The Skewing of the Vote

This morning brings a new poll from CBS/New York Times/Quinipiac that covers the states of Pennsylvania, Ohio and Florida. The results show Obama with a big lead in each state and the Democrat candidates for senate in each state with similar big leads. The result seemed like such a big switch from the past polls in these same locales that I decided to look a bit further. The results of that investigation were startling.

In Florida, the CBS poll questioned "likely" voters. The group consisted of 36% Democrats, 27% Republicans and the rest Independents. This is way off the mark. In Florida, registered voters are roughly 40% Democrats and 36% Republican. Likely voters are normally more Republican than registered voter samples. Other polling outfits that have questioned likely voters in Florida have found the samples to be almost evenly split; for example last week Survey USA found the likely voters were 40% GOP and 39% Democrat. That is a swing of 16% compared to the latest poll from CBS and the New York Times.

Normally, one would expect a variation in the poll samples. The results from CBS and the New York Times, however, are not just a normal variation. The pollsters at Quinipiac surely knew that a sample in Florida that favored Democrats over Republicans by 36 to 27% was way off the mark. Indeed, the sample is so skewed that the poll results are meaningless. Any pollster interested in accuracy would either adjust the results to reflect a proper sample or else take a new sample. That Quinipiac did not do either seems like proof that the poll was released to satisfy the need at CBS and the New York Times to affect perceptions of the current state of the race.

It is too bad that bogus polls like this one get released as "news". The real story here is media bias for Obama, not the poll "results".