Why hold debates at all? Several students have asked me this in light of my gentle chiding of the pundits for their predilection to declare a victor in the debates on the basis of dubious evidence (see, for example, the remarks of FiveThirtyEight’s Nate Silver here or those at the Daily Kos here.) By now, I trust, you have read enough of my posts to understand why the partisan-colored debate reaction is misleading: instant polling results typically reflect the underlying partisan composition of those polled because most people “score” the debate through their own partisan predispositions. It is for this reason that debates rarely, by themselves, change anyone’s vote. For example, the CBS Knowledge Poll post-debate instapoll had Obama winning the most recent presidential debate, 40%-26%, over McCain, with 34% calling it a tie. Putting aside for the moment that the “fine print” in the survey data indicates that an unspecified number of these “uncommitted” voters actually have expressed a preference for one of the two candidates, the results – which were widely reported in the press – obscure the more important fact that the debate failed to persuade the vast majority of viewers to support either candidate. Buried at the very end of the CBS story (literally the last paragraph!) was this: “Immediately after the debate, 15 percent of them said they are now committed to Obama, and 12 percent are now committed to McCain. But most – 72 percent – remain uncommitted.” In raw numbers, that means of the 516 “uncommitted” viewers polled by CBS, 77 were moved by the debate to support Obama, 62 decided to back McCain, but 372 remain undecided! Assuming that the composition of the viewing audience mirrors that reported in CNN, with Democrats slightly outnumbering Republicans, and factoring in the margin of error with the sample, we can safely conclude that the debate had almost no measurable national impact in terms of changing the balance of support for either candidate. Presumably, that’s the statistic and related point that should have led the discussion in the Daily Kos, or in Nate Silver’s FiveThirtyEight, but both sites conveniently forgot to even address the methodological or substantive issues I raise here. One can forgive them, of course, because neither presumes to provide an objective analysis of the election – they are partisans who are advocating for a candidate (which is why I don’t link to their sites on my blog and why I urge caution for anyone who reads their election analyses.) But CBS is supposed to be a nonpartisan source, and it is simply bad journalism not to lead with this statistic, never mind burying it at the bottom of the story. (Let’s be clear here: the title of the CBS article is: “CBS Poll: Uncommitted Voters Favor Obama.” It’s not clear to me that this is even factually accurate, never mind a serious distortion of the poll results. Here’s the link. )
The relatively inconsequential impact, in terms of changing votes, of the three debates so far raises a more important issue than the well-documented partisan bias of the blogosphere: why hold debates at all? Are they simply meaningless exercises? Not at all. The key to understanding the significance of debates is to view them as one piece of a larger mosaic that each voter must piece together before deciding how to vote on November 4. That is, rather than a single event, to be won or lost by the candidates, debates are better understood as part of an ongoing process by which voters both become activated to pay attention to the campaign and better informed regarding the relative merits of the two candidates. In this light, candidates aren’t trying to “win the debate” – they are using it as a forum in their continual effort to frame the campaign in a way that plays to their political interest. Pundits may make a great deal of Obama’s “cool” demeanor, McCain’s reference to Obama as “that one”, or his failure to look at Obama during the debate, but there’s not much evidence that undecided voters care a whit about these issues. Instead, they are fitting the candidates’ debate performance into an ongoing internal dialogue that is based on the myriad sources of information each voter is exposed to during the course of the campaign. Viewed in this light, debates need to be assessed not in terms of who “won” or “lost” the individual event, but rather how well each candidate did in integrating his performance into this larger campaign strategy. That means, in particular, whether the candidate effectively contributed to their ongoing effort to frame the fundamentals that will determine the presidential vote. The fact that over 90% of Republicans watching the vice presidential debate thought Sarah Palin cleaned Biden’s clock, while over 90% of Democrats believed it was Biden that did the cleaning, doesn’t mean the debate was a useless exercise. There remain a vast number of voters – literally millions – who are only now beginning to pay attention to the race. The debates are an important source of information – but only one source – that these uncommitted voters draw on as they begin to think about how to vote, and that committed partisans use to reevaluate their support and/or to decide whether to vote at all.
Debates matter – but not in the way that the punditocracy, with their focus on winners and losers – suggests. I don’t expect Nate Silver or the denizens at the Daily Kos to understand this. But you should.
Does the idea that debates create an opportunity for unscripted lasting images (Dan Quayle: “You are no Jack Kennedy”, Clinton “I feel your pain”, GHWB checking his watch, Al Gore “Lockbox”) hold any water? These moments are replayed in political satire (which seems to be an active part of this campaign) and seared in the brains of voters.
