So now the Washington Poll shows Dino Rossi slightly ahead of Sen. Patty Murray among those voters interviewed during the week of Rossi's announcement. Good news for the Rossi campaign? Certainly. But when it comes to polling, you always have to look at methodology. For example, the sample size during the week of May 24-28 that shows Rossi ahead was only 221 voters; it's not a statistically valid sample for a statewide poll.
The Washington Poll is administered by UW faculty, not a professional polling firm and their methodology is often unusual. For instance, for this poll, they conducted interviews for 25 days. Most pollsters will conclude their interviews in one to three days. Events and trends can change opinions, so to be accurate, you generally want a poll to be quick snapshot in time.
In February, 2008, the Washington Poll released a survey showing Christine Gregoire significantly ahead of Dino Rossi at a time when every other poll showed the race dead even. When you looked inside the poll, however, you found that the UW had tried to contact the same panel of voters they had interviewed in October 2007, but they apparently only reached half of them, yielding only 300 interviews. Again, a very odd methodology and a very small sample size for a statewide poll.
All polls are different. Some use automated calls (SurveyUSA and Rasmussen), some use live callers. Some interview all adults, others interview just registered voters, and some, particularly Rasmussen, are interviewing only voters they think are likely to vote this year. It is best at this point to take the Real Clear Politics approach and average the polls together. (Current RCP melded average puts Murray ahead of Rossi, 46-43.)
What are all the polls telling us about the Rossi/Murray race right now? It'ês very close and Patty Murray faces a very tough road to reelection. That'ês all we can say for sure as filing week begins.