SurveyUSA: Merkley is surging (and here's one reason why...)

Kari Chisholm FacebookTwitterWebsite

Last night, KATU released the results of the third SurveyUSA poll in the Democratic primary for the U.S. Senate.

With a four-point margin-of-error, it's true that Jeff Merkley and Steve Novick are statistically "tied" - but only barely. Merkley leads Novick by four points - 31 to 27 - so Novick's best-case-scenario is +4, whereas Merkley's best-case scenario is +12.

[Update: After this post went up, the Tribune and KPTV released a Davis/Hibbits poll that shows Novick up 29-23, with a 4.8% margin-of-error. No trend lines yet, however, since this is the first (and only?) poll they've done.]

Here's a chart of the three SurveyUSA polls - with shading to illustrate the MoE.

Surveyusamay12

Compared to the same poll just eleven days earlier, the net shift has been +6 in Merkley's favor.

As with any campaign, there are lots of factors at work here. Some of them are obvious and public (like TV ads) and others are very nearly imperceptible (like hundreds of campaign volunteers knocking on doors across the state). In-between are the effects of direct mail, campaign events, personal emails from supporters, and yes... radio commercials.

On the one hand, radio is very public. After all, it's broadcast. On the other, because radio is so much cheaper to produce, it can be much more narrowly targeted than television. If you're not in the target market (or if you listen exclusively to public radio), you're not likely to hear a particular radio spot.

Despite my role on the campaign, I had no idea the Merkley campaign was running radio ads targeting women voters - until today, when I heard one while fiddling with the dial. This 60-second radio spot has been running for several weeks now.

That might explain in part the strong surge in support that Jeff Merkley's seen from women voters in Oregon. Here's the chart. (Unshaded, because I think the smaller sample changes the MoE, but I'm not sure how much. Oh, to remember my grad-school statistics...)

Surveyusamay12women

On the jump, the same chart for Oregon male voters - for those who are curious.

Surveyusamay12men

[Once again: my firm built Jeff Merkley's website but I speak only for myself.]

  • Daniel Spiro (unverified)
    (Show?)

    We'll see what effect Novick's new ad (unleashed only yesterday) has on these polls. It is the perfect response to the negative ads that Merkey has been running lately.

    If the people of Oregon want to choose politics as usual over the inspiring and non-traditional approach Novick offers, so be it, but I have the sneaking suspicion that they will vote for change this go around.

  • (Show?)

    Yeah, uh huh. Jeff Merkley is "politics as usual"? The guy that finally figured out how to take back the Oregon House - after sixteen years? Right. He doesn't know anything about winning and delivering on change. Mmmm hmmmm.

  • Daniel Spiro (unverified)
    (Show?)

    Kari posted: "Yeah, uh huh. Jeff Merkley is "politics as usual"? The guy that finally figured out how to take back the Oregon House - after sixteen years? Right. He doesn't know anything about winning and delivering on change. Mmmm hmmmm."

    I'm just going with the views of the vast majority of the newspapers in the state ... people who are impartial, as opposed to folks (like Kari) who are on Merkley's payroll.

    If Merkley were truly the cats meow, given his bankroll from the DSCC, he wouldn't be in a statistical dead heat with Novick, he'd have Novick way, way in the rear view mirror. That's one of the reasons why so many impartial people are voting for Novick. They realize how much more he could do if and when he has DSCC money in his pocket, and not in the pocket of his opponent.

  • Dan (unverified)
    (Show?)

    Claims that Merkley was somehow singlehandedly responsible for flipping the state legislature are ridiculous and don't take into account the major anti-republican sentiment here and around the country during the time. Merkley is hardly responsible for that.

    George W. Bush bears much of the responsibility for what happened all over the country in 2006. If Democratic strategists somehow think they had some incredible gameplan in that got them the gains 2006, and will this year as well, then we are making short term gains indeed. The Republicans dramatically overplayed their hand and basically handed us the game.

    We as progressives have to get smarter and change the way the game is played. Sending out establishment candidates won't initiate the drastic changes we need made in the system. That's why I'm voting for Steve Novick.

    I don't think Merkley is a terrible candidate, just an okay one. Steve Novick is a major talent and that is why he should be our next senator.

  • (Show?)

    Well, I believe that Jeff is also a major talent, and he will make a fantastic United States Senator after he beats Gordo in the fall. He can do it.

    And, as for his sucesss in the legislature, Jonathan Poisner says it all, with respect to the environment:

    When Jeff Merkley joined the Legislature, he immediately took up the environment as an issue. He was smart enough to figure out quickly that with hostile Republican leadership, he had to think small – so he focused and made progress on an important, but narrow issue – reducing water pollution from outboard motors. As he moved up in seniority and led the Democrats back into power, his vision expanded dramatically. He was always a good vote and leader, but when the Democrats took control, he shined. The 2007 Session was the Legislature’s greenest in 30 years – passing 16 separate environmental bills, with the highest average environmental rating since 1977. I don’t think folks recognize what an extraordinary achievement that was and how much of the credit is owed to Merkley. It’s not just that there was only a 31-29 Democratic majority in the House, with virtually all of the Republicans hostile. It’s that some of the Democrats also had a track record of voting for the environment less than half the time. Merkley succeeded in getting them to a yes vote on bill after bill by working closely with Environment Committee Chair Jackie Dingfelder using all the tricks of the trade – personal relationships, unusual allies, coalitions, and hard-nosed pressure. These are precisely the skills that separate successful U.S. Senators from those who go there and make speeches. If you’re electing a President, speeches really matter. As a Senator, not so much. There’s a chance Steve Novick might turn out to be great at moving an agenda in DC, but it’s just that – a chance. With Merkley, we’re not taking a chance.

