Polls this cycle have been all over the place, yet they consistently show how momentum is derived largely from debate performancesCaution, everyone. All of us who are following the election closely are very much aware of the debates. But that's simply not the only thing going on. Since December, there are ad wars in all of the early primary states, with saturation levels of TV ads, presumably accompanied by other forms of advertising. There are also candidate appearances and local news hits. Each early state has its own local conservative talk radio hosts. And then there's the various nationally syndicated radio hosts, plus Fox News. On top of all that, there's straight news coverage, including on those local news shows and within the GOP-aligned partisan press, of campaign developments including not only debates but also polls and primary results. All of these may produce direct and indirect effects, including pure "momentum" effects when inattentive voters who like all the candidates simply tell pollsters they support the candidate they most recently heard something good about. And of course saying that "debate performances" moved polling numbers is also dicey just by itself; impressions of those performances can be unmediated for those who watch the debates in full and nothing else, or mediated by the press and party actors for those who learn (or remember) what happened from news recaps, sound bites, and whatever pundits and talk show hosts choose to talk about.
Getting back to debates vs. everything else: most of that, to most of us who are not in the state that's preparing for a primary this week, is mostly or entirely invisible. The debates, on the other hand, are extremely visible. Later on, with any luck we'll get some good analysis that can attempt to sort out all these effects. Including why we're getting all these large swings (Nate Silver today suggested the old Mickey Kaus effect; I'm sticking with the "no heavyweight" theory). But for now we should all be very cautious, in my view, about leaping to the conclusion that the most visible thing to us is the one that's pushing all the polling.
Posts
How was it that so many debates were scheduled in the first place? (Or is this an illusion? Were there this many for both parties in 2007-8?) Did the RNC direct it, or did state parties ultimately push for it with the blessing of the RNC? I feel like I've seen a fair amount of pundit articles that have complained about the number of debates, but wasn't this a conscious decision of parts of the hierarchical party apparatus, so there was probably a rational reason behind doing so?
ReplyDeleteI'll correct myself here: it seems that there were essentially the same number of GOP debates by this point in 2007-2008. The party or the TV networks or whoever have not been *adding* to the number of debates.
DeleteSo their seeming salience this year is either an illusion of political junkies and party actors or it's due to something else. That said, since perceptions do play a role in primaries, there's a notable reality and significance to even the mere illusion among political junkies/party actors that there have been too many debates.
My theory is too many polls over too short a time.
ReplyDeletePolls aren't THAT accurate to begin with-- while an average of polls definitely measures something, a lot of people answer polls despite not really having thought about their vote. They are just spouting a worthless opinion. People don't prepare to answer a polltaker's question the way they prepare to determine who they are voting for in an actual election.
But all these "tracking" polls are worse, because they have tiny samples (usually no more than 300 voters) and are taken too often, just measuring random statistical noise rather than actual swings in public opinion.
We are asking polls to do more than they can really do. And that's why they keep on going up and down.