Skewed Polls

Posted by Jeff Carter
on September 27th, 2012

There is a lot of talk about the polling data we are seeing today. Polls are all about statistics. If you understand stats, then you can also understand how they can manipulate them. For example, global warming statistics were hyper manipulated to form the distribution that they did. As a matter of fact, the math equation they set up would spit out the same data no matter what data you put in it! Polls for the Presidential Election may be manipulated today.

In the polling data we are seeing the mainstream media push today, it all shows Obama with a tidy lead in some cases. In other cases it’s within the margin of error. What we are finding out though is that those polls are weighted. This means they weight the data, or skew it, to show a pattern they expect in the electorate. For example, Democrats had record turnout in 2008. College kids had record turnout. Will it happen again in 2012 or will turn out resemble historical levels? No one really knows. But the only polls we are seeing are skewed heavily in favor of the Dems. Sometimes a sample as high as Dem +14!

There is a sight, UnSkewed Polls that you can look at. It takes the polls, and normalizes them. They are transparent in their process. If you believe their analysis, Romney is way ahead.

I haven’t seen an analysis of the way the pollsters are questioning their respondents, but I am going out on a limb and assuming their isn’t any bias. But the way things are being played right now, that’s probably a bad assumption.

There is a defense to skew a poll, or data. If you know something about the population you are sampling, you can get better inference by skewing. If the population of likely voters is really +14 Dem, it pays to skew it. The electorate is not 50/50. There are always more Democrats than Republicans-but the people that identify themselves as Democrats are down in this election cycle. That’s why the skews on the polls may be incorrect.

The polls were all wrong back in 1976. Reagan creamed Jimmy Carter. They were wrong again in 1992. They were incorrect in 2000. Even in 2004, people told fibs to the exit pollsters. 2008 was a very different election. But then, there was 2010. Different again and it was a midterm election. That kind of midterm election can’t be trusted for turnout. Can it?

The information in this blog post represents my own opinions and does not contain a recommendation for any particular security or investment. I or my affiliates may hold positions or other interests in securities mentioned in the Blog, please see my Disclaimer page for my full disclaimer.