    We can't understate the importance of what he helped craft. Sure, he wasn't solely responsible. But he should get a lion's share of the credit, for causing the flip to happen and then piloting a narrow majority to one of the best progressive sessions on-record.

  • Jim Ross (unverified)
    (Show?)

    Radio is a great persuasion tool, especially to women, who tend to vote much more rationally than male voters, who vote more viscerally. Women tend to be late deciders and this spot at this time could be decisive. This race is going to come down to the wire.

  • (Show?)

    kari, did you let Kevin write this one? Because your senses have left you in your zeal to prop up Merkley.

    The last poll was a tie. This poll is a tie. In fact, the last poll could have easily been 31-27 Merkley, and this one could be 30-28 Novick--and here you're potentially crowing about the wrong poll.

    You why else it represents no real change in the race? Undecided haven't budged; still 24%. With essentially the same pool of voters, they got a different answer within the same range as before.

    Is it nicer to be up than down? Sure---but you can't hang your hat on it at all. Novick had a 60% of being ahead last time; maybe it was actually in the 40% for Merkley. Same situation here.

    To declare that Merkley has "surged" is unsupported by the data, and an irresponsible claim. And to further justify it on the basis of Merkley actions even less grounded in evidence. And in any case, after the petulant display Canter gave after the first SUSA poll, merkley doesn't get to claim later polls are suddenly valid because he likes the toplinebetter.

    And finally, 43% already voted? Right.

  • (Show?)

    I dunno, TJ. There's been a lot of talk about poll results throughout this campaign. But I think that it's fair to note that it's still very close and with many undecideds. And it's absolutely a race to the finish.

    But, you know, the trends are noteworthy. I disagree with the notion that it's irresponsible to claim upward movement where it exists.

  • trishka (unverified)
    (Show?)

    it's irresponsible if the statistics don't back it up.

  • BOHICA (unverified)
    (Show?)

    a close italics tag would be nice

  • (Show?)

    "But, you know, the trends are noteworthy. I disagree with the notion that it's irresponsible to claim upward movement where it exists."

    Ben, that's what I'm saying--there's no basis to claim it DOES exist. The last two polls could have been run on the same day with the same voters, and gotten the two results you see here.

    It IS very close, it IS a race to the finish with apparently 1 in 4 still undecided. Those conclusions are well supported by the data. What is NOT supported, is that Merkley is "surging," between the last poll and this one.

  • (Show?)

    This is GREAT news for the Merkley campaign.

    Not only are all the trend lines running in his favor but some of the earlier prognostications - that Merkley's ad would hurt Obama - simply are not reflected in the accompanying SurveyUSA presidential poll where Obama has increased his lead.

    As for the cross-tabs on this Senate poll, I was most struck by Merkley's surge and lead in the 18-34 demographic which appears to rebuff another prognostication that Novick & Obama would benefit from young voters. Since the May 2nd poll Obama gained 13 points, Merkley gained 17 points and Novick lost 13 points among the 18-34 group.

  • (Show?)

    and the MoE on women is around 5.5-7, probably in the low 6s depending on the female universe (I did 410,000 and put the number of women in the sample at 315, giving them a slight numeric edge both times).

    Which means you can't say much about how they're doing with women, either. If the spread ain't 20 points at least, I wouldn't conclude anything much at all. Crosstabs you can't count on.

  • (Show?)

    If you look at the methodology of this report, and look carefully at the results, there's reason to be skeptical of the conclusions being drawn. First of all, this was a telephone poll, which means the "random" phone numbers were drawn from directories of landlines. What is the percentage of young voters with landlines? Second, the breakdown of likely voters is 19% for the critical vote under age 34, further indication that the survey essentially missed young people entirely. And finally, what group is responsible for the huge increase in Democratic voter registration?

    It's been well established that young people are registering and voting because of Obama's candidacy, and that same anti-establishment enthusiasm is more likely to respond to Novick than Merkley. It's going to be very interesting when the primary votes are actually counted, but I think you're going to see a marked variance from these results, and it's going to come from young voters.

  • (Show?)

    "Since the May 2nd poll Obama gained 13 points, Merkley gained 17 points and Novick lost 13 points among the 18-34 group."

    If you believe that much swing really happens in one week, you can't be helped.

  • (Show?)
    "Posted by: Daniel Spiro | May 13, 2008 3:50:26 AM I'm just going with the views of the vast majority of the newspapers in the state ... people who are impartial.."

    LOL.

    It's so cute to when the little ones still believe in Santa.

  • (Show?)

    Ah, questions of methodology indeed. I think we've all taken turns with various polls arguing the ifs and buts on the results.

    It might not be a set-in-stone readout, perhaps, but what poll is? As it stands, trends are still overwhelmingly positive for Jeff, if still quivering with a degree of undecideds. But, you know what, I'll take it. And I think it does show people are responding to Jeff as they get to know him, his body of work, and his efforts.

  • (Show?)
    "Posted by: Dan | May 13, 2008 5:24:57 AM Claims that Merkley was somehow singlehandedly responsible for flipping the state legislature are ridiculous and don't take into account the major anti-republican sentiment here and around the country during the time."

    Of course nobody is actually claiming Merkley flipped the legislature by himself. However it is a fact that he was the pivotal person leading, strategizing and recruiting for the effort to flip the leg.

