Why did polls give Hatch the edge?

11/11/2006

Republicans allege that polls are flawed, with Democrats being favored by the pollsters and Republicans more able to avoid being polled.

Sharon Schmickle, Star Tribune
Last update: November 09, 2006 – 12:07 AM

Eight public opinion polls by different organizations gave DFLer Mike Hatch a lead in Minnesota’s race for governor during the final two weeks of the campaign.

Hatch lost. What happened?

The answer is a study in the controversial science of polling as it applied in Election 2006.

Nationwide, polls signaled accurately that Democrats were heading toward a victory that could dramatically reorder the political landscape, said Mark Blumenthal, publisher of Pollster.com, which tracked polls for 80 key races. In almost all cases, he said, polls gave an accurate picture of “who was clearly ahead, who was behind and which races were going to be close.”

The same could be said for the Minnesota governor’s race, which ended in a near tie with a slight edge for Republican incumbent Tim Pawlenty. The polls reported that Hatch’s “lead” generally was within the margin of error.

But the fact that it was reflected so many times suggested it was real. Think of measuring a tied race as akin to flipping a coin. Chances of getting heads (or in this case, Hatch) on a given flip ought to be 50-50. Getting heads eight times in a row would be extremely unlikely.

While most polls suggested that the race was tightening in its final days, not one showed Pawlenty leading.

One possible explanation is that the polls were flawed. State Republican Party chairman Ron Carey said that pollsters favor the DFL in the technical adjustments they make to identify likely voters. He also said that Republicans are more likely to use caller ID and other technology that puts them outside a pollster’s reach.

But the same polls accurately reflected Minnesota’s U.S. Senate race, if anything over-estimating support for the Republican who lost. Same samples of likely voters, same interviewers, same number-crunchers as were involved in the governor’s race polls.

“If there were bias in one race, it should have been reflected in the other,” said Lawrence Jacobs, who directs the Center for the Study of Politics and Governance at the University of Minnesota’s Humphrey Institute. One of the polls in question came from the center.

Another possibility is that something changed in the homestretch to tip the balance toward Pawlenty. Carey said the GOP’s private polls also showed Hatch in a down-to-the-wire lead, but with a narrower margin than the public polls reported.

“If the election had been held a week earlier, Mike Hatch may very well have won,” Carey said.

One last-minute wild card involved fence-sitters --undecideds and independents who appear to have tilted Democrat in other races. In Minnesota’s U.S. Senate race, voters who told exit pollers they were independent went overwhelmingly for Democrat Amy Klobuchar, helping propel her to victory.

But 54 percent of Minnesota’s independents said in the exit poll that they voted for Pawlenty or Independence Party candidate Peter Hutchinson.

The reason may have been missteps that Hatch and his running mate, Judi Dutcher, committed a few days before the election, said Rob Daves, who directs the Star Tribune Minnesota Poll. Hatch blew up at a reporter and Dutcher revealed a lack of knowledge about E85, an ethanol blend that is important to rural voters.

“I believe that is what did the trick for Pawlenty,” Daves said.

But why did polls that went on through the weekend miss the turn in attitude? Unless an event is seismic, it tends to settle gradually with voters, pollsters say. Some miss the first reports and catch the news later. Some hear about it from friends.

“There are very few examples of stories that are so huge they turn everything upside down immediately,” Blumenthal said.

When a fast-moving trend hits the final hours of a tossup race, he said, “there is no political survey I’m aware of that’s capable of forecasting the winner.”