Complex machinery of polls show tight fall races

Andrew Pritchard

A flurry of recent polls shows Minnesota’s gubernatorial and U.S. Senate races are too close to predict, with just over a month remaining for candidates to rally voters.

The most recent survey results released Sunday show U.S. Senate candidate Norm Coleman, the former Republican mayor of St. Paul, leading Democratic incumbent Paul Wellstone 47 percent to 41 percent.

The poll’s margin of error is plus-or-minus 4.5 percent.

The MSNBC/Zogby poll also showed Independence Party candidate Jim Moore with 2 percent support among likely voters and the Green Party’s Ray Tricomo drawing 1 percent.

A Star Tribune Minnesota Poll released last week shows a closer race – a 4 percentage point lead for Wellstone that could be accounted for by the poll’s 3.2 percent margin of error.

The 3 percentage point Wellstone lead found by a Pioneer Press/Minnesota Public Radio poll released last week was also within that poll’s 4 percent margin of error.

Both papers’ polls also show a tight three-way race for governor.

The Star Tribune poll found state Senate Majority Leader Roger Moe, DFL-Erskine, in the lead at 30 percent, with state House Majority Leader Tim Pawlenty, R-Eagan, at 28 percent and Independence Party candidate Tim Penny receiving 25 percent.

These gaps could be accounted for by the survey’s margin of error, as could Penny’s lead in the Pioneer Press/MPR poll.

Thirty percent of the public backed the Humphrey Institute Policy Forum co-director in that poll, while 28 percent supported Pawlenty and 27 percent said they would vote for Moe.

Green Party candidate Ken Pentel received 3 percent support in the Pioneer Press/MPR survey and 2 percent in the Star Tribune Minnesota Poll.

Understanding the polls

a poll’s margin of error is determined by the number of people surveyed, said Larry Harris, a principal at Mason-Dixon Polling and Research, an independent Washington firm that conducts polls for the St. Paul Pioneer Press.

Once a pollster has found the percentage of people surveyed who expressed a particular opinion, the margin of error above and below that number creates a range within which the pollster believes resides the percentage of the entire public holding the opinion the poll measured.

A survey of 500 people has a 4.5 percent margin of error, Harris said. However, doubling the sample to 1,000 people reduces the margin of error only to 3.2 percent.

“It’s not a dollar-in, dollar-out formula,” he said.

A pollster then calculates the survey’s confidence measure – how confident the pollster is that the public’s actual opinion is within the survey’s margin of error.

A typical confidence measure is 95 percent.

“That means that 95 times out of 100, if you sampled the entire universe … the number would be correct within that 4.5 percent margin,” Harris said.

To know if a poll is reliable, Harris said, a person should look first to the polling company’s track record of accuracy and independence.

“Knowing where the pollster was coming from, who paid for it, why the poll was conducted is the first thing to look at,” he said.

A person should then look at how the survey’s questions were worded, Harris said.

Polls determine which respondents are “likely voters” in different ways, and Harris said it’s also important to know what the poll taker said to the respondents before asking the survey questions.

“Bottom line,” he said, “buyer beware, and reader beware.”

Penny tours campuses

penny met with Chancellor Don Sargent and student leaders at the University’s Crookston campus Thursday morning on a swing through northern Minnesota.

Sargent said Penny toured campus while students showed him how several state programs benefited the campus.

“I tried to take advantage of it, and the students did too, to show him what the state of Minnesota does for us as a higher education institution,” Sargent said.

Penny is scheduled to tour the Morris campus and meet with students Friday afternoon.