    To ignore that he was the one behind our wins is to be oblivious to what actually occurred. To dismiss or downplay that reality only reveals the person suggesting it doesn't really know what they are talking about.

  • (Show?)

    If you believe that much swing really happens in one week, you can't be helped.

    Oh I dunno, Mark. You touted the high quality SurveyUSA's polls so often back when they showed Novick ahead that you've made a believer of me.

    :-)

  • jeff Merkley's conscience (unverified)
    (Show?)

    Maybe the saddest thing here is watching people basically say, "Yay Merkley's disingenuous smear campaign is working. Yay America. Go Jeff!"

    There aint no "progress" involved in winning because you're more willing to lie and spending more money.

    Thaat's more of the same ol' same ol', and on that note, Merkley's negavitve ad is the perfect encapsulation of his campaign: Off the shelf, just like Kitzhaber says.

  • (Show?)
    Posted by: torridjoe | May 13, 2008 8:23:59 AM Ben, that's what I'm saying--there's no basis to claim it DOES exist.

    Except for the actual poll numbers.

  • (Show?)

    "Oh I dunno, Mark. You touted the high quality SurveyUSA's polls so often back when they showed Novick ahead that you've made a believer of me."

    That's a nonresponsive answer. I have been consistent in believing that SUSA does generally good work. Unlike Merkley folks, I didn't pretend the poll must be fucked up in some way because my guy wasn't nominally ahead. What I'm saying is that swings of which you describe are extremely unlikely events in a week's period. Voter sentiment just doesn't move that fast in most cases, even with a smear ad in play. I would welcome your reprinting of my pieces on the previous two SUSA polls, in order to point out where I have gone beyond what the data will allow.

    Quality of the polling doesn't have anything to do with how to read one. A little knowledge is a dangerous thing, and with polls you are Chuck fucking Norris, Kevin.

  • Fair and Balanced (unverified)
    (Show?)

    How about going with the views of those who are in a really good position to judge the merits of a candidate for a legislative body; i.e. Oregon's state legislators?

    Every single one of the Democrats in the House, from the most liberal to the most conservative - and there's quite a span - have endorsed Merkley. It's not like they have to; he won't even be a member next session, win or lose. They were just impressed with his performance.

  • (Show?)

    Mitch, what ARE the actual poll numbers? I wasn't aware you'd contacted all 800,000 voters this weekend. Failing that, "the actual poll numbers" are subject to error. In this case, 8 points of it across the two candidates.

    And...OH NO! Another poll, from Hibbits, that is fucked up and worthless! Why? Because it shows Novick nominally ahead! Ack! It's dirty, demean it before someone believes it!

  • (Show?)

    Is there anything more compelling than an anonymous poster calling a candidate a liar with no links or reference to specifics?

    <hr/>

    Lestat, loved the Santa reference.

    <hr/>

    I tell, ya, with my wife in the tank for Clinton and the Forever War in the Senate race, I can't wait for next Tuesday. My head is about to explode from the stream of illogic and the willful disregard of any and all countervailing facts among the Movers and Shakers in this state.

  • (Show?)
    Posted by: jeff Merkley's conscience | May 13, 2008 8:36:12 AM Maybe the saddest thing here is watching people basically say, "Yay Merkley's disingenuous smear campaign is working. Yay America. Go Jeff!"

    You are right. After all Merkley ads employed the Jedi mind trick (aka Novick's own words) to "smear" Novick into calling Obama the "great white hope" and a "fraud" because, well that's just Novick being a good progressive and focusing his sharp mind on the substantive issues that affect people's lives.

    I know I hate it when Merkley "smears" me by quoting my own words and "smearing me" into being an asshole towards my co-workers.

  • orexpat (unverified)
    (Show?)

    http://www.portlandtribune.com/news/story.php?story_id=121065483391101200

    According to the Tribune Davis, Hibbitsts & Midgall is calling it 29-23 for Novick with 40% undecided. But they are just a small local firm, so what do they know? ;)

    Regardless, it's going to be interesting down to the wire. It would be nice if supporters on both sides could refrain from throwing rocks in the meantime.

  • (Show?)

    And by the way, I should have much more cred now that I have switched allegiances and now firmly back Jeff Merkley for Senate.

    No really, go look! I honest to God have a "Merkley for Senate" post that I wrote, at LO!

    If only you'd have realized what it would take before, Kari!

  • (Show?)

    Cool down, TJ. Yes. The Hibbits shows Steve narrowly ahead. And, yes, the SUSA shows Jeff narrowly ahead. And, yes, it's all somewhat of a sprint-to-the-finish tie if you want to look at it like that.

    In SUSA, where we get excellent cross-tabs, we can see Jeff's growing success. I'd love to see whatever cross-tabs Hibbits has. They sure show a lot of undecideds (over 40%), but that's okay too.

    The Hibbits poll is not worthless, but neither is this one. But I'll take Jeff's movement as I can see it; a month ago, he was close-but-behind in the polls and now he's picking up steam and leading in several key categories (if SUSA is to be believed).

  • (Show?)

    I've updated this post with the Davis/Hibbits poll info. Thanks for the tip, TJ.

  • (Show?)

    You also screwed up the font, Kari!

  • (Show?)
    Posted by: torridjoe | May 13, 2008 8:43:58 AM Mitch, what ARE the actual poll numbers?

    Ahhhh... the numbers which are the basis SurveyUSA poll, you know... the subject of this thread. You can claim they are wrong, but your panicked dismissal and about face on the veracity of SurveyUSAs polling, who you went on at great lengths to tout back when they showed Novick in the lead, if you like, but I must say it is bemusing to see you playing dumb about the treand-line within SurveyUSAs numbers and claim there is no basis to for people to point to the clear movement in Merkley's direction in the SurveyUSA numbers.

