Archive for the ‘polling’ Category

We get more speculation on the effectiveness of the red phone ad.

Did it work, the way pundits assume it worked in Texas and Ohio?
There’s no way to know. Ever. There are too many other variables. That’s a classic problem of causal explanation — and at the very least “analysis,” even political analysis, should try to rule out other possible causes.

It can’t, because even a poll that’s positive for Obama about who would make a better commander in chief doesn’t rule out the effectiveness of the ad. It could have been a significant factor in Obama’s meager support among whites, for instance. And race or candidate likability could have been the predominant factors, and the red phone ad entirely irrelevant in all states. Pre-established voter loyalty or their choice of candidate based on other factors will skew any answers about readiness to be commander in chief or even specific questions about the ad itself. Even if the red phone had been asked about explicitly, are respondents really good evaluators of whether they were affected by this or other factors? Nope. Again, they would be likely to claim the ad was effective or even “changed their mind” simply because they already leaned toward Hillary or had preconceived doubts about Obama as commander in chief anyway. It’s a foregone conclusion that he will win that poll in any state that he wins, and lose it in any state that he loses. If there is a disparity between the margin of victory and the commander in chief percentage, can we assign a cause to that? No.
There’s no way to control other variables barring running and not running the ad in separate Texas’s, Ohios, and Mississippi’s in alternate universes respectively. (Even then we can’t be completely sure, because it’s possible that other factors derived from elements of pure chance (chaos theory) have screwed up the control). Whatever the case, there’s no way to establish causality via statistical methods.

Not only does polling about the ad produce meaningless results, but it assumes that voters are dumb sheep, marketing victims, easily manipulated. That may be true, but if it is we ought to spend our energy on elevating public discourse rather than cynically reinforcing its stupidity.

The sham science of polling should be put out of its misery. It won’t only because charlatans make money off it, because it gives politicians something a little better than astrology with which to soothe their anxieties, and because it gives pundits something to make meaningless pronouncements about.


Read Full Post »