The first debate of the 2012 general election is over and the verdict is pretty clear.  CNN is reporting that 67% of a sample of 430 poll watchers said Romney won, while just 25% said Obama won.  By Thursday evening, the conventional wisdom in the electorate at large will be that Romney won the first debate.  The pressing question will become:

How will the outcome of the debate impact the candidates' polling numbers?

We'll have to wait a couple days to see, but for now, we can look at some data from past debates and try to create some realistic expectations.

In 2000, the National Annenberg Election Study (NAES) conducted panel interviews for each of the individual debates between Al Gore and George W. Bush.  The second debate, held on October 11, strikes me as a reasonably good analogue.  About one-third of respondents reported watching the entire debate.  Of those, 56.3% said Bush won, while 29.4% said Gore won.  Interestingly, among those who didn't watch the convention at all, 50.4% thought Bush won, while just 23.4% thought Gore won; in the days after the debate, opinion about who won was relatively consistent throughout the electorate.

If we look at pre- and post-debate vote intention, we can see that there was a small but substantively important shift in Bush's favor after the debate:


Vote intention before and after the Oct. 11, 2000 debate (NAES 2000 Debate Panel, N=670)

	Pre-debate vote intention	Post-debate vote intention
Bush	43.3%	46.0%
Undecided	12.5%	9.9%
Gore	44.2%	44.1%

Individual-level analysis suggests that this small increase in support for Bush came from "undecideds" who concluded that Bush won the debate and changed their vote intention accordingly:


Determinants of post-debate vote intention
Ordered probit

	Coefficient (standard error)	Significance
Pre-debate vote intention	1.645 (.104)	.000
Bush won (dichotomous)*	-.547 (.190)	.004
Gore won (dichotomous)*	.178 (.223)	.427

*excluded category:  No clear winner
N=508
Prob > chi2=.000
Pseudo R2=.587


After controlling for pre-debate vote intention, holding the opinion that Bush won the debate moved post-debate vote intention in favor of Bush to a statistically significant degree.  Examination of predicted probabilities shows that "undecided" voters who thought Bush won the debate had a 53.8% chance of switching their vote intention to favor him.

The major difference between this 2000 example and the current situation is that there are far fewer undecided voters in the 2012 electorate (about 6% according to the latest Gallup poll), which would diminish the overall impact of a comparable proportion of undecideds breaking towards Romney.  However, this example demonstrates that perceptions of the debate outcome do have the potential to shift polling numbers when there is a clear winner.




for more, follow me at @JosephCera