Top polls prove reliable in picking election winners

Several major polling organizations have impressive track records of accurately predicting the outcomes of presidential elections, even though it might have been impossible to predict the chaos of November 2000.

Although the Harris Poll was the lone organization to predict an exact tie in that election, a Zogby poll and one by Gallup/CNN/USA Today both came within two percentage points: Zogby had Al Gore winning, Gallup had George Bush winning, each by two points.

Be it Gallup, Zogby, Harris, Pew Research Center for the People and the Press or any of the myriad major news organization polls, on average they’ve gotten it right within three percentage points of the real election results since 1984.

And, there’s no evidence of any systematic bias toward Republicans or Democrats during that time.

Voters are hesitant to rely on polls, said Harris Poll chairman Humphrey Taylor, which serve as the backbone of most of the punditry filling newspapers and airtime in the lead up to the one poll that actually counts.

“People are skeptical,” Mr. Taylor said. “It’s counterintuitive to many people that you can interview 1,000 people and get an accurate cross-section of the population.”

The main reason that people don’t trust polls, he said, is that “most Democrats talk mostly to Democrats, and Republicans talk mostly to Republicans. … Both sides think the polls are against them.”

Lee Miringoff, president of the National Council on Public Polls (NCPP), said somehow, the polls “are perceived as being something different than collections of people’s attitudes.”

Mr. Miringoff said a similar phenomena can be seen in the way “people watch news and read newspapers, but don’t trust the media.”

“They read their newspaper religiously and they watch their news shows and they participate in polls and are interested in poll results; they just don’t like reporters or pollsters,” he said.

Mr. Miringoff said the general public sees reporters and pollsters as filters that have the potential to somehow skew the truth.

But, he said, there is “no evidence” that polling organizations intentionally do that because “accuracy is what puts food on the table.”

The pre-2000 election polls were the most accurate in recent history, with several organizations predicting that Mr. Gore and Mr. Bush would end up within two percentage points of each other.

By contrast, 1996 was one of the least accurate. Although every major poll predicted that Bill Clinton would defeat Bob Dole, most were off by more than three percentage points in predicting how much of the vote each candidate would get.

Meanwhile, 1996 and 2000 saw an increase in the number of major polls conducted, with nine in 1996 and 10 in 2000, according to the NCPP.

Story Continues →