Even the best polls underestimate Republican vote

Rasmussen Reports, the U.S. pollster has a self-congratulatory article about its polling for the 2004 presidential election.

It is true that Rasmussen forecast the overall result in terms of the popular vote, getting within half a percent of President George Bush's score and even closer with Senator John Kerry's.

However, is it really clever to publish a list of 25 states that shows that Rasmussen underestimated George W. Bush's vote in 23 of them? With only New Mexico spot on and New Jersey the only state overestimated for the Republican candidate?

I would have a lot more time for opinion polls if in a tight race the over- and underestimates showed more consistency. As it is I expect polls to underestimate Republican voting figures, which is why commentators may believe that I'm biased against Democrats.

According to Matthew Dowd, Bush's campaign strategist: "Scott's [Rasmussen] polling data was dead on this election. Both nationally and at the state level, his numbers were hard to beat."

No comments: