Princeton Election Consortium

Innovations in democracy since 2004

Outcome: Biden 306 EV (D+1.2% from toss-up), Senate 50 D (D+1.0%)
Nov 3 polls: Biden 342 EV (D+5.3%), Senate 50-55 D (D+3.9%), House control D+4.6%
Moneyball states: President AZ NE-2 NV, Senate MT ME AK, Legislatures KS TX NC

On the wisdom of crowds of pollsters

October 31st, 2010, 11:22am by Sam Wang

Last-minute polls will be available tomorrow. The final Senate prediction will move (indeed it already has). Also, I haven’t applied variance minimization (VM) yet for tiebreakers. In the meantime…

I should express the idea of the previous post more carefully. Let us pose a hypothesis: Pollsters sample voters with no average bias. Their errors are small enough that in large numbers, their accuracy approaches perfect sampling of real voting. This hypothesis held up well in 2004, 2006, and 2008.

The 2010 test occurs on Tuesday. Polls predict an outcome (95% confidence) of a House Republican majority of 227-233 seats. For the Senate Democrats it’s 50-52 seats (assuming Lieberman continues to caucus with them). If either outcome falls outside this range, it’s time to re-examine the hypothesis.

Tags: Politics

2 Comments so far ↓

  • PolyisTCOandbanned

    Think the House numbers will be the bigger test for you. Seems like everyone’s Senate numbers are very similar, but there has been a wild array of House numbers estimated.

    • Sam Wang

      Not entirely true. Firstly, most claim a Republican House and a Democratic Senate. However, I think the stated probabilities by others are too low – in my view it’s a lock. Strictly speaking, the person who claims the greatest certainty should get the most credit. But I understand that’s not how most people think.

      Secondly, my Senate estimate is more than 1 seat less than other estimates. Proportionally, this is a larger difference than the differences among House models.

      In any event, my November 1 posting gives an estimate and confidence intervals on House numbers. Anything outside the confidence interval is grounds for asking more questions. Nobody else is using confidence intervals so it’s impossible to tell whether they are “right” or “wrong.” Note that I am posing this year’s calculation as a test of a hypothesis that passed in 2004, 2006, and 2008. If you think it will be different this year…well, check in on Wednesday.

Leave a Comment