Battleground State Poll Analysis: Part Two - Mason-Dixon and Gallup

Part Two (read Part One here) of my battleground state poll analysis continues with the final polls conducted by Mason-Dixon (M-D) and Gallup.

These two firms are among the biggest names in the field. M-D went 21 for 21 in 2000 and 22 for 23 in 2002. Gallup has a well-deserved reputation for national polling, particularly its likely voter model, but its state polling has been questioned.

The battleground states selected for this study are Iowa, Florida, Michigan, Minnesota, New Hampshire, New Mexico, Ohio, Pennsylvania and Wisconsin. All results are listed with the support of President George W. Bush first and the support of Senator John F. Kerry second.

For Ohio, M-D conducted two polls very near in time to each other and election day. One for the Cleveland Plain-Dealer with 1500 LVs and one as part of its "swing state" polling with 625 LVs. The latter concluded its sampling one day later than the former and, thus, its results will be used.

Florida:
Results: 52-47
M-D: 49-45
Gallup: 47-50

Iowa
Results: 50-49
M-D: 49-44
Gallup: 48-46

Michigan
Results: 48-51
M-D: 45-47
Gallup: n/a

Minnesota
Results: 48-51
M-D: 48-47
Gallup: 44-52

New Hampshire
Results: 49-50
M-D: 46-47
Gallup: n/a

New Mexico
Results: 50-49
M-D: 49-45
Gallup: n/a

Ohio
Results: 51-49
M-D: 48-46
Gallup: 46-50

Pennsylvania
Results: 49-51
M-D: 46-48
Gallup: 50-46

Wisconsin
Results: 49-50
M-D: 46-48
Gallup: 52-44

Predicted Winner
M-D had a record of 8-1-0 and Gallup had a record of 2-4-0. Big advantage M-D.

Average Error In Spread
M-D had an average error of 1.6 points. Gallup had an average error of 5.7 points. Big advantage M-D.

Bias Towards Each Candidate
M-D underestimated Bush's support by an average of 2.2 points and underestimated Kerry's support by an average of 3.3 points. Gallup underestimated Bush's support by an average of 2.0 points and underestimated Kerry's support by an average of 1.5 points. Thus, M-D had a pro-Bush bias of 1.1 points and Gallup had a pro-Bush bias of 0.5 points. Although this appears to be a small advantage for Gallup, that firm had three results that had a pro-Kerry bias. Because two of its results had a huge anti-Kerry bias, the average is pro-Bush. However, due to one-half of its results overestimating Kerry's support, they cannot be given the advantage in this category. No advantage.

Conclusion

Although this analysis contains a fair amount of rounding (and rounding of rounded numbers) and is based on state vote totals that are subject to change as absentee ballots are counted and other revisions are made, it appears that each polling firm did demonstrate results consistent with each's reputation.

M-D had excellent results including three races in which it correctly called the winner and the spread and three other races in which it correctly predicted the winner and missed the spread by only one point. Gallup's results were awful. It completely miscalled Florida, Ohio, Pennsylvania and Wisconsin. It predicted the wrong winner with spread errors ranging from six to nine points. In Minnesota, it called the right winner but still missed the spread by five points. Yet, this result with a five point error actually is its second best result.