September 10, 2004
FAKING IT:
State GOP lashes out at Minnesota Poll (Patricia Lopez, September 11, 2004, Star Tribune)
The state Republican Party has called on the Star Tribune to fire its longtime pollster, Rob Daves, saying that the paper has consistently underestimated Republican turnout since he was hired in 1987.State Party Chairman Ron Eibensteiner, at a news conference at party headquarters this morning, said that "the final poll results for Republican candidates have been, on average, 5.2 percent under the actual results of the election. "The inaccuracy of the Star Tribune's Minnesota Poll, he said, "is a direct result of their pollster's insistence upon underrepresenting Republican voters in his samples." [...]
Displaying a chart that showed poll results going back to the Minnesota Poll's inception in 1944, Eibensteiner said that particularly since 1998, the poll has demonstrated a pattern of underestimating Republican presence in the state.
That year, the final poll showed then-Republican gubernatorial candidate Norm Coleman with 30 percent of the vote and Democratic challenger Hubert Humphrey III with 35 percent. On Election Day, Coleman had 34 percent, Humphrey 28 percent and the winner was Independent Jesse Ventura.
In the 2000 presidential race, the Minnesota Poll showed Republicans George W. Bush and U.S. Sen. Rod Grams each with 37 percent. On Election Day, Bush took 45.5 percent of the vote, Grams 43.29 percent. In the 2002 gubernatorial race, the poll showed Republican Tim Pawlenty with 35 percent. His actual share of the vote turned out to be 44 percent.
Polls are not intended as predictors of election results, only snapshots in time, but Eibensteiner said the Minnesota Poll had so consistently shown Republican support at less than what it later turned out to be that it was the party's conclusion that the poll's methodology was flawed.
"When a newspaper conducts a poll on a political race," Eibensteiner said, "...it is doing more than reporting the news. It is also creating the news ... and could ultimately impact the outcome of an election."
This kind of thing probably explains a good deal of the discrepancy in current national polls too. Posted by Orrin Judd at September 10, 2004 3:38 PM
Mixed blessing. Granted you want the public to fall in behind the winner and a momentum to develop for your party, but I personally get a kick out of sneaking up on over confident Democrats and crushing them on Election Day.
Posted by: h-man at September 10, 2004 3:45 PMNational pollsters at least have their big media clients to satisfy come Election Day -- satisfy in this case meaning, "don't embarass" -- so their final polls Halloween weekend will have to come within some reasonable distance of the truth, in order to retain their big-money contracts for the next election cycle. But before then they can tweak the polls questions and/or survey pool any way they want to nudge the results this way or that.
On the other hamd, companies that focus on state polling are fewer in number and their clients aren't as powerful. So outside of the big media markets like California and New York, they don't have to be held to the same level of accuracy, which may explain Minnesota's polling problems.
I'd be reluctant to indict the entire polling effort just because of one little slip up.
Posted by: luciferous at September 10, 2004 4:46 PMH-man - I disagree. I agree with OJ that polls can be used to shape opinion and a campaign race. People who don't pay close attention can see a poll and say "this guy is winning big - I'll go for him/her" when in fact the race is deadlocked.
Posted by: AWW at September 10, 2004 11:18 PMAww
I defer to the expertise of OJ and yourself. I still love the "shock and awe" of an upset.
Posted by: h-man at September 11, 2004 5:29 AMThe Red Star is about as trustworthy as the Brezhnev era Pravda. The reality is that everybody in Minnesota knows it and discounts its claims accordingly.
Posted by: Bart at September 11, 2004 7:18 AM