  • (Show?)

    "In SUSA, where we get excellent cross-tabs, we can see Jeff's growing success. I'd love to see whatever cross-tabs Hibbits has. They sure show a lot of undecideds (over 40%), but that's okay too."

    I don't know what you mean by "excellent," but it's not a synonym for "precise." As I said, MoEs are 6-7 at a minimum--and for the youth vote it's more like 10+.

    You cannot see Jeff's growing success, is the point--it's not supported by the data. Especially when the undecided has stayed exactly the same.

  • trishka (unverified)
    (Show?)

    you all understand statistics and margin of error, right?

    what the poll data from SUSA show is that merkley may be moving ahead - but he may not be.

    it's a big fat unknown at this point.

  • (Show?)

    "You can claim they are wrong,"

    I can, but I didn't. I have no problem agreeing with the prospect of a +12 Merkley - +4 Novick scenario, any more than I did when it was +10 Novick - +6 Merkley. You are avoiding the point, which is pretty clearly that neither poll establishes a defined lead for either man, and in fact could easily be different sides of the same coin, variant by sheer random differences in who responded.

    The numbers that are the basis of this poll do not show a Merkley surge. They show a tie last time, and a tie this time. And the same proportion undecided.

  • Pat Malach (unverified)
    (Show?)

    Hey lesta, where ya been for so long?

    Hope everything is well.

    Or, as Jeff Merkley might say:

    "...lesta ..be..long ...in... ...h..ell."

  • (Show?)

    Okay, maybe I've been playing too many dark video games in whatever remains of my spare time. But that actually cracked me up Pat. Also, I recently finished reading the lovely Stephen King novella "Apt Pupil." Talk about dark!

    How macabre!

  • (Show?)
    Posted by: torridjoe | May 13, 2008 9:00:31 AM "You are avoiding the point, which is pretty clearly that neither poll establishes a defined lead for either man, and in fact could easily be different sides of the same coin, variant by sheer random differences in who responded."

    Except that is not what is being discussed here. The trend-line shows movement towards Merkley. THAT is what is being discussed and what is shown within the SurveyUSA numbers. Since the Hibbits poll is a singular poll on the race, we cannot deduce any trendline for movmeent from it whether its numbers are accurate or not.

    "The numbers that are the basis of this poll do not show a Merkley surge."

    Except that they do.

    "They show a tie last time, and a tie this time. And the same proportion undecided."

    Clap louder.

    Posted by: Pat Malach | May 13, 2008 9:00:56 AM "Hey lesta, where ya been for so long?"

    Been slammed so hard at work. The corporation I work for had a total rebradning roll-out a few months ago, and so we on the creative services team have been drowning in redesigning everything form business cards, to product brochures and feel to trade-show signage.

    "Hope everything is well."

    Other than being inundated at work, things are going great. Paid off my last credit card and my last car payment so now I am 100% debt free... and Obama is going to be our parties nominee, so all is good right now with me.

    And you?

  • (Show?)
    Which means you can't say much about how they're doing with women, either. If the spread ain't 20 points at least, I wouldn't conclude anything much at all. Crosstabs you can't count on.

    In the NE Hassalo Street sample it is a statistical tie for sure. %^>

    Here's what we know:

    Nobody knew who either of these guys was until they started advertising.

    Steve went on the air first so he got his name recognition up first.

    Jeff came to TV later and got his name recognition up to rough parity, so that in the polls their results were within the margin of error.

    Then Jeff took out a second mortgage to fund a barrage of negative ads.

    But guess what? Now there are two brand name polls, one leaning each way, but both still within the margin of error.

    There are also a lot of undecideds and Steve is on the air with a great new ad that defuses the attack ads and lays out his biggest endorsements.

    We've seen Merkley's worst and we're still in a dead heat, with Steve's best only now coming online.

    It's going to be a fun week.

  • (Show?)

    "The trend-line shows movement towards Merkley."

    Then point it out, specifically. What numbers indicate to you a trend towards Merkley? Spell out your claim. Explain literally how you count out movement towards Merkley.

  • Robert G. Gourley (unverified)
    (Show?)

    I campaigned for Steve at a Clinton event last night and lots of folks told me they'd already voted for Steve - many more than those who said they'd already voted for Merkley. The main demographic about this event was it was well attended by Mexicans.

  • Shirley (unverified)
    (Show?)

    All I have to say is Jeff is for the people, He stood up for the workers at the Hilton with their boycott even though he took at lot of heat from everyone and still surports us for the last 5 and a half months. Talking about the add the one I can''t stand is the one with three guys saying they are Steve and then he comes on saying he's the little guy standing up for the people.

  • Bill R. (unverified)
    (Show?)

    Off-topic- Tim Hibbitts has a KPTV poll out this morning showing Obama ahead by 17 pts. http://www.kptv.com/yourvote/16247266/detail.html

    SUSA has him up by 11 yesterday. I would go with Hibbitts on an Ore. race.

  • (Show?)
    Posted by: torridjoe | May 13, 2008 9:21:07 AM "The trend-line shows movement towards Merkley." Then point it out, specifically. What numbers indicate to you a trend towards Merkley?

    Pretty straight forward, Merkley's numbers improved from the previous SurveyUSA numbers while Novicks and the undecideds went down.

