Using the final RCP averages in battleground states:
http://www.realclearpolitics.com/...
and comparing them to the actual results we get:
Average amount off by: 1.47%
Standard deviation: +/- 1.33
Only one state (Wisconsin), had a different result than the polling
average indicated, and while 5 of 17 results were effectively dead on,
only two polling averages were off by 4%, and none were off by more.
State RCP Avg Actual Off By
FL +1 +5 4
OH +2 +2 0
PA -1 -2 1
IA 0 +1 1
MI -4 -3 1
MN -3 -3 0
WI +1 -1 2
MO +4 +7 3
NH -1 -1 0
NV +6 +3 3
WV +9 +13 4
NM +1 +1 0
OR -5 -4 1
CO +5 +5 0
ME -10 -8 2
NJ -7 -6 1
AR +7 +9 2
I could not find comparable 2000 polling data. If anyone could point
me to such data I'll love to look at it.