jump to navigation

Experiment on election prediction markets August 5, 2009

Posted by Sverre in : Methods in political science, Political behavior , comments closed

I’ve recently become involved (as a participant) in an interesting experiment performed by PhD student Sveinung Arnesen at the University of Bergen in which we are asked to predict the election result through a market model, buying and selling fictive “shares” in the outcome based on our own evaluations. This is based on prior experiments like Iowa Electronic Markets experiments by the University of Iowa in connection with American Presidential Elections, and the work of Robin Hanson.

Participants have been recruited through the political party organizations (at least I was), and appear to only have the option of buying or selling “stock” in our own party and/or government coalition. I assume part of the reason why we are restricted to our own party is the need for keeping the results secret to avoid incentives for strategic attempts at driving up the predicted value.

(more…)

NYT polling standards March 17, 2009

Posted by Sverre in : Methods in political science , comments closed

I recently found out that the NYT actually have published standards for what polls they are willing to publish. It doesn’t appear to be an entirely new thing (the article is from september 2008), but I found it to be an encouraging surprise. I wish more media were as quality conscious.

The standards document seems to be very basic, but still lay down some important ground rules for minimum requirements for the credibility of a poll. I’d like to see it go into some more detail also on what is required of the questions, but that would of course have to be less concrete and authorative.

10 points to the NYT for a good start.

A tip of the hat to Bård Vegard Solhjell for bringing this to my attention.

Pollster biases revealed November 3, 2008

Posted by Sverre in : Methods in political science, US Presidential election , comments closed

The Monkey Cage led me to a paper by Len Adleman and Mark Schiling that compares the election polls made by American networks. They’ve compared them to the polls made by Gallup And Rasmussen, and show that the political inclination of the networks seem to influence the polls. Fox’s polls show a trend of predicting more to the right, and CBS/NYT more to the left. This is really interesting.

I agree with Andrew Gelman who comments that this surprises him. I would expect their coverage of the polls to show some bias, but had expected the polls themselves to be done in a professional way eliminating personal biases. This apparently goes to show that being completely neutral is difficult if not impossible, even in quantitative analysis.

What if the whole world could vote? October 30, 2008

Posted by Sverre in : Methods in political science, US Presidential election , comments closed

asks the Economist and tests it. They’ve asked their online readers to vote and constructed a worldwide electoral college. Lo and behold! the world electorate map is shockingly enough painted bright blue. It appears most nations in the world have a distribution in excess of 80-20 in Obama’s favour. Some people (no serious political scientists I hope) take this as evidence that the world supports Obama.

Economist.com voter map screenshot

Economist.com voter map screenshot

The world at large probably prefers Obama, but this “survey” does in no way confirm that. Why not? It’s really quite simple. The survey is conducted at the Economist.com website. And who are going to claim that worldwide readers of the Economist represent a fair approximation to a random distribution of the population? None, I hope. For example, I would expect the Economists’ readers to have vastly higher than average levels of education. This is further accentuated by the fact that not every visitor to the website can vote, just registered Economist.com members. This ensures that even those casual visitors not normally reading the Economist are even less likely to vote.

There is no easier way to prove the bias of the survey than just looking at the scores for the US in this survey. In this survey the US supports Obama by 81-19.

This isn’t even very original. I have seen several such maps made on the basis of different surveys already, and must have read a dozen different online articles on it.

So has the Economist suddenly gone naive and stupid? Of course not. This was never intended as a survey by the Economist, so just don’t make the mistake of treating it like one. What the Economist wanted was more hits to their website, and more registered members on their website. I’d guess this got them hundreds, if not thousands. It worked on me. 😉 (more…)