While I understand the flaws of his argument, I can imagine that a noted Red Sox fan such as yourself Professor Dickinson would have a special place in your heart for fivethirtyeight’s Nate Silver. He is, after all, one of the Editors of Baseball Prospectus and the inventor of the PECOTA system which forecasts the future performance of major leaguers–Theo Epstein is one of his biggest fans.
So while he might have some flaws in his forecasting of the Presidential election, we must never forget that he spends the majority of his time forecasting something far more important to the average American…baseball players.
Silver got the snap poll issue wrong, and as I hinted in a previous comment (apparently not strongly enough, given the feedback I received), you need a well-balanced diet of political readings. Over-reliance on any one source of information is a horrible idea.
But please don’t lose track of the argument from Markos — the ‘kos’ in “dailykos” in case you don’t follow this stuff, though you should know that he is not the author of all or even most of the posts there.
He’s one of the best analysts of the rough-and-tumble realities of politics. Let’s stipulate that his faith in the accuracy of the snap polls is misplaced.
Kos’s more interesting point (to me, as a member of the evil evil MSM) had to do with the post-debate debate. He argued that the snap polls made it impossible for the GOP to win the post-debate debate by focusing on relatively immaterial things like Gore’s “sighing” in his first debate with Bush in 2000. For those of you who don’t recall such ancient history, early polls and early commentary found Gore “won” that debate. But the Bush team and conservative pundits waged a remarkably effective campaign to focus the political media on Gore’s repeated sighs at some of Bush’s comments, and his alleged exaggerations, working to fit the debate into the existing narrative of, well, “Gore is arrogant and exaggerates.” The result? If you asked people in the immediate aftermath of the debate, they said Gore won. If you asked four days later, they said Bush won. Kos’s underlying point is, I think, broadly correct: Snap polls shape the media narrative and give Democratic partisans a pretty potent weapon in the post-debate “real debate” over who won, why, etc.” Sometimes, as I told one emailer, the illusion of social science is enough.
“McCain’s reference to Obama as ‘that one,’ or his failure to look at Obama during the debate, but there’s not much evidence that undecided voters care a whit about these issues.”
First, argh. It’s not only about the coverage you get, it’s about the coverage you don’t get. So you can dismiss that debate as trivial and even dumb, but minutes spent discussing those issues are minutes not spent discussing the candidates’ health care plans, and so yes, that discussion has a definite impact on undecided voters. (See the ‘sighs’ above). Please don’t dismiss this as pointless or immaterial.
Second, is there really no social science evidence showing that undecided/late-deciding/low-information voters cue in on relatively intangible things like “shares my values”? I’m surprised. Keeping in mind (always!) that the plural of “anecdote” is not “data,” I’ve talked to voters (voters I don’t know personally) of the late-deciding variety in this cycle, and I ask them what they think of the election, and I hear back pretty intangible “I feel like this guy has my back on the economy” arguments, as well as “oh, candidate X seems so (insert unflattering adjective here).” Campaigns fight to shape media narratives — something I will define here as “frameworks for understanding new nuggets of information about the candidate or his party.” So when Obama is painting McCain as “angry and erratic,” a discussion of “he wouldn’t look him in the eye” matters.
The usual rebuttal is that the exit polls show little evidence of this.
But one thing political scientists and reporters alike struggle with is how to deal with the accuracy of information that is self-reported. People like to be seen as making “smart” or “wise” decisions. (Anecdote: The Baltimore Sun found decades ago that no-one was reading their World section. They set about phasing it out. The backlash was instant, and brutal. Turns out that their readers didn’t read it, but they didn’t want to be seen as reading a paper that didn’t have a World section). So a voter tells you “I voted for McCain because I think he can better manage the economy.” Is it true? How can you tell? Did that voter really just worry that Obama was a secret Muslim socialist?
I’m not saying that polls are not, or cannot be, predictive. Plainly they can be. But there’s a difference between telling a pollster “I will vote for Candidate X” and telling a pollster a reason that may be seen as unpalatable. I know voters who are happy to tell pollsters that they’re voting for McCain, but won’t say why. Hint: I included it above.
I am relatively new to your blog and emails (they were occasionally forwarded to me), but I am interested in your overall opinions on how voters form their opinions on who to vote for. This notion of voters forming opinions within a larger mosaic of a campaign seems encouraging to me: it makes voters seem thoughtful. However, I wonder how you would balance this thoughtfulness with your earlier accent on the importance of demographic factors I read about particularly in your emails regarding the primaries. I look forward to more thoughts on how voters make up their minds…