    (5/9-11, likely and actual voters, 4/28-30 in parens): Jeff Merkley (D): 31 (28) Steve Novick (D): 27 (30) Candy Neville (D): 11 (8) David Loera (D): 2 (2) Roger Obrist (D): 2 (4) Pavel Goberman (D): 2 (2) Other/Undecided: 24 (26)

  • (Show?)

    "Jeff Merkley (D): 31 (28) Steve Novick (D): 27 (30) "

    That is literally no more the truth than this:

    Jeff Merkley (D): 28 (31) Steve Novick (D): 30 (27)

    Do you understand the concept of survey error? Both outcomes above are possible.

    And because of that, you cannot say there is a surge for Merkley--not when it is well within the parameters of these polls for Novick to actually be ahead, with Merkley actually having been ahead in the last poll.

    And if you needed any more data to prove that point, we now have an independent poll showing ANOTHER tie, with Novick nominally leading this time.

    If you aren't intellectually honest enough to get "still tied, lots of undecided" out of those figures, either you aren't familiar with survey error or you're just letting hope invade your brain and crush the rationality cells in an epic battle.

  • (Show?)
    Posted by: trishka | May 13, 2008 9:00:21 AM you all understand statistics and margin of error, right? what the poll data from SUSA show is that merkley may be moving ahead - but he may not be. it's a big fat unknown at this point.

    Trishka,

    Your assertion only holds true if you isolate this single poll, cover your eyes and refuse to compare it to the trends revealed by comparing it to previous SurveyUSA polls.

    Furthermore, one needn't have a PhD in statistics to understand that polls use the numbers that they do because they reflect the statistically most likely scenario.

    Yes, Merkley MIGHT actually be up 12 points, and yes, Novick MIGHT actually be up 4 points. But Merkley is statistically MOST LIKELY up by 4 points over Novick - thus the poll stats released yesterday.

  • (Show?)

    "But Merkley is statistically MOST LIKELY up by 4 points over Novick "

    Again--patently false, Mr. Norris. The reported result is NOT the most likely result, it is the midpoint on a range that (at the 95% confidence level) extends about 2 standard deviations away, 1 SD in either direction. (It's actually 1.96 SD, but let's not quibble).

    You don't know what you are talking about, and you are spouting off in front of people who do it for a living. I urge you to stop.

  • Miles (unverified)
    (Show?)

    But Merkley is statistically MOST LIKELY up by 4 points over Novick - thus the poll stats released yesterday.

    Actually, Kevin, that's not true. Polls use single data points because that's the easiest way to explain the poll (and get press coverage). There is no more probability that Merkley has the support of 31% of voters than there is that he has the support of 27% or 35%. In fact, the probability that Merkley is really at 31% is extremely low -- as it would be if we assumed any single data point within that range.

    Assuming they are using 95% confidence intervals, this poll claims with 95% certainty that Merkley's support is between 27% - 35% (likewise, there is a 95% certainty that Novick is between 23% - 31%). The probablity that Merkley is up by 12 points is exactly the same as the probability that he's down by 4 points.

    All of this assumes a valid random sample, which I would quibble with, but that's another post.

  • Miles (unverified)
    (Show?)

    Most interesting to me is the 40%+ that is still undecided in the Davis/Hibbits poll. That number is shockingly high with only one week to go in the campaign. If the poll is right, this may be one of the most volatile races in Oregon's history. It will be fascinating to see if late-deciding voters break towards the experienced establishment candidate or the rookie change agent.

    In my totally invalid sample, things are breaking towards Novick. Even my MOM voted for him, which in my sample means Merkley's in trouble.

  • (Show?)

    Miles, I believe there is a natural asymtotic line away from the midpoint, where the maximum error amount is generally a less likely result than one of lower error. But you're correct that the midpoint is not defined as the most likely result.

    Does anyone else want to hazard a guess how SUSA found a sample where 43% have already turned in their ballot?

  • (Show?)

    should say voted already, not necessarily turned it in...but still, that's like 4 times the actual turnins so far...

  • Dan (unverified)
    (Show?)

    "To ignore that he was the one behind our wins is to be oblivious to what actually occurred. To dismiss or downplay that reality only reveals the person suggesting it doesn't really know what they are talking about."

    Thanks lestatdelc, but I'm very familiar with FuturePAC and was before making my claim (I'm pretty sure that I'm arguably as "in the know" about this as much or more than you). Again, you are giving WAY too much credit to Merkley.

  • (Show?)

    Again, you are giving WAY too much credit to Merkley.

    That would be if you ignored the wretched mess that was FuturePAC in the pre-Merkley years.

    As a guy in an HD that has been Republican for several cycles, and with our current HD candidate Suzanne Van Orman on her second effort, you can bet your ass that I've been all up in FuturePAC bidness for a long time.

    Merkley was absolutely key in revamping, staffing and funding this worse than moribund organization.

    And yeah, we're intelligent enough to notice what's going on at national, but we're also intelligent enough to understand that under other leadership we might well still be in the minority.

    <hr/>

    For the undecided all three of you these are a few of the facts that should weigh heavily in Jeff's favor.

  • Eric Parker (unverified)
    (Show?)

    "The guy that finally figured out how to take back the Oregon House - after sixteen years"

    It should have been sooner - regardless of the republican attitudes.

