Sunday, August 24, 2008
Polls, Predictions and the Presidential Race
They have a certain scientificity about them that Op-Ed pieces don't.
Polls are grounded in Statistics, in the formal sense of the word as a subset of Mathematics. You can't argue with Mathematics.
As this is written, the polls, irrespective of the pollster, unanimously declare the presidential election to be a very close one...a virtual tie.
This astounds Republicans, of course, because they know all too well that they should be light-years behind the Democrats, given that their nominee is effectively the incumbent and therefore must defend the record the Republicans have assembled for themselves over the past eight years. That record needs no elucidation here.
The Democrats, on the other hand, seem to be interpreting the Polls as an indication that there is something amiss with their candidate, or least with his campaign strategy. They too think they should be light-years ahead, given the brutal and concrete facts that Bush & Co.'s tenure have bestowed upon the Democratic Party.
The poll question that matters most, and that causes all the hand-wringing on one side and jubilation on the other is: Who would you vote for for President of the United States today?
But away from Gallup, Zogby, Pew there are at least two other organizations (which I have blogged about and cited in the past) that are have been engaged in a similar yet different enterprise than the mainstream polling organizations.
These are www.intrade.com (intrade) and the University of Iowa's Prediction Markets (Iowa.)
Intrade and Iowa allow participants to buy and sell, with real money, "futures" contracts on various propositions, including but not limited to the outcome of the presidential race. Over an (statistically) significant period of time, these two market makers have had an uncannily high success rate when it comes to predicting outcomes of political races.
Iowa and Intrade are not asking their clientele, Who would you vote for today? Rather, they ask the question, Who will win the election? I have not heard of any mainstream polling polling organization that asks, or at least reports on, that question.
If one were to visit the websites of Iowa and Intrade (I will supply the url's at the end of this post), one would discover that one both sites the answer to that question is approximately, Obama: 60%, McCain: 40%.
This result is much more in line with what the socio-economic-political conditions of the United States in 2008 would lead an objective observer to surmise regarding the chances of the Democrat vs. the Republican candidate.
Why the disconnect between Intrade/Iowa and the mainstream pollsters?
Are the questions being asked so vastly different? If the pollsters would ask the same question as does Intrade/Iowa, would the results be the same?
There is one difference, at least, that can be pointed out. It doesn't cost anything to answer a Gallup pollster's question. But to "answer" an Intrade/Iowa question, one has to put one's money where one's mouth is and purchase a McCain or Obama "future's' contract.
Which produces the more accurate snapshot of reality?