When I was in Business School, several courses were designed to teach us statistics and most of those convinced me, in market polling and customer surveys, that is is extremely difficult to get good poll results when working with the general public. I tend to distrust poll numbers, especially in political races simply because people will lie to you. It's one thing to measure widgets coming off an assembly line and get an idea about your production standards. It's another thing entirely to measure people, and any pollster can inject bias into a sample, even when they're trying very, very hard to be fair. It's part and parcel of the business.
Polls tend to favor incumbents. I put the level of confidence of all political polls at 90%, which in my mind tells me that they're skewed 10% in one direction or another. Nowadays the pollsters call that the margin of error. I've studied the math behind human polling and I'm here to tell you that no pollster can tell you within 10% what the public is thinking. Don't let them bullshit you. If you're paying attention to the current discussion among news sites, you'll see some question about the level of confidence we should have in polls. I"m not alone here.
Last night at the family gathering, we were talking politics and I predicted that Mitt Romney would win with about 55% of the popular vote. Lots of the folks present thought that I was wrong, but I'll persist in my prognostication, for the following reasons.
The economy sucks. Incumbents don't do well in a bad economy. Anyone who has bought gasoline or groceries with their dollars lately knows that those prices have risen dramatically. The middle class, and the lower end of the middle class is hurting right now. They're stretched pretty thin.
For those of us who pay attention to foreign policy, the President's foreign policy is in tatters, to the point where knowledgeable pundits are saying that the President actively lied about the Benghazi debacle.
At this point in the 1980 race, President Carter was ahead of Ronald Reagan by 4 points. In late October, Carter was ahead by 7 points. When the polls closed at the end of the voting, Reagan had won by 4 points and a huge electoral college victory.
UPDATE: As I was posting this, Instapundit linked to this report on PJ Tattler that only 9% of people today are cooperating with political pollsters. That is a tiny number and illustrates the difficulty of conducting political polls. I stand by my prediction. Romney is going to win by at least five points.