As Goldy pointed out, there was a new poll for the Washington gubernatorial race released today by Rasmussen Reports. The poll was taken on May 12 on a sample of 500 likely voters, giving it a margin of error of about 4%.
The poll shows Governor Christine Gregoire is leading Dino Rossi 52% to 41%. This double-digit (+11%) lead is substantially better that her +5% lead in a late-April Elway poll poll and her +4% lead mid-April SurveyUSA poll. Finally, the previous Rassmussen poll, taken in late March gave Gregoire a slim +1% lead.
My Monte Carlo analysis of the Elway poll indicated that Gregoire had an 89.9% chance of beating Rossi (but only if the election were held when the poll was taken). It is precisely because that probability is less than 95% (a rather arbitrary number offered as a reasonable cut-off “significance level” by the great statistician Sir. R.A. Fisher early in the last century), that the results are considered a “statistical tie” or “within the margin of sampling error.” But having a nearly 90% probability of winning is not the same as having a 50% probability of winning—something that is implied by the phrase “statistical tie.”
So lets repeat the Monte Carlo exercise to estimate Rossi’s or Gregorie’s probability of winning a hypothetical election held right now. I simulated a million gubernatorial elections of 500 voters each election. Each person had a 52% chance of voting for Gregoire, a 41% chance of voting for Rossi, and a 7% chance of voting for neither.
The result? Gregoire won every single one of the one million elections. In other words, the poll results indicate that, in an election held right now, Gregoire would win with certainty. There are some assumptions involved: the Rasmussen poll is unbiased and the sample was truly selected at random. So violations of those assumptions may offer Rossi slight chances. Furthermore, there is no question that things can and will change before November. But right now the best evidence available suggests that Gregoire wins.
There was another interesting finding in the poll. When the likely voters were asked their views on each candidate, 49% viewed Rossi favorably; 55% viewed Gregoire favorably; 45% viewed Rossi unfavorably; and 41% viewed Gregoire unfavorably. That gives Rossi a +4% spread between favorable and unfavorable, and Gregoire a +14% spread.
I’m thinkin’ the “We Wuz Robbed” meme isn’t really working out for Rossi….
(Cross-posted at Hominid Views)