    Jeff represented me and my district in the Oregon legislature. When he replaced Frank Sheilds, he promised a lot but didn't have the backbone to deliver them when he actually needed to deliver them. It was like he just sat in the back like he was told to so he could 'learn something'. Only when it was safe and in vougue to actually do something (i.e. without retaliation and reprisals from his fellow big wigs), he did. In short - he waited too long to invoke his promises to his constituents. It may look good now, but it should have come sooner. As constituents, we can't wait another batch of years for Jeff to get off his behind and go forward like he did as a rep. I do not think he will act on his promises until he feels it is safe to do them without him getting reprisals and retaliations from other big wigs that are important to him, but by then it will be too late.

    Thats why I feel Merkley is Milquetoast.

    Novick, however, will dive in head first and deal with anything no matter what. He won't wait - he will do.

  • bird (unverified)
    (Show?)

    Would it be too much trouble for the person posting to first say who they support so that the following text can be seen for what it is? I just don't understand someone talking about how one candidate is "surging", yet not really as the margin of error shows it a wash, and then further going on to say that the other candidate actually has the lead.

    I mean, are you kidding? Does the author own this site? I stopped following another site because there was no disclosure about reality. Do this site a favor, put the facts up front.

  • (Show?)

    Let's put this thread to bed... The voters of Oregon will decide this election, not the punditry, get over youselves already.

  • (Show?)

    You folks are underestimating Merkley's appeal across the rest of the state. Novick has weak support outside of Portland.

  • Eric Parker (unverified)
    (Show?)

    That's because, Karen, the Kool Aid is better when it's got more "sugar" in it - especially to those in Eastern Oregon.

    But I digress...

  • (Show?)

    You don't know what you are talking about, and you are spouting off in front of people who do it for a living. I urge you to stop.

    A statistical analysis, properly conducted, is a delicate dissection of uncertainties, a surgery of suppositions. ~M.J. Moroney

    With all due respect, Mark... the fact that you do it for a living is all the more reason for people to analyze the poll themselves. As another Mark famously paraphrased Disraeli, "there are lies, damned lies and statistics."

    Darrell Huff, author of "how to lie with statistics," was more succinct. He coined a word to describe the professional analysis of statistics by one with an axe to grind: statisticulation

  • (Show?)

    "With all due respect, Mark... the fact that you do it for a living is all the more reason for people to analyze the poll themselves. As another Mark famously paraphrased Disraeli, "there are lies, damned lies and statistics.""

    That MIGHT be true if I were interpreting the poll. I am not doing so here; I am offering you information about the science of survey research, that you are happy to look up and verify on your own. It doesn't matter what the results say; the points Miles and I are making are true no matter the numbers.

    Stick to rainsticks. You suck at polling science.

  • LT (unverified)
    (Show?)

    Margin of error with 24% undecided--I would hope even partisans realize that the power to decide this race lies with that 24%.

    Would be interesting to know why Neville took that dip at the end of April and then bounced back.

  • (Show?)

    It doesn't matter what the results say; the points Miles and I are making are true no matter the numbers.

    Stick to rainsticks. You suck at polling science.

    And that, Kevin, is the bottom line. Nothing you can say, no matter how reasonable or logical, will ever be correct or relevant.....because TJ and Miles say so.

    I'm totally persuaded now. I'm off to Gladstone to retreive my ballot and bow to the unassailable arguments of the Novckians.

  • (Show?)

    "And that, Kevin, is the bottom line. Nothing you can say, no matter how reasonable or logical, will ever be correct or relevant.....because TJ and Miles say so."

    or because easily verifiable precepts of polling research say so, take your pick.

  • trishka (unverified)
    (Show?)

    the way i understand it, polls are like a baby's due date.

    baby's are rarely born on their due date. when the doctor assigns a due date, what is being said that there is some statistical probably (can't recall what it is) that the baby will be born sometime in the window between 2 weeks before the due date & 2 weeks after the due date.

    that's the margin of error.

    saying "the baby is going to be born on X day" is ridiculous. it has just as much chance of being born on X-2 weeks, or X+1 week, or any other day in that window.

    same with the polls. saying merkley has 31% support means that there is a statistical likelihood(not 100% but high) that he has somewhere between 31% - MOE and 31% + MOE.

    there's no interpretation going on here. it's just the math.

    oh, and i'm a novick supporter, but the same holds true for the last poll that showed novick ahead. and apparently merkley did have a boost between the poll before that and that one. but between this one and the last one? nope.

    they were tied before and they're still tied.

    and LT nailed it - the election will be decided by those who identified as undecided.

  • (Show?)

    LT brings up a point that I think merits consideration - the listed percentage of Undecideds.

    SurveyUSA lists that number at 24%. Hibbits lists 43% undecided. That is a HUGE difference.

  • (Show?)

    "SurveyUSA lists that number at 24%. Hibbits lists 43% undecided. That is a HUGE difference."

    The difference being 19. Your point?

  • Runtmg (unverified)
    (Show?)

    Really, it comes down to this; Who is voting in this election and why.

    I think that if the election happened in March or April this is Novicks race. But my gut feeling and as TJ will gladly point out elections aren't won by gut feelings I think that Novick has two points that are hurting him regardless of how it is spun.

    The first is, Novick is a bit of a loose cannon. It is one thing to change from politics as usual and quite another to get boxed into hyperbole filled statements that in my opinion Novick has become legendary for.

    The second point is taxes. I don't care how you spin it, people don't want to hear about tax increases. When you go around talking about tax increases regardless of how fair you might feel they are people get nervous.

    Case and point, the gax tax in portland has left major shortages for pdot and odot. Everyone knows that better roads and great infrastructure leads to a better community but because of higher gas prices people are turning to Max and the bus lines which means less money for the maintenance of our roads.

