By now the public polling done in North Dakota’s election races – results released by SayAnythingBlog.com/Valley News Life/DFM Research and Forum Communications/University of North Dakota – is old news. But one lingering issue is how far off the Forum/UND polling seemed to be.
That polling was widely derided by political observers – including this one – as just not passing the smell test. The Forum/UND polling showed Republicans with double-digit leads so massive that not even the Republicans I spoke to were buying it.
So what happened?
Well, two Fridays ago I was guest hosting the Jay Thomas Show on WDAY and I had on Dean Mitchell from DFM Research and Bo Wood from the University of North Dakota to talk about the respective polling results, and we figured out what the problem was.
Wood pointed out that the results for Republican candidates were even – almost identical in fact – across the board between the two polls, but the UND/Forum poll showed lower support for Democrats and a lot more undecideds (see the table of results for the three most competitive races on the statewide ballot to the right). So what could have caused that very specific difference between the two polls?
It turns out the difference was in how the polling respondents were asked to answer the questions. The SAB/VNL/DFM Research polling didn’t give respondents “undecided” as a choice. They were only marked down as undecided if they offered that choice.
Wood said the UND/Forum poll did offer “undecided” as a choice for respondents, and as a result got a lot more people saying they were undecided.
Mitchell said – and I agree – that it’s important to present polling questions in a manner that is as close to what voters will see on the ballot as possible. Since the ballot doesn’t have an “undecided” option, our polling didn’t give respondents that option. The UND/Forum polling did, and that ended up skewing the results.
But it’s very interesting that the skew was so heavily against Democrats. It seems a lot of Democrat-leaning North Dakotans, if given the choice, would rather say they’re “undecided” than say they’re supporting a Democrat.
That might be a product of North Dakota’s large Republican advantage. Democrats here are a little shy about identifying themselves as Democrats.
Some might be tempted to see this “break” in undecideds between the two polls heavily favor Democrats, but I think it’s fair to say that the rest of the undecideds – the “true” undecideds, I think – aren’t going to break that way.
If we assume that this polling are were things stand on the eve of the election – remember, these polls are weeks old now – Republicans will likely win benefitting from most of the undecideds breaking their way in this Republican state voting in a year favoring Republicans.