed Party ID in Likely voters)
BUSH 0.342699387 + 0.017560976 + 0.08516129 = 44.5%
KERRY 0.020981595 + 0.319609756 + 0.110709677 = 45.1%
OTHER 0.016319018 + 0.022829268 + 0.024129032 = 6.3%
Final Result..
Bush 44.5%
Kerry 45.1%
Other 6.3%
NOTE - I pulled 38/36/26 TOTALLY out of my hat, I would have to look at election data for a day or two to know the "right" number to use...
Again, this is simply an educational example, but it does show the huge impact that the weighting pollsters do can have on the final result.
The way some of these polls are so massively weighted, at the end of the day you are trusting the pollster as much as you are trusting their survey and sample....