    However, no one will support a change in the gas tax no matter how well intentioned it will be. I give Novick credit for his well meaning attempt to have an honest dialog about taxes and the programs people support but it hasn't helped him in this campaign.

  • (Show?)

    Aside from the fact that for my money, Tim Hibbitts is the best pollster in the Pacific Northwest (disclaimer: Tim is a friend that I go camping with annually), I have serious questions about the methodology of the Survey USA poll. Survey USA uses robo calls, not human callers, to ask its questions. No other major polling firm in the US uses this methodology. I believe it's getting tougher and tougher for pollsters to get good samples, given the fact that so many younger people don't have landlines. Adding the element of recorded calls makes assuring a good sample all the more difficult. I think there have to be huge questions for pollsters about how to get a good sample in an election cycle where so many new voters are showing up.

    I believe Hibbitts--Survey USA not so much. My two cents.

  • Bill Bodden (unverified)
    (Show?)

    Darrell Huff, author of "how to lie with statistics," was more succinct. He coined a word to describe the professional analysis of statistics by one with an axe to grind: statisticulation

    On the other hand, a statistician, Dr. W. Edwards Deming, developed management theories based on statistics that he tried to interest major American corporations in. They turned him down, and he went to Japan where his theories were accepted and put into practice. That led to the Japanese ascendancy in business. He was, and probably still is, regarded as a god in Japan where the Deming prize is as coveted as a Nobel Prize. "The Reckoning" by David Halberstam tells this story, but if you search the Internet for "edwards deming management japan" you will get many shorter articles.

    Mark Twain's, "There are lies, damned lies and statistics" is cute but not true.

  • (Show?)

    TJ: It suggests that late breakers are breaking for Merkley.

  • (Show?)

    "TJ: It suggests that late breakers are breaking for Merkley."

    What does?

  • (Show?)

    "Survey USA uses robo calls, not human callers, to ask its questions. No other major polling firm in the US uses this methodology."

    Rasmussen also does.

    I can't agree that robopolling is unreliable. It's shown its mettle the last couple of elections, and has shown indications it reduces bias rather than adding to it, because people are more honest talking to a machine.

    As much as I'd like to elevate Hibbits on that score, I can't.

  • (Show?)

    I'm sorry, but despite TJ and Miles also being Novick supporters, the MOE point is being spun a bit out of control here.

    <hr/>

    The first point (so to speak) is that though the data aren't perfect, they are data. If we took the arguments being made here literally, there would be no point in polling. The fact that Jeff is up in this one sample means that it is more likely that he really was up than if the numbers were equal or if Steve were up at the time of the poll.

    TJ and Miles are writing as if the probabilities are equally distributed within the MOE / confidence interval. But they are not. That's why a normal curve is higher in the middle than at the ends. Within the interval some results are more likely than others.

    TJ points out that what pollsters call MOE represents what other statisticians call confidence interval. Also that conventionally a standard of 95% confidence is used. Also that the points that bound 95% confidence are found on a normal curve are just under 2 standard deviations from the midpoint.

    (A normal curve is a certain type of shape of curve produced by the distribution of probabilities in a random sample, the famous bell curve; a standard deviation is a measure of spread -- the wider the spread of probabilities, the wider and flatter the curve, and the bigger the standard deviation.)

    What he does not say is that at the points marking out 1 standard deviation. i.e. half the distance from the midpoint of the 95% confidence points, the confidence level is 68%.

    What this means for a poll with a MOE of 4% is that we can have confidence that 95% of a large number of samples would fall within 4% of the midpoint on either side (midpoint=the poll sample %, in this case 31% for Jeff, 27% for Steve), but also that 68% of such samples would fall within 2% of the midpoint.

    So with Jeff 4 points up from Steve, if this were a random sample, it's roughly twice as likely that he was actually ahead of Steve in the population sampled, than that Steve was ahead of him, at the time of the poll.

    The boundaries of the 68% confidence interval would be a tie, on one hand, and Jeff up 4 points, on the other.

    (This is a bit crude about confidence intervals & wouldn't do as an answer on a test in a statistics class, but is o.k. for present purposes).

    The actual data points in a sample do matter, and it does matter with a 4 pt MOE whether the actual data points are the same or 4 pts apart.

    It is still a matter of probabilities, not certainties, and there is a substantial chance that Steve actually was ahead in the voting population a the time of the poll if the vote had been taken then. But given a choice betting whether the actual result would fall within the 68% confidence interval or in 32% confidence range outside it, I'd bet on inside.

    Now this gets mucked up in uncertain ways by non-randomness of sample, as Miles points out, particularly if there's a systematic bias that's related to voting preference distributions. But we don't know if such a problem exists or whether it skews the sample toward Steve, or Jeff (or Candy).

    <hr/>

    On the other hand, the fact that the Hibbetts poll shows the inverse result from the SUSA poll, though they are not exactly comparable due to different sampling methods at different times, gives the "statistical tie" argument more weight.

    It reduces the likelihood that either sample accurately reflected the underlying population distribution of candidate preferences.

  • (Show?)

    Thanks, Chris. I did say in response to Miles that the distribution is asymptotic, to the extent where maximum error is less likely than minimal error as a generality.

    And I've also agreed that while a tie is a tie, you CAN talk about probabilities within that tie. As I said, it's always better to be up or down, for this reason.

    But at root, there's really no way to know whether this poll represents a Merkley lead or a Novick lead. And you simply cannot record a surge from one tie to another--it suggests that both "true" results are actually known. A "surge" by definition would be statistically significant movement, and there wasn't any.

  • (Show?)

    that should say "always better to be up instead of down."

