Election Exit Polls Were Flawed

Exit polls for the 2004 election were "not as accurate" as polls produced in the past, according to a study by the firms that conducted the polls.

The exit poll estimates "overstated John Kerry's share of the vote nationally and in many states," says the report released Wednesday by Edison Media Research and Mitofsky International, which conducted polls for major news organizations, including CBS News.

The report notes, however, that "there were no incorrect… winner projections" based on the exit polling, and finds "no systematic problem in how the exit poll data were collected and processed."

The study also finds no support for "allegations of fraud due to rigging of election equipment," which arose in a number of states, including the decisive electoral vote state of Ohio.

Exit poll estimates overstated the vote for Kerry in 26 states and for President Bush in four states. The primary reason for this discrepancy, the report concludes, was a greater willingness of Kerry voters to answer interviewers' questions than Bush voters.

"Kerry voters were more likely to participate in the exit polls than Bush voters," the report says.

It attributes the higher response rate of Kerry voters to a number of things including "motivational factors that are impossible to quantify," as well as interactions between respondents and interviewers, restrictions placed on interviewers by state and local election officials, and weather conditions at certain polling spots.

The report suggests several ways to improve future exit polls, including better cooperation with state and local election officials, better interviewer training, and changes in how interviewers are hired and monitored.

The report "is a very blunt examination of the issues involved when you do exit polls," says Kathy Frankovic, director of surveys for CBS News.

The Edison/Mitofsky study also acknowledges that early exit poll estimates on the afternoon of November 2 overstated the proportion of women voters, which led to speculation – particularly on Internet blogs – that Kerry, who did better with women than men, was headed for a victory.

The report attributes that problem to "a programming error" which was corrected before any news organizations broadcast or published exit poll results.

By Joel Roberts