9 October 2018 — 1609 mdt
The Big Sky Poll was conducted in August —
but its results weren’t released until yesterday
The Big Sky Poll is a teaching project administered by two University of Montana professors and executed with student labor. From 13 through 31 August the poll interviewed 618 registered voters in Montana. The sample for the U.S. Senate and House horserace questions was 466, and has a margin of error of 4.5 percent. The sample was weighted by geography and gender.
The poll reported Sen. Jon Tester was leading his Republican opponent, Matt Rosendale by 24 points, 56 to 32 percent, with two percent preferring Libertarian Rick Breckenridge and nine percent undecided. I consider that 24-point lead highly improbable.
That was approximately six weeks ago. Since then, as displayed in the graph below, five professional polls have been conducted. None reports such a whopping lead for Tester.
I have two problems with the Big Sky Poll. First, the results were released on academic time, not political time, and are too late to be of much use for reporters and political activists. Second, the weighting is rudimentary, not on a par with the weighting of professional polls. That widens the error bars considerably.
Although the average of the last five polls has Tester leading by 4.4 points, none of these polls was conducted after the Kavanaugh nomination donnybrook. We therefore don’t know what effect that had on the election, and won’t until polls taken after the festivities concluded are released.