Winning the Election vs. Winning the Polls: Possible vs. Impossible
Turnout assumptions make a McCain lead in the polls impossible
Many of us have been carefully watching for McCain to move ahead in this year’s national polling. However, we must realize that based on the turnout assumptions underlying all of these polls, it is nearly impossible for McCain to move ahead. That does not mean that the election is out of reach.
This year’s polling assumes a turnout model that, if it were to hold true, would require McCain to win upwards of 25% of Democrats in order to just pull even. Note that if such a thing were to happen, the actual voter turnout could not possibly look like what is assumed by the model. The model assumes that McCain is doomed to lose, and only seeks to explain why.
If the Joe the Plumber campaign and the Sarah Palin selection have indeed won McCain some votes, then we would expect to see a small effect in the numbers reported in the major polls. However, the greater actual effect to expect is that the actual turnout will ultimately look very different from the current assumptions.
If GOP turnout is within a few points of Democratic turnout in the swing states, then McCain wins. If Democratic turnout is ten to fifteen points greater, then McCain loses. If the assumptions hold true and young and single voters turn out in record numbers while older and married voters stay home, then Obama wins. If turnout is more conventional, then McCain wins.
It is very hard to develop a turnout model based on polling. It is even harder this year, in which Obama supports will naturally tend to want the talk about their support, while McCain supporters are naturally more reserved.
McCain may very well win on Tuesday. If he does, we will eventually learn that all of the polls were indeed right, but that the data was incorrectly interpreted because the pollsters based their assumptions on an erroneous reading of the dispositions of the American electorate.