The Moderate Washingtonian

Outlook on politics and elections in the state of Washington from an overall centrist viewpoint. My views tend to be libertarian in nature, but at the same time are largely nonpartisan.

28 September 2006

New polls

In both the US Senate race and the 8th district Congressional race, both SurveyUSA and a partisan pollster has released polls recently. In both races, the partisan polls seem to not have been to their party's advantage to release, and SurveyUSA continues to be Maria Cantwell's favourite pollster. The data:

Senate race
SurveyUSA 9/21-9/25 +/-4.5 MOE:
Maria Cantwell [D] 54%
Mike McGavick [R] 42%
Aaron Dixon [G] 3%
Bruce Guthrie [LBT] 1%
Robin Adair [IND] 0%

Strategic Vision [R] 9/22-9/24 +/-3 MOE:
Cantwell 49%
McGavick 40%

8th District race
SurveyUSA 9/24-9/26 +/-4.3 MOE:
Dave Reichert [R] 50%
Darcy Burner [D] 48%

Grove Insight [D] 9/18-9/21 +/-4.9% MOE:
Reichert 44%
Burner 43%

SurveyUSA's result for the Reichert-Burner race is much more in-line with what I was expecting from their previous poll, and while I think this one is closer to the truth, the partisan identification change between the two could have easily been the reason for the change. Since their August poll, their Republican sample was virtually unchanged while the Democratic respondents jumped 6%. Even with this, Reichert draws more Democrats than Burner does Republicans, and took self-described Independents 53-40. I think this is a silver lining for the Reichert campaign from this one, it seems as if they should still win as long as they can get voters to the polls (though this year that is no certainty). The Grove Insight poll was commissioned by EMILY'S List.


At 8:43 AM, Blogger Daniel Kirkdorffer said...

You got one of your numbers wrong. The 53-40 split for Independent votes was from August. That's now 50-45, an 8 point swing in Burner's favor.

You also fail to mention that a more likely reason for the change in numbers has to do with the fact that the August SurveyUSA poll for WA-08 was of registered voters, while the current one is of likely voters.

You also don't explain why it wasn't a good idea that the Grove Insight poll was released. It shows the race a tie, just as the SurveyUSA poll does. With Burner still yet to grow her name recognition, that's only good news for a challenger of a well known incumbent.

At 8:39 AM, Blogger TMW said...

My bad on the independent numbers, an unfortunate error from having both windows open at the same time. The registered/likely scenario is a possible explanation of the swing from Independents to Democrats from the August numbers, but it could just as well be a result of the random telephone sampling of the pollster.

The reason I think the Grove Insight poll should not have been released is because as a partisan poll people should and do take it with a grain of salt. The SurveyUSA aside, the first thing I would take from seeing the partisan firm numbers is that it's apparent that Burner cannot beat Reichert even in a favourable poll. Granted, it's a statistical tie, but I don't think it's unfair to figure a partisan poll has more potential to inflate the candidate's numbers and deflate the opponent's numbers.

At 4:37 PM, Blogger Island Republican said...


The SurveyUSA poll is a robo call poll system according to Randy Pepple, CEO Rocky Company. He stated this fact in a Greater Seattle Chamber of Commerce speech last night. His question was what type of voter was willing to listen to a computer call for over 90 seconds to make their choices. He believes this type of polling is not very accurate in a close race.

Island Republican

At 10:23 PM, Blogger TMW said...

I've heard those criticisms of SurveyUSA from both sides each time they release a poll that is unfavourable for them (Goldstein said much of the same when its August poll had Reichert leading big). In my experience, however, their polls are credible and more often than not accurate, so until another independent pollster comes to a different conclusion, I think the race is probably somewhere between a dead heat and a small Reichert lead.

At 12:15 AM, Blogger Will said...

No matter which way the poll cuts, I have found robo-polling to be no less accurate than any other kind of polling. Blogs I read that focus on polling (especially DailyKos) have noted that robo-polling does pretty well. People will tell a machine something that they won't tell a person. (such as on school funding, where normal polls skew higher on support for school funding. Folks tell the guy on the phone that they're for more money for schools but will end up voting 'no'. A similar skew upwards can be seen for 'minority race' candidates and even female candidates. Folks don't want to come off as bigoted.)

In short, I think dissing SurveyUSA beause it is automated is a cop out, no matter who does the dissing.

At 8:11 AM, Blogger TMW said...

I agree, William. I've also heard of poll skew such as you stated happening in other states with anti-gay ballot measures. The end result of many of those measures run contrary to polling because some people are embarassed to admit to others their personal biases.


Post a Comment

<< Home