Click on the years to see earlier postings and news clips. Numbers in [square brackets] indicate the 68% confidence interval for the Democratic-side outcome.

2004 (original site):
Final polls – Bush 286 EV, Kerry 252 EV.
Final outcome – Bush 286 EV, Kerry 252 EV.

2006:
Senate polls – 50.5 D/I, 49.5 R (50% chance of Democratic takeover).
Final outcome – 51 D/I, 49 R.

House polls – 234.5 D, 200.5 R, [232,237].
Final outcome – 233 D, 202 R.

2008 (results pending a few remaining results):
Final single-day snapshot – Obama 352 EV, McCain 186 EV. My prediction: Obama 364 EV, McCain 174 EV, [352, 367]. Popular vote 53.0% to 46.0%, [52.5,53.5]%.
Outcome – Obama 365 EV, McCain 173 EV. Popular vote 52.7% to 45.9%.

Senate polls – 58 D/I, 42 R (16% chance of a 59-41 split).
Outcome – 58 D/I, 41 R, MN recount pending.

House polls – 257 D, 178 R, [254,260].
Outcome – 257 D, 178 R.

Remaining updates for 2008 are available from CNN and DailyKos.

Where comparisons exist, meta-analysis by straightforward reduction of poll data outperformed econometric predictive models, poll aggregators (electoral-vote.com, the 3BlueDudes average, and FiveThirtyEight), and the InTrade electronic market. Overall, the results support the conclusion that the collective “market” of pollster methods contains sufficient information to converge extremely well on final outcomes. Adjustments based on undecided-voter assignment, voter demographics, and pollster reliability have the effect of reducing accuracy.

I’m about to fade back into academic mode, which means reduced posting. I hope to write up this whole project for a regular peer-reviewed publication. In the meantime, for one last bit of bloggy goodness, including some pretty, um, robust comments, let’s look at the mailbag.

First let’s review what a jerseyvote is. In 2004, I said that in a close election, resources were best deployed in states where the outcome is uncertain, with win probabilities between 20% and 80%. That campaign was quite closely fought, as evidenced by a plot of median EV estimate:

Meta-analytic EV estimator for 2004 Kerry-Bush race

For most of the campaign, neither Kerry or Bush spent much time outside the 95% confidence band. The exceptions were a few weeks in August (when Kerry was definitively ahead, between the Democratic convention and the Swift Boat Veterans for Truth attacks) and September (when Bush was definitively ahead, between the Republican convention and the first debate). Therefore both candidates had a vital interest in the same battleground states, any of which could be critical to their efforts.

I pointed out that the disparity of influence by different voters could be quantified using the Meta-Analysis. I defined the “jerseyvote,” a measure of the power of an individual voter on influencing the election outcome, relative to a voter in New Jersey (me, for instance). By this measure voters in Pennsylvania, Ohio, and Florida had thousands of times as much power as I did. This suggested that activists should try to get out the vote in those states, and campaigns should advertise there – which they did.

This year, the dynamics are different. Until recently, Obama has spent the post-primary season clearly ahead of McCain:

Meta-analytic EV estimator for Obama-McCain race

For most of the last two months, Obama’s 95% confidence interval has been entirely above the magic 269 EV threshold. For this reason I had to redefine voter power since when the race is not close, individual voters don’t have much power anywhere. I redefined it as the amount of power that voters have if the race swings sufficiently to make either candidate’s win probability 50%. The results are in the right sidebar. You’ll see the expected suspects: Ohio, Virginia, New Hampshire, Nevada, and so on. These match the places where McCain is spending the most money. What this tells me is that the McCain campaign is assuming that if they win, it will be tight. Therefore they are running a “Kerry/Bush” campaign, in which the goal is to get just over the threshold. They are not expecting a blowout; they just want to win.

In this same light the Obama campaign’s spending suggests a more optimistic view. Their pattern of expenditures makes sense if they are assuming that they have many opportunities to win. In this scenario, voters in many states can potentially contribute to an electoral victory, and are worth pursuing. This pattern of behavior is seen in candidates who expect substantial victories; they will even go so far as to campaign to help downticket races, in the hopes of winning more Congressional seats to build a larger caucus on their side.

Last week the fact that my jerseyvotes calculation (as previously defined) identified non-battleground states seemed like an error. But it gave the expected results for a race that is less close – and evidently that’s what the Obama strategists are expecting, despite Obama’s recent slippage to near-parity. In summary, the Obama campaign’s spending makes sense if they believe that the recent decline in their candidate’s fortunes is transient, and the advantage they enjoyed in July will return in the fall. To put it another way, they are acting as if they have enough eggs to put into multiple baskets.

The Fox News appearance was an interesting dip into the world of punditry and the right wing. The producer, who arranged the interview, was pleasant but showed remarkably little knowledge of the statistical analysis. As far as I could tell, she was going on the strength of the fact that it had made the front page of the Wall Street Journal. I sent graphics of electoral maps and of the jerseyvotes calculation, which were promptly lost in the mad shuffle before going on air.

On the day of the interview a car (driven by a Russian neoconservative who wanted to retire to Israel) brought me to the News Corporation headquarters in midtown Manhattan. In the Green Room were three Halloween pumpkins. One was carved in an intricate pattern that could have been Ann Coulter. The other two were painted, one with Bush (”Four More Years!” “My Fellow Americans…”) and one with Kerry (”They Call Him Flipper,” “I voted for the $87 billion…”). Fair and balanced. Hair/makeup by two terrifically nice women, whose schedule included Morton Kondracke, Fred Barnes, and Rich Lowry – the usual suspects.

The appearance itself was remarkably short. For those of you who have seen the piece ([2 MB version] [30 MB]), I was preceded by an insightful blurb for a Web site called zombiepumpkins.com. Then I gave a description of the meta-analysis and spit out the raw poll result (Bush ahead in EV) and the undecideds-allocated result (Kerry ahead in EV). The URL of this site was not flashed on the screen, even though this had been promised. And then we cut to Tom Ridge saying we weren’t going to condition orange.

Marginal notes (none surprising): The sound stage is quite ratty in person. Same, to a lesser extent, for the interviewers. The on-set laptops really work, and give semi-useful information. Finally, I found cable “news” television to be remarkably shallow, though more driven by print media than I expected, which is good.