The First Debate: Who Won?

Murphy’s Law might explain why a week after starting a blog, my home DSL and phone service has been out for three days running (perhaps that should be Verizon’s law?).  As such, I’m a bit behind schedule posting these thoughts on the polls on last night’s debate.  Here’s my take.
 
Determining who “won” the debate is often a matter of definition.  I’d argue there are four criteria:
1) Who did viewers perceive to be the “winner” (an assessment that requires looking separately at the views of each candidate’s supporters)? 
2) Which candidate did the most to improve the way voters perceive him, regardless of the initial impact on the vote?
3) Did the debate change the vote preference of those who watched?
4) Will the debate change the dynamics of the race in a way that will cause vote preference to change over the next week?
 
The “quickie” polls done last night can tell us a lot about the first two and a little about the third.   The fourth is for now, purely a matter of conjecture.  (Much more after the jump)

First, were the surveys released last night by ABC, CBS and Gallup/CNN/USA Today reliable and worthy of our attention? Wisely, none of the pollsters involved tried to reach a representative sample of the entire electorate last night. All three tried to survey just debate watchers, and for that task their designs were reasonably reliable and trustworthy under the circumstances.*

The surveys done by ABC and Gallup used a very similar design. Both called back respondents they had previously surveyed and re-interviewed only those who watched the debate. ABC reached 531 debate watchers out of 969 registered voters originally interviewed September 23-29. Gallup reached 615 debate watchers out of a pool of registered voters first contacted September 28-29 (Gallup did not report the size of the initial sample).

CBS again used a unique Internet based methodology, only this time with a twist: They interviewed only debate viewers not firmly committed to either candidate. The CBS methodology used a “panel” of Americans recruited to periodically take surveys over the Internet. Unlike most Internet based research, the CBS panel (maintained by Knowledge Networks, Inc.) is arguably representative or “projective” of the US population because it recruits respondents by traditional RDD sampling and provides free Web TV to those respondents who lack Internet access. According to their release, CBS was able to identify 209 “uncommitted voters – voters who don’t yet know who they will vote for, or who have chosen a candidate but may still change their minds – who have agreed to watch the debate” (emphasis added).

So what do these polls tell us about my four criteria?

1) Who did viewers perceive to be the winner? – As has been reported extensively, all three showed that Kerry was perceived as the winner, especially when you consider Kerry and Bush supporters separately. Thus, on the ABC poll, 89% of Kerry supporters thought their man won, compared to 70% of Bush supporters who gave the nod to the President (overall, Kerry “won” 45%-36%). On the Gallup survey, 93% of Kerry voters thought Kerry the winner, compared to 63% of Bush voters who judged Bush the winner (overall, Kerry “won” 53%-37%). The CBS survey of uncommitted voters rated Kerry the winner by a 43% to 28% margin.

2) Which candidate did the most to improve the way voters perceive him, regardless of the initial impact on the vote? Here the surveys again show Kerry at a clear advantage. On the Gallup survey, 46% said their opinion of Kerry grew “more favorable” as a result of the debate, only 13% less favorable. For President Bush, 21% were more favorable, only 17% less. The useful USA Today release shows that Kerry’s 46% “more favorable” number is higher than for any candidate after the 2000 or 1992 debates except Ross Perot in 1992.

The CBS poll of uncommitted voters showed essentially the same result: More than half (53%) said their opinion of Kerry had “changed for the better” as a result of the debate, only 14% had changed their opinion “for the worse.” For President Bush, 22% had changed for the better, 20% for the worse.

3) Did the debate change the vote preference of those who watched? Here the results show no clear significant change, at least for now, among those who watched the debate.

The ABC survey showed no real change: Each candidate gained a point, with Bush holding the same four point lead with the debate watchers after the debate (51% to 47%) that he had held before (50% to 46%). On the CBS survey of uncertain voters, both Kerry and Bush gained ground. Before the debate Kerry had a 28% to 19% advantage; after Kerry gained 10 points and Bush 12, narrowing the Kerry slightly to 38% to 31%.

Gallup did not ask about vote preference after the debate, though they showed no change on questions of “who would better handle the situation in Iraq” (Bush preferred 54%-40% before and 54%-43% after) and “who do you trust more to handle the responsibilities of commander in chief ” (Bush preferred 55%-42% before and 54%-44% after)

4) Will the debate change the dynamics of the race in a way that will cause vote preference to change over the next week?

This question is ultimately most important, as the initial reaction of debate viewers is typically less important than the coverage that follows. Consider the first debate in 2000. Narrow pluralities gave Al Gore the nod as the “winner” (see the historical data here), yet ensuing coverage of his audible sighs and factual missteps took him from a two point lead (48%-46%) just before the debate to a three point deficit afterwards (45% to 48% on the ABC/Washington Post surveys). In the Annenberg tracking study, the percentage rating Gore as “honest” fell from roughly 57% to 50% over the same period.

Obviously the impact of this debate on the larger campaign dynamics, for now, a matter of hot debate and speculation. I’ll leave that to others for now, but let me offer one point for consideration: Kerry has been hampered in part by what the Post’s Morin & Deane described yesterday as an “Enthusiasm Gap” seen in most of the pre-debate surveys. In the ABC/Post poll, for example, 65% of Bush supporters were “very enthusiastic” about their candidate on the last ABC/Washington Post survey, compared 42% of Kerry supporters. Most of the stories I have seen this morning play the debate as a Kerry clear win, a trend that will likely help re-energize the Democratic base. Such a result – if we can go back to methodology for a moment – may also influence who gets through the “likely voter” screens in the Kerry’s favor.

To sum up: Kerry clearly “won” in terms of improved perceptions among debate viewers, but the jury is till out on the eventual effect on the horserace both among debate viewers and the larger electorate.

*Some raised valid questions here yesterday about the drawbacks of these “panel-back” surveys involving multiple interviews. All respondents interviewed last night were interviewed a few days before the debate, which theoretically may have had some influence or bias on their view of the debate itself. I believe that risk is minimal and an acceptable compromise given the time constraints involved a post-debate survey. As always, I encourage those with conflicting views to post them in the comments section.