  • (Show?)

    TJ, Apologies, I missed where you said it's always better to be up instead of down. Bad reading on my part.

    But now you've said it at least three times, so maybe folks will be clear that you mean it. :->

    Also, "The boundaries of the 68% confidence interval would be a tie, on one hand, and Jeff up 4 points, on the other," should have been "Jeff up 8 points on the other."

    <hr/>

    Do you have any insight into how much of a large, late undecided poll population ends up not voting? It will be interesting to see if there's an undervote for the Senate primary compared to Obama vs. Clinton.

    I guess it could mean a lot of people having a hard time choosing between two good progressive candidates, but it feels more like whoever wins is going to have to keep doing a lot of name recognition / identity establishment work.

  • Miles (unverified)
    (Show?)

    Okay, my econometrics training was over 10 years ago, and I overstated the case in my above post. (And TJ did immediately point that out.) But to be honest with both of you, I don't recall the probability of any particular number within the MOE being on an asymptotic curve relative to the midpoint. I could certainly be wrong. My understanding is that the confidence interval represents the band within which sampling variability affects the results. So in 95 out of 100 samples, Merkley's number would be between 27%-35% (and in 68 out of 100, it would be between 29%-33%). But in any one sample, can you really say that it's more likely to be 31% than 27% or 35%?

    As for an overall trend line, aren't there only three data points here? As such, given the MOE, I find it hard to put much faith in a "trend". Kari's graph is pretty, but it's misleading since you could straighten out or even reverse either of the candidates' trend lines based on pure statistical variability. If Survey USA were doing weekly polling that showed a trend, that's a different story.

    Finally, as Chris points out, the Davis/Hibbits poll showing the opposite results means that one of these polls has serious sampling problems. And I guess us partisans will just have to wait another week for the truth.

  • Robert G. Gourley (unverified)
    (Show?)

    My bet is Corvallis will go for Steve, based upon two times handing out literature to folks standing in line to see Bill Clinton. Most of the folks who said they voted for the other guy didn't even know jeff's name. Driving around town you hardly see any Merkley signs, and while Steve's could hardly be called "all over the place" they are many times Jeff's. I look forward to seeing whether I'm right.

  • (Show?)

    I am way behind on this post, and don't have time to read the 70 comments, but apropos of the Chris/TJ discussion (interesting and well-stated), I'll add this: the biggest problem pollsters have had in this election is identifying exactly who will turn out. The discussion immediately above talks about the math, but the big issue involves the assumptions the pollsters make when they weight the sample to reflect the expected voting population.

    Generally, polls are pretty accurate. In past election cycles, taken in the aggregate, they got things right. That's why pollsters were so stunned in the early primaries when they kept getting it spectacularly wrong (NH was a moment of truth, when pollsters appeared to have lost their confidence in the laws of the universe).

    The caveat in these or any polls in 2008 is: who will show? Pollsters are getting better; there have been 46 primaries already, so they're refining their methodology. But even in NC and Indiana, they went pretty haywire. So, caveat emptor. And of course, it makes watching the returns a hell of a lot more interesting.

  • (Show?)

    Miles,

    You're remembering half the story, but Chris describes the whole story. The standard error is based on randomized measurement error around a "true" but unknown mean value. If the errors are generated by a truly random process, the estimated value deviates from the true value with probability described by the normal curve.

    So we can say, assuming all the assumptions hold, that 65% of the time, the estimated value +- 1 SD contains the true value; 95% of the time, the estimated value +- 1.96 SD contains the true value.

    What you remember correctly is that you can't say with certainty that the "true" value is 31%. The probability that the true value is any precise number is vanishingly small. You can say, however, that the true value is more likely to be closer to your estimated value (that is, in the range 29-33) then, say, in the range 27-35.

    ===

    Jeff, I also didn't wade through all the comments carefully, but to my eyes, two things pop out:

    1) Merkley is coming on (TJ you are not correct to state that we can't infer anything from two or more polls in a row--think of these as two random throws of a die, yes, seven is equally probable in each throw, but 2 sevens in a row is less probable, and 3 is less probable, etc. Same goes for a poll trend)

    2) BUT the bigger story is how low Merkley's recognition / support levels were three months ago. I don't care much that he wasn't advertising--he's the head of the state senate, and (with all due respect to Steve, for whom I have great respect, and whose run a great campaign), some City Club attorney from Portland is out polling him by 15 points a month before the election? Wow.

  • (Show?)

    don't care much that he wasn't advertising--he's the head of the state senate...

    And if we needed any more information about how much that position is in the public spotlight, it's worth noting that the head of the Reed Poli Sci Department appears to be confusing President of the Senate, Peter Courtney with the House Speaker, Jeff Merkley.

  • not piled high (unverified)
    (Show?)

    Just a brain fart on the prof's part I'm sure. But that's what's best about Merkley supporters lik PH D above. They'll attack anyone!

    Hillary/Merkley '08 !!!

  • trishka (unverified)
    (Show?)

    robert, you're in corvallis as well? that's my take on it, too. i've seen one merkley lawn sign, and too many novick signs to count. plus the response when i rang doorbells was either "i don't know anything about the senate race" or "i'm voting for novick".

  • (Show?)

    I don't care much that he wasn't advertising--he's the head of the state senate, and ...

    I once refered to Paul as a prof at Lewis and Clark--transposing my alma mater for that other liberal arts college...grrr, damn Reedlies--and the good professor took great umbrage. Well, maybe if he were at LC, he'd know better. :-0

connect with blueoregon