Weather Forecast


Methodology of Forum Communications' Senate race poll questioned

FARGO - Saturday's publication of a Forum Communications Co. poll that shows U.S. Rep. Rick Berg leading Democrat Heidi Heitkamp by 10 percentage points - with 10 percent undecided - has some calling foul on the poll's methodology.

Although experts and campaign insiders on both sides of the North Dakota U.S. Senate race agree the independent poll of 500 likely voters is not going to sway undecided voters, they disagree that the polling methodology is an accurate reflection of voter opinion.

"On the whole, it's an unreliable poll from a polling firm that has shown in the past it doesn't follow professional polling standards," said Brandon Lorenz, a spokesman for the Heitkamp campaign.

The Forum Communications poll was completed by Essman Research, a Des Moines, Iowa, firm that polled residents statewide between Oct. 12 and Oct. 15. It used a sampling margin of error of plus or minus 4.4 percent.

The same firm was hired by Forum Communications prior to the June primary elections. The poll - which also showed Berg as a clear favorite - was criticized for polling primary voters about whom they would vote for in the general election.

Of the phone numbers randomly called in the October poll, 75 percent were landline numbers and 25 percent were cellphone numbers.

Mark Jendrysik, University of North Dakota professor of political science and public administration said the polling methodology appears fair, especially with the inclusion of cellphone numbers.

Calling cellphones is important for an accurate population representation, he said.

Retired Minnesota State University Professor and poll analyst Jim Danielson agreed with importance of using cellphone numbers, but cited the poll as not having enough.

"The latest guesstimeate is that 41 percent of adults in North Dakota are cell phone users only. That's their only number, they don't have a landline," he said.

Danielson said young voters, college students, and especially new residents living in western North Dakota are more likely to exclusively have a cellphone and many may not have a 701 area code; excluding the voters from being polled.

Danielson said the response rate of the poll - or how many of those called actually remained on the line to take the poll - is also very important.

"Some polls nationally and statewide have a response rate of 5 percent," Danielson said. "Only 5 percent of numbers you call. That's a very low rate and it's a consequence of the country being over-polled."

The sample seems to under-represent women, Danielson said. The sample reported polling 51 percent of women. He said general polls usually show a female sample of nearly 53 percent.

Danielson said he is reminded that in Heitkamp's 1992 successful bid to become the state attorney general, 30 percent of female voters sided with Heitkamp, the largest gender bias Danielson said he has ever seen.

"That degree of gender favoritism is not working to an advantage for her," Danielson said.

He said that bias could return on Nov. 6.

Berg Spokesman Chris Van Guilder said the Berg campaign is pleased with the results that show what his camp views as a realistic attitude of voters.

"I'll let others debate the merits of the methodology," he said. "We're confident that Rick's message of standing up to Barrack Obama's failed policies and repealing Obamacare and balancing the budget is resonating with voters. We feel good about the support we've received."

This is the first poll to show Berg with such a commanding lead over his challenger. A July Rasmussen poll showed Berg ahead with 49 percent to 40 percent in Heitkamp's favor.

A Mason-Dixon Poll in June showed Heitkamp leading berg 47 percent to 46 percent.

"This is not a poll that we would rely on for how we would conduct our campaign," Lorenz said. "We don't believe this poll reflects the state of the race."

To back up the claim of unfair inclusion in the poll, Lorenz pointed to the 19 percent of those polled identifying themselves as Democratic. Lorenz said the number is too low, considering 2008 exit polls showed 28 percent of the voters in the general election identified themselves as Democrats. He said the discrepancy shows a lack of accurate methodology which would favor Republican Berg.

Thirty-five percent of those polled identified themselves as Republicans, 42 percent claimed to be independents and 2 percent refused to answer the question.

Jendrysik said it's hard to tell if partisan numbers have a great impact on the polling result, especially in a state of unregistered voting.

He said he believes the poll is an example of one well-done.

Neither Danielson nor Jendrysik were surprised to see Berg land on top, but they are cautious of a 10-point lead.

"My guess is Berg's lead is less than 10 percent," Danielson said.

Readers can reach Forum reporter

Wendy Reuer at (701) 241-5530