Princeton Election Consortium

Innovations in democracy since 2004

Oct 21: Biden 357 EV (D+5.4% from toss-up), Senate 53 D, 47 R (D+4.8%), House control D+5.0%
Moneyball states: President NV AZ PA, Senate AK MT IA, Legislatures KS TX NC

Is the bounce fading? National polls a tie, 0.0 +/- 1.0%

September 9th, 2008, 9:34pm by Sam Wang

In five polls released today, all spanning September 6-8, the Obama-McCain margin is 0.0 +/- 1.0%, a perfect tie. Is this apparent tightening the end of a bounce – or just random fluctuation?

The five polls used (Pollster/RCP) are:

  • NBC/WSJ (Obama +1%)
  • American Research Group (Obama +1%)
  • Rasmussen Daily Tracker (tie)
  • Diageo/Hotline (McCain +1%)
  • Gallup Daily Tracker (McCain +5%)

Recent Gallup data points have been outliers. They are said to emphasize voter enthusiasm in their likely-voter screen more strongly than other organizations, a plausible reason for the difference. In any event, my use of median-based statistics allows outliers to be used without giving them excessive influence. (Click the link – in Canada even 7th-graders know this! We need more of that here. To learn for yourself see these resources.)

The previous measurement, based on eight weekend polls (September 5-7), was McCain +2.0 +/- 1.2%. Based on these error bars, there’s about a 1 in 10 chance that the 2-point shift is a chance fluctuation. Not enough to reach statistical significance, but it bears watching. If it keeps up, I might be able to stop doing the adjustment to state polls by the end of the week.

When applied as a correction to the state-poll data, the resulting EV estimate is Obama 265 EV, McCain 273 EV. The confidence intervals are Obama [232-296] EV, McCain [242-306] EV.

Tags: 2008 Election

No Comments so far ↓

Like gas stations in rural Texas after 10 pm, comments are closed.