My biases are well-known in this venue, but I think the real story here is that none of the "experts" anticipated that as of late February Steve Novick would still be pacing Jeff Merkley, much less potentially slightly ahead.

Steve is a much stronger candidate than the pundits were willing to give him credit for.

With all the structural and economic advantages Merkley's campaign has enjoyed, they should have put Steve away by now. The fact that they haven't, and it appears they just can't, gives me great encouragement.

I attribute much of this to the distinctions visible between them on certain issues, and the rest to Steve and Jeff personally in terms of style and approach.

"The only noteworthy conclusion you can draw from all of these polls (on the Senate race) is they all have Smith under 50 percent," he said.

I agree with Mapes. The fact that Smith's numbers didn't budge with either match-up is indicative of both his weakness and of a general lack of name recognition for the Dems. A generic poll pitting Smith against a nameless Democratic opponent very likely would have yielded nearly identical numbers - 48% for Smith and 30-35% for the nameless Dem.

I want to preemptively chime in here that this is my post, and I had hoped to put this up a few days ago but have been slammed with work, ect. So, hopefully better late than never...

In a year widely expected to be a Democratic clean sweep and repudiation an unpopular, corrupt presidency and its fawning supporters in Congress, the most baffling thing here is that the Bush-loving incumbent tops either challenger by more than 10%.

It's going to take real resources ($$$+feet) after the primary to launch a real challenge, even if the current poll merely reflects Smith's greater name recognition.

I think either Merkley or Novick will have no trouble getting the support he needs after the primary.

Merkley, because it has already been promised to him by so many parts of the Democratic establishment.

Novick, because if he shows he can beat Merkley in a primary, he will be a HUGE story and will have shown he can win.

Steve Novick clearly is a distinct change motivated by principle and the law and not by playing political games that have gotten this nation in its current mess.

Except that a variety of ethically questionable issues directly linked to his campaign seem to indicate that his campaign at least is very comfortable playing political games.

Novick staffer - caught altering Merkley's Wiki page - Novick campaign says nothing.

Novick staffer - caught allegedly trying to game an organization's endorsement for her boss, has apparently been removed from any formal role with said organization - Novick campaign says nothing.

Novick staffer - pointedly declined to practice disclosure of anything beyond her first name in response to a direct question - Novick watches, says nothing.

Novick - has repeatedly coopted GOP-originated smears on Jeff Merkley.

By contrast...

Merkley staffers - zero allegations of ethical malfeasance of any sort.

Merkley - has not attempted to criticize, let alone smear, Steve Novick in any way.

One of those two campaign profiles appear to be consistent with age-old political gamesmanship

Here we go again.

Novick staffer - caught altering Merkley's Wiki page - Novick campaign says nothing.

You know, the whole point of wikipedia is that it is open and editable by anyone.

It is also transparent. That's how we know who edited what, and when.

I've read through the Merkley article history and I have to say that I think Henry's version was at least as honest as the current "official" version, which omits from the text of the article the extremely inflammatory preamble to the resolution (all the whereases).

Kevin has expressed repeatedly that he believes that Jeff Merkley handled HR2 in the exact manner Kevin considers appropriate. That's nice for Kevin. For many of the rest of us it is still an issue and the preambles are a big part of the problem (not the totality of the problem, just a big part of it). So I would consider Henry's handiwork to be more in the nature of "full disclosure" than today's version. Among other things, Henry's version was honest enough to include an ellipsis to indicate where one clause was omitted. The current version says only,

In 2003, Merkley voted for Oregon House Resolution 2, which resolved: “ [we] (1) Acknowledge the courage of President George W. Bush, the President's cabinet and the men and women of the Armed Forces of the United States, and express our support for the victorious removal of Saddam Hussein from power; and (2) Praise the courage, dedication, professionalism and sacrifices of the men and women of the Armed Forces of the United States and their families in the defense of freedom.

That's plenty bad enough, but not complete, and it is masquerading as the complete text of the bill.

And just for fun, here's the whole text again, because I just never get tired of mocking it.

HOUSE RESOLUTION Whereas the dictatorship of Iraq has continued to develop weapons of mass destruction in violation of United Nations Security Council Resolution 1441; and Whereas the dictator Saddam Hussein has demonstrated a willingness to use weapons of mass destruction against neighboring nations and the citizens of Iraq; and Whereas Saddam Hussein threatens the Middle East and the