Dec 31 2007
Prior to the Christmas holidays going into full swing I noted the deteriorating nature of the polls and how to prove samples over the holidays were a bunch of garbage. Here is what I said on December 20th, 2007:
Never before has the Presidential election attempted an incursion into Americaâ€™s sacred time to give thanks and pray for peace and solidarity. Never before has politics intruded so much on Christmas (and hopefully never again). The candidatesâ€™ desperate attacks and dirty tricks are ringing a harsh dissonant note against the backdrop of Christmas caroling, and friends and family uniting to celebrate a man who tried to bring people together in peace – not to divide them with heated rhetoric.
I also cannot help but notice the polls started going haywire as we entered this last week before the big end of year break. We all know the problems with polls over the summer holidays when Americans are distracted with vacation and family. We all know weekend polls are highly suspect and erratic because of the fact catching people at home willing to talk politics becomes nearly impossible. They focused someplace else. So why would Christmas be much different?
But not to worry, like all good scientist testing a hypothesis I will make a prediction which, if it comes true or not, will prove or disprove my theory. My prediction is that after the New Year is past the polls will show a â€™surprisingâ€™ shift back to about where they were at the end of November. They may not go all the way back, but they will get back to within 3-4 points for each candidate.
To some this might have seen to be a bit of wild speculation. But to an engineer/scientist who understands the difference between haywire sampling and changes in the underlying dynamic being measured it was simple to predict the samples would be screwed up by the Christmas break. The only thing I could not predict was some major event (like the Bhutto assassination, but closer to home) throwing a major new dynamic into the mix. I do not think the assassination of Bhutto has really penetrated the average American’s gray matter, what with the Patriots going 16-0 and many other things worth contemplating this holiday season. So I am not surprised to see all these ‘surprise’ polls showing a return to normalcy as America begins its exodus out of the Christmas Season and back into the day-to-day grind.
Down more than 10 points in the polls here only weeks ago, Mitt Romney appears to be surging back days before Iowans head to their caucuses.
Two polls released over the weekend showed Romney ahead of rival Mike Huckabee by several points, while another survey out Sunday has Romney in a statistical tie with the former Arkansas governor.
The fact is pollsters, for the first time in a long time, have no way to gauge the electorate. The polls, and all the premature analysis based on them, are all garbage. It is just as bad as the ‘science’ behind global warming where misuse of statistics is causing amateurs to jump to conclusions based on incomplete and biased data. Between the diversion of Christmas and the rejection of politics as usual Iowa is not a good measure of the country right now. I could care less what its partisans do in this year of the anti-partisan, it is irrelevant and futile. Watch New Hampshire, South Carolina and Florida if you want to get even a inkling of where the political forces are heading.