Introducing a New Open Collection of Danish Polls

Display image for Niclas Darville

Today, I think my collection of Danish opinion polls are mature enough to share with the public. They can be found on this site.

The political debate in Denmark is not doing well; pundits and talking heads take up all TV, print, and radio.

The problem with this model is that many of those people aren’t basing their comments and analyses on any solid foundation centred around a proven, scientific model. They are not what people in the U.S. would call “wonks”—political-science nerds.

Another fundamental problem is that the media and punditry don’t know how an opinion poll actually works. A poll is a snapshot with a statistical uncertainty. Most “movements in the polls” are results of short-lived moods and a statistical uncertainty that can make parties vacillate in polls.

The New York Times’s wonk vertical, The Upshot, explained the consequence of the statistical uncertainty brilliantly in this article.

As such, it is a problem that the media don’t get that you need to analyze the development in Danish politics based on a trend, and not by going from one poll to the next. You’ll keep seeing polls presented in the media as a bar chart—not a trend line over time. A bar chart at that, which doesn’t display the statistical uncertainty either.

As of this writing, a panic has seized the Danish media world and British politics, because YouGov presented a poll showing more Yes votes than No votes to Scottish independence. Meanwhile, Sam Wang is estimating the probability of a Scottish No to be 95%—based on an aggregate of polls.

There are several reasons why the American development towards a scientifically-informed political discourse didn’t come from the major new organisations, but from individuals with an academic background. One reason is that the media in Denmark tend to collaborate with a single polling institute; e.g. DR with Epinion and TV 2 with Megafon. It constrains them to basing their coverage on their own polling institute; meanwhile, the rest of us are free to include all public polls and draw a trend from them, which yields a more accurate picture:

Trend line

We all deserve to participate in the political discussion, and by making polling data public, we can contribute to a salience sorely missed in the Danish coverage of politics.

So start using the polls here.

American Examples

Swedish Examples

Great Books

  • The Gamble (free, downloadable chapters here)