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

National update: Obama ahead by 5-8%, with a post-convention bounce of 3-4% (and its effect on EV estimate)

September 1st, 2008, 11:24pm by Sam Wang

It’s now possible to examine the Democratic post-convention bounce. I define the bounce as the shift in the Obama-McCain margin as of this weekend compared with after August 21st (the date of McCain’s number-of-houses gaffe and the “Seven” ad).

I have four matched comparisons in which an organization polled both soon after the 21st and again over Labor Day weekend. A fifth comparison (CBS) uses data from 8/15-19. (Update: a sixth poll from Diageo/Hotline) The results are…

Using the median and median-based error bar in the way that I do for the Meta-Analysis, the current national margin is Obama over McCain by 6.5 +/- 1.3% (range 1 to 9%). The size of the bounce is Obama +3.5 +/- 0.9% (range 0 to 5%).

The data sources give margins (Obama minus McCain) of

Diageo/Hotline: Obama +9% (from +4% 8/18-24)
CBS: Obama +8% (from +3% 8/15-19) (story and PDF)
USA Today/Gallup: Obama +7% (from +4% 8/21-23)
Gallup Daily Tracker: Obama +6% (from +2% 8/22-24)
Rasmussen: Obama +3% (unchanged from 8/24-25)
CNN: Obama +1% (from a tie one week earlier)

This national bounce can be applied as an adjustment to the current Meta-Analysis, which currently uses pre-Democratic convention state polls. I’m doing it just this once because the Republican convention has already started, and I wanted to get a quick EV snapshot.

If the bounce is taken into account, the resulting EV estimate would be Obama 343 EV, McCain 195 EV. The 95% confidence interval is Obama 324-362 EV, McCain 176-214 EV.

It is interesting to examine the Gallup Daily, which shows the largest swing between the 8/23-25 sample and the 8/24-26 sample. The second sample picks up people surveyed on August 26th, who were called after Michelle Obama’s 8/25 speech, and a mix of before and after Hillary Clinton’s 8/26 speech. It appears that some combination of these two speeches may have been very effective in moving public opinion.

Tags: 2006 Elections · Site News · Uncategorized

2 Comments so far ↓

  • Jack

    Mr. Wang: So the needle on the EV Estimator should go up to around 343? That would be a huge bounce and would bring Obama to an all-time high. But it’s now 9/4 and the needle has not moved appreciably since after the “seven” ads. Que pasa? Is it because we’ve go no new state polling completed since the Dem convention?

  • Sam Wang

    The point of this post is that we can do the Meta-Analysis using state polls alone, but they are out of date. In order to get a more current estimate, one can use national polling data to estimate how far the Electoral College snapshot would move if we had current state polls.

    So the answer to your question is yes, but it’s the premise of the post, not a consequence.