Poll: Merkley 48-42. Already voted: Merkley 53-39.

In their very last state-level poll of 2008 for DailyKos, Research 2000 has a new poll in Oregon's U.S. Senate race. It shows Merkley leading 48-42, with a margin of +/- 4%. Two weeks ago, it had been Merkley 47-41.

The R2K poll was conducted Monday through Wednesday. Among the 39% who said they had voted already, Merkley led 53 to 39.

Some interesting notes from Kos about the remaining undecideds:

The bulk of the undecideds -- 8 percent of 11 percent still remaining -- are Democrats. They've probably voted for Smith in the past, told he is a moderate, and are conflicted. All we need is half of them and we put this thing away.

We told you previously about the SurveyUSA poll released Friday, but the details are now available online.

SurveyUSA's poll was conducted Wednesday and Thursday, with a margin of +/- 3.8%. It showed Merkley leading 49-42, but among the stunning 71% who claim they've already voted, Merkley leads 52-40.

Discuss.

  • (Show?)

    I have an op ed distributing that criticizes polling firms for releasing what are essentially exit poll data of early voters, and here's a good example of why.

    Survey USA sucks. The "stunning" 71% who claim to have voted compares to the real <href=http: www.sos.state.or.us="" elections="" nov42008="" daily.pdf="">37% of ballots processed by the state.

    Do we really think another 34% of ballots are either in the mail or sitting on mantles? Highly doubtful.

    If we assume a total turnout of 85%, then the percent of ballots returned = .85 x .37 = .3145.

    Standard errors increase geometrically. If you want to reduce your standard error by 2, you have to increase the sample size by 2^2=4. By 3, then by 3^2=9.

    So if my back of the envelope calculation is right, a sample that is 30% smaller has a standard error not of 3.8 but more like 11-12%.

    This race is a lot closer than 12%.

  • (Show?)

    So, Paul, if they did a much larger sample of early voters - say one with a margin of error for that group was down to 3%, then you'd be happy?

    Or is your objection more than just the precision issue?

    I want the rest of the country to follow our lead and go 100% vote-by-mail, but I'm not sure I'd like to see polling firms essentially announcing the results days ahead of time.

    It's bad enough when they announce the results on the East Coast before the West Coast polls have closed...

  • Bill R. (unverified)
    (Show?)

    @ paul

    I dispute your assumption that there is going to be an 85% turnout. I think it's more likely to be in the low 70s. The nasty ad campaign in the senate race and the multitude of ballot measures will suppress turnout. In truth SUSA didn't say it was a 12% race, they said it was a 7% race, and I think that's probably pretty close. I will guess about 5% by Tues.

    I see your name on CNN site connected with the "Early Voting Information Center." How come no posts on that here?

  • rural resident (unverified)
    (Show?)

    If we assume a total turnout of 85%, then the percent of ballots returned = .85 x .37 = .3145.

    Paul G ..... I'm confused about this calculation. Based on your assumption about total turnout, the 37% returned as of Wednesday would represent about 44% of the total votes to be cast (.37 divided by .85).

    Does the .3145 have some other significance?

  • Jim Oleske (unverified)
    (Show?)

    Paul,

    If, as you say, "Survey USA sucks," then all polling is pretty much worthless. Survey USA has one of the best records in the business, by far.

    Notably, both their topline spread (7) and their early voting spread (12) are statistically indistinguishable from the R2000 spreads (6 and 14).

    They do have a much higher return rate than R2000, but their poll was taken later (10/29-30 as opposed to 10/27-29), and their question may have allowed for more people to answer yes if they had filled in their ballot, but no dropped it off.

    In any event, I choose to look at the Rasmussen spread -- 3 -- and conclude there is no cause for complacency and we must Get Out the Vote.

  • Bill R. (unverified)
    (Show?)

    What Survey USA shows is mirrored by the Oregonian study of ballots returned. http://www.loadedorygun.net/showDiary.do?diaryId=1449

    That, as of mid-week, 49% of Ds had returned the ballots sent them, 41% of Rs had returned their ballots. This suggests a dispirited Republican party that will not turnout the way Dems are across the country. If that percentage of ballot return holds up anything like that in Oregon, it will translate into a blowout for Dems.

  • (Show?)

    their question may have allowed for more people to answer yes if they had filled in their ballot, but no dropped it off.

    And of course, people self-report all kinds of things. They may say that they have "voted" if they've merely decided how to vote. Or if they've filled it out, and haven't signed, or stuffed, or stamped, or sealed the envelope - or put it in the box, or it's in the mailcarrier's truck - or it's sitting in a box unprocessed at the county.

    And keep in mind that they may just be lying -- after all, there are GOTV callers calling and by saying "I voted already" you get them off the phone. Wouldn't be a surprise if people were doing the same thing to pollsters - even automated ones.

  • Snipaw26 (unverified)
    (Show?)

    I have already voted but I can't wait until Wednesday. That way all the endless annoying TV ads change from mud slinging to endless analysis.

  • (Show?)

    In the general elections between 2000 and 2006, we've seen between 33-38% of the vote come in between the weekend before the election and Election Day.

    The numbers from the voter file show us just under 50% turnout so far for the state. If the same trend as usual holds, we'll definitely see a lot higher than the low 70's.

    What people have to remember is that voter turnout will always be lower than the percentage of people in a poll who say they have voted. There are a lot of people who like to turn in their ballots on Election Day. There are a lot of people who have voted their ballots, but just haven't turned them in yet for other reasons. I voted my ballot the very first day, but was waiting for my sister and husband to vote theirs (which they did yesterday). Now I have to take the trip via TriMet to go and turn in our ballots. And of course that's not so easy for those of us in east county - there are only 4 official drop locations east of I-205.

    I know over the next three days I'll be working hard to drive up turnout here in Gresham.

  • (Show?)

    "Survey USA sucks. The "stunning" 71% who claim to have voted compares to the real 37% of ballots processed by the state."

    Thank you Paul (well, not about SUSA sucking as a generality; it's true they're usually very solid)--I said the same thing when their survey came out. They reported 50% had voted at a time that not even 18% had been received. I don't think there's much worth in these pseudo-exits; as you note there is a smaller sample size, and they consistently seem to overstate what we know are the true extent of returns.

    A ballot is not returned until it makes it to the county clerk. Otherwise it may not end up being a ballot at all, even if it's filled out, signed and sealed. Or maybe THAT race has been filled out, but not some others, and there it sits until too late.

    However, as long as you keep a caveat in your head about it, and don't do anything stupid like start projection how much of the remaining vote the underdog now needs to win, it's an added piece of information.

    <h2>As for turnout, we may be short of 85%, but we'll be well ahead of 70%. As LoadedO reader SteveL has shown in this handy chart from three days ago, returns are between 2000 and 2004, which suggests something about 80%. As of yesterday 3pm we were at 48%. In 2004 it was 51% after 10 days; in 2000 it was 44%. So if anything we're inching closer to that 85%. Turnout's gonna be big, folks.</h2>
in the news

connect with blueoregon