04 November 2008

More Reasons For Optimism

From Jim Geraghty at Campaign Spot, his friend, the man they call "Obi-Wan." Let the Democrats and moderates get their poll-predicting information from complex statistical modeling and empirical evidence. As for me, I'll take the word of a Star Wars hero:

Look, the real drama to this election is being provided not by the candidates but the polling community. By which I mean the decision they made to stake out — as Campaign Spot has noted — a remarkably bold position, that the Democratic Party turnout is not only going to exceed a recent historic advantage of 4 percent but go to 6.5 percent (Rasmussen) to 8 percent in many polls to even 12 percent in one.

I keep looking for the justification for this. Not easy to find. Rather like the academics' one-time belief in the Aristotlean spheres and an earth-centered universe, it just seems to be a pretty good working theory — some sort of way to make sense of observable phenomena and keep all the smart people talking agreeably and pleasantly among themselves.


In the old days the networks had political directors like Marty Plissner and Hal Bruno who kept an eye on the tendency to politicize the number-crunchers. What happens when that sort of internal check is lost was evidenced by the spectacular embarrassment – the debacle — of the exit polls in 2004. (Obi ) The polite explanation was that the skewing resulted from the fact that Democratic voters are more likely to talk to polling representatives at the polls. What got buried was the fact studies found that the cultural-political backgrounds of the exit-poll employees was a big factor.

Anyway, back in the days when exit polls were reliable — if a first or second wave of numbers were seen similar to the McCain-Obama battleground polls that came out today the network insiders would have been saying: hold everything, this is a very close one. (That's because they usually wacked two or three points off the Democrats' total since urban areas get better represented in the exit polls.)

So, if the polling community is basically right in their turnout models, this is looking like '64 — a nightmare scenario for the GOP. But if they are off to any significant degree, the state polls seen today (even though some of them favor a high Dem turnout model) make this a very different race. And what about the outlier polls in Pennsylvania and even one in Minnesota showing a close race?

And there are other questions. What about the reaction to media bias (Obamamania, the resentment towards Sarah Palin and Joe the Plumber) driving up Republican turnout? What about the extent to which Reagan Democrats in Florida are being urged to the polls by the McCain campaign?

And the reason that the polling community, not to mention the Obama campaign should be uneasy is that finding the justification for their heavy Democratic weighting isn't readily accessible. And that is the point – along with failing to themselves take note that in this period of unprecedented economic turmoil and therefore any predictions this year might be questionable or at least hugely complicated, the pollsters and media gurus never really put their own premises about voter turnout front and center and asked for questions, objections and evaluations.

They seem to have slipped into a world of easy assumptions. Always dangerous for those whose job is to quantify and track the stars and planets of an ever-changing, ever-moving political universe.

Ignore the exit polls, wait for the real voting tallies to come in.

For those of you who prefer "stats" to "Star Wars," this seems like as good of a time as any to re-link to Iowahawk's little ditty on the supposed scientificness of polls and stats behind them.

Let me put it this way: Things with a lot fewer variables and even lower probability happen all the time.

If you have tips, questions, comments or suggestions, email me at lybberty@gmail.com.