Monday, November 12, 2012

Post-election: Prediction success, or not

One of the questions I asked before the election was whether polling was still useful. If pollsters' methods could no longer capture what was likely to happen on Election Day, that would signal the loss of an important tool.

Well, we don't have to worry. The pollsters provided helpful information, and the aggregators like Nate Silver used this raw material to create fairly accurate predictions. The best pollster was PPP. Rasmussen and Gallup were near the bottom of the pack, tied at 24th out of 28 (source: Atlanta Journal).

Here's a roundup of interesting commentary on the pollsters:
  • Sizing up Nate Silver's results.
  • Forbes lets UnSkewedPolls' Dean Chambers blather about how he didn't do so badly--getting only 3 or 4 states wrong. Hilarious.
  • LA Times lets the winners and losers explain themselves. Good and concise.
  • Votamatic, an aggregator who made an early and accurate prediction, and also has pretty graphs.
  • Business Insider lists the assumptions that led the unskewers to their blazing wreckage. Short and pithy.
  • HuffPo skewers a few fools too.
  • Atlantic Wire enjoys listing many conservative pundits' predictions, and shouting "WRONG" after each. It is juvenile, but somehow cathartic. (Scroll to the bottom.)


On a personal note: My favorite way to follow the predictions was by reading Meep at the Conservative Commune. I'm disappointed that 1) she didn't follow up on accuracy as she said she would, 2) they don't let me comment at the site anymore. Too dangerous to let an intelligent alternative like me speak there. 

Update on the same day. Here's Michael Barone's prediction. He projects a calmer GOP partisanship than most, but he was just as wrong. 

No comments: