KPLU is reporting that a new poll in the Washington senatorial race will be released today from The Washington Poll. The poll shows Sen. Patty Murray (D) leading real estate opportunist cum foreclosure seminar motivational speaker Dino Rossi (R) by 50% to 42%.
The sample size is 500 giving a margin of error of 4.3%. I’ll provide a more detailed analysis when the poll results are released.
The Washington poll is made using live calls (like Elway). This poll differs from most other polls in this race in that the results are of registered voters, rather than likely voters. The poll was sponsored by KPLU and KCTS-9 and conducted by the University of Washington out of the Political Science Department. (In the interest of full disclosure, I work for the UW; however, I have nothing to do with this poll.)
5th consecutive poll in the too close to call/race of her life election showing Murray ahead.
[Deleted — see HA Comment Policy]
Sometimes the typos should not be fixed
“leading real estate opportunist cum foreclosure seminar motivational speaker Dino Rossi (R)”
[Deleted — see HA Comment Policy]
Mitch @3,
Thanks. Ahhh, the many perils of blogging with a low blood-caffeine level….
Rossi will get 42%. Just about anyone running with a “R” next to his name gets 42%. Hell, cigaret smoke got 38%! Four out of 10 Washington voters would vote for Gary Ridgeway if he ran for governor, U.S. Senator, or president with a cigaret in his mouth and an “R” next to his name.
Although the Washington Poll selects from registered voter lists, it further narrows the search by selection just as Elway does choosing only a mix of “perfect voters” and those who voted in 2 of the last 4 elections with a “sprinkling” of the newly registered. So those it is drawn from lists of registered voters, it is a bit of a hybrid of registered and likely voters.
“cum”?
sarge,
“Cum” is Latin meaning “with.” It means something like “plus” in this context. (Admittedly, it is a more common in British English but it isn’t uncommon in American English.)
Any polling info on the 3rd legislative district? Heck vs Herrera?
More good news (but off-topic): Dick’s announced it will build it’s new drive-in in Edmonds, apparantly in the Top Foods complex at 2100 Hwy 99. For those familiar with edmonds, this is near Stevens Hospital and Lynnwood Honda.
That should put a big crimp in the nearby McDonalds. But it’s a bit far for the kids from the high school to walk, they will probably still go to Burger King and DQ closer to the school.
Oh, the Murray/Rossi debate is now national news, featured on the top of the MSNBC page (although that changes so often, it might dissapear in an hour). It goes under the title “Democrat’s talking point: Her support for health care bill”.
Here’s a link to the article:
Murray touts health care bill in first debate with Rossi
Darryl,
How do people determine what would be an adequate sample size for a poll?
re 9:
Oh, cum on, Darryl. The ‘other cum’ meaning still holds — something like ‘plus’.
@14: Spooge works too.
Mary Plante @ 7
“Although the Washington Poll selects from registered voter lists, it further narrows the search by selection just as Elway does choosing only a mix of “perfect voters” and those who voted in 2 of the last 4 elections with a “sprinkling” of the newly registered. So those it is drawn from lists of registered voters, it is a bit of a hybrid of registered and likely voters.”
Not in the results that were released today. These results are the raw numbers for registered voters without screens.
Michael @ 13,
“How do people determine what would be an adequate sample size for a poll?”
In a close race (i.e. close to 50:50 split), once can determine the sample size (N) required for a particular margin of error (MOE) as
N = (0.98/MOE)^2
Hence if knowing the results with a margin of error of +/- 2.5% is necessary one would need a sample of (0.98/0.025)^2 = 1537 individuals.
If you can get by with only a 5% MOE, the needed sample shrinks to (0.98/0.05)^2 = 392 individuals.
Michael, what Darryl is really saying is that sample size often depends on how much money the pollster or sponsor wants to spend on the effort.
