By: Will Lester, Associated Press Writer
(AP) Polls conducted in races for governor and the Senate in 2002 were generally accurate, according to an analysis by the National Council on Public Polls.
The National Council of Public Polls monitors the polling industry and news media coverage of it.
The study of 159 polls in gubernatorial and Senate races found a generally good performance, contradicting some reports at the time that pollsters were having problems getting accurate results because of declining response rates. The analysis found that those reports were based on results of a handful of inaccurate polls and did not accurately reflect the larger situation.
Of the 159 public polls reviewed, 22 (14%) had the wrong winner. Four of five of the polls differed from the final results by less than the margin of error.
Some pollsters who had problems blamed a late surge of Republican support related to President Bush’s intense campaigning in battleground states in the closing days.
Most organizations conducted polls in only one or two states. Five organizations working in three or more states conducted 64 polls.
* Mason-Dixon conducted 23 polls in 16 states and had only one poll with the wrong candidate winning.
* Zogby International did 17 polls in 12 states and had five incorrect winners.
* Research 2000 did 13 polls in nine states and had two wrong winners.
* The Gallup Organization did seven polls in four states and had one mistake.
* Quinnipiac College did four polls in four states and had no mistakes.
* The 95 polls done by the other polling organizations had 13 wrong winners.
The mistaken polls refer to situations where the wrong candidate was ahead in the last release before the election.