Don’t take the DNT online polls to the bank

Rob Karwath, The Duluth Trib’s editor, wrote a column on Saturday downplaying the validity of their online polls. Specificially he was referring to a poll in which 65% of the participants said they favored the passage of Mike Jaros’ bill to require a referendum on the Red Plan. Its no longer on their website but I believe about 650 people voted – over twice as many as Johnson Controls surveyed last year.

None of the heavy hitters for the District seem to want to know what Duluth’s voters think about the Red Plan. The Board has nixed a vote two years running. It has sought to prevent a non binding referendum from taking place after rejecting Let Duluth Vote’s petition. A professional survey would only cost about $3,000. This year they have spent three times that much on advertising in the Trib alone.

The Trib’s reporter assures me that under no circumstances would the Editorial staff allow their coverage of the story be affected by advertising revenue.

Tell that to the folks who watch Rupert Murdoch’s Fox News.

I’m not sure to which other polls he was referring. [ here and here ]But when it comes to results of our polls at duluthnewstribune.com, the simple answer is: Don’t take them to the bank.

Our online polls, like those of most other Web sites, are far from scientific. Readers who respond vote voluntarily. And the sample size can vary from a few dozen to thousands on a hot subject.

They’re not the kind of polls you occasionally see in our paper that we take pains to make statistically significant. That means ensuring a random sample. It also means collecting enough respondents to say the results would be the same, within an acceptable margin of error, as if we had asked everyone in town.

About the author