A little 15-minute study of whether polls have any connection to reality.
I took
this page which lists the "last-minute" polls taken just before election day. I decided to limit it to the senate races.
The polls were exactly right 3 times, within their published margin of error 7 times, and "wrong" - outside their margin of error - 10 times.
No wonder the elections were such a surprise.
The biggest offender was Zogby, which got it wrong 7 out of 9 times.
South Dakota
Zogby GOP 5%
USA/CNN/Gallup GOP 3%
Actual DEM 1%
Zogby wrong, U/C/G at outskirts of margin of error
Minnesota
Zogby DEM 6%
Actual GOP 3%
Zogby wrong
Missouri
Zogby GOP 8%
U/C/G GOP 4%
Actual GOP 1%
Zogby wrong, U/C/G within margin of error
Colorado
Zogby DEM 5%
U/C/G GOP 2%
Actual GOP 5%
Zogby wrong, U/C/G within margin of error
New Hampshire
American Research Grp GOP 4%
Univ of NH GOP 1%
Actual GOP 4%
ARG exactly right, UoNH within margin of error
Arkansas
Zogby DEM 13%
USA Today/Gallup DEM 8%
Actual DEM 8%
Zogby wrong, U/G exactly right
New Jersey
Zogby DEM 13%
Quinnipiac DEM 11%
Actual DEM 10%
Zogby and Quinnipac within margin of error
Texas
Zogby GOP 4%
Actual GOP 12%
Zogby wrong
Georgia
Zogby DEM 2%
Actual GOP 7%
Zogby wrong
N Carolina
Zogby GOP 10%
Actual GOP 9%
Zogby right
Iowa
Survey USA DEM 22%
Des Moines Register DEM 9%
Actual DEM 10%
Survey USA wrong, Des Moines Register within marg of error
South Carolina
Mason Dixon GOP 17%
Survey USA GOP 1%
Actual GOP 10%
Both Mason Dixon and Survey USA wrong... in different directions.
Tennessee
Survey USA GOP 11%
Mason Dixon GOP 10%
Actual GOP 10%
Mason Dixon exactly right, Survey USA within marg of err