Princeton Election Consortium

A first draft of electoral history. Since 2004

A look back at poll aggregation, 2004-2014

May 6th, 2015, 1:11pm by Sam Wang


My article on Presidential poll aggregation is now published, in the International Journal of Forecasting. You can read it here. It’s part of a special issue on Presidential forecasting; when I have the other articles I will link those as well. Read about the origins of a rather odd hobby!

Tags: 2004 Election · 2006 Elections · 2008 Election · 2012 Election · 2014 Election

6 Comments so far ↓

  • Matt McIrvin

    That’s an excellent summary of the action over the past few election cycles. I’ll probably send people to it as a reference on the subject.

    Drew Linzer also did remarkably well at calling the 2014 midterm, if I recall correctly. I’m going to be closely watching whatever he does over the coming year.

  • Amitabh Lath

    You mentioned Bayesian predictors, specifically Linzer’s Votamatic. One feature of Votamatic is the almost eerie lack of movement. There may be some scatter at the state level but his Obama vs. Romney prediction did not change for several months before the election.

    The PEC, by contrast, is full of cliff-edges, such as the one after the first Obama-Romney debate.

    Given that Votamatic and PEC both nailed the EV, how significant are the movements of the PEC? Did Romney really have an even chance after that debate?

  • Amitabh Lath

    This is going to do wonders for your h-index.

  • David D.

    I think there’s a typo in the link titled ‘here’: surely it should be http://math.princeton.edu/~sswang/wang15_IJF_origins-of-poll-aggregation.pdf?