Rearranging the sample size equation he wrote, you can calculate the sample’s MOE as
MOE = 0.98/sqrt(N)
However, for the poll discussed here (N = 500), that equation gives MOE of 4.4%. More precisely, 4.38269…%. Paging Dr. Holman…
N in Seattle @ 18,
When I heard the first news blurb on the poll this morning, all they said was a MOE of 4.3% and I added that the sample size would be something like 525. A half hour later they had a story (online) that gave the MOE (4.3%) and sample size (500), and I modified the post (it was a few minutes after the initial posting). So those are their numbers, not mine.
Having tabulated the results of many, many hundreds of polls, I frequently find such discrepancies.
As you know, the MOE is computed from a binomial distribution. Most pollsters fix p = q = 0.5, giving the most conservative (highest variance) estimate for MOE. It’s possible that the WA Poll uses the observed p and q in their estimate.
I see there are not a great number of cum laude’s around here. In cursu honorum, some strive for extremes.
@8 In context, it’s a perfectly respectable word, correctly used, and with no sexual connotation. English is a language in which many words have multiple languages. It appears that when English was invented they ran out of words before they ran out of meanings.
You humans would be happier if you did less fighting and more fucking, IMHO. Just a rabbit’s perspective.
@20 Well, not among the trolls, anyway. Most of them got kicked out of grade school for stealing the chalk from the classroom. They ate it because their parents couldn’t afford the free school lunches.
Roger Rabbit,
“They ate it because their parents couldn’t afford the free school lunches.”
Or fed them cigarettes when President Reagan cut subsidies to school lunches and increased tobacco subsidies….
@17,18
Um… Thanks, but could you rephrase that for someone that failed algebra 101, twice.
Gee it is a real head scratcher why Rossi isn’t talking about his Tlingit heritage in this race. I wonder why? Isn’t being Indian a good thing still for him? Or is it a liability now?
# 26: The Tea Party folks probably figure the Indians are illegal immigrants, and want to kick them out. Rossi figured that out, and decided to be quiet about it. It’s easier for him to go along than try to explain it to them – not that they would listen, anyway.
Which is why we don’t have a proper second-person plural in the English language. The duplicative use of “you” and “you” in both the singular and the plural context creates confusion. Norther attempts to substitute “you guys” is rather offensive to any ladies in the group (should they feel included, or ignored?). Therefore the Southern use of “Ya’ll”, short for “you-all”, makes perfectly good sense. It’s probably the only good export to ever come out of the South (well, aside from the blues, rock-and-roll, etc.).
rhp6033, the South isn’t the only region to offer a second-person plural. For instance, there’s Pittsburghese:
To which I’d add the New York/Philadelphia youse, and probably others.
28
And, bourbon. Don’t forget about bourbon.
Michael @ 25,
“could you rephrase that for someone that failed algebra 101, twice.”
Sure…there is an equation that allows you to compute a sample size for whatever level of uncertainty is acceptable (@17).
Likewise, one can compute the uncertainty for whatever sample size one can afford to survey @18).
In practice, most pollsters know that 400 is about the minimum reasonable sample size. Most polls doing statewide head-to-head races have samples sizes ranging from about 400 to 2,500. More often than not, the samples are from 450 to 800 individuals.
@31
Got it. Thanks!
Adding to Darryl’s clear explanation @31, there’s little value to drawing really big samples. Using my rearrangement of Darryl’s equation, here are the MOEs associated with some sample sizes:
N = 400 … MOE = 4.9%
N = 500 … MOE = 4.4%
N = 600 … MOE = 4.0%
N = 800 … MOE = 3.5%
N = 1000 … MOE = 3.1%
N = 1500 … MOE = 2.5%
N = 2000 … MOE = 2.2%
Polling 4 times as many responses only cuts the MOE in half, and it costs waaaaaaay more to do.
@28, 30; That’s about right, but don’t forget Cajun food, Higgins boats and Viking stoves. Credit where credit is due. But, no more than is due. Just read a chunk of Grant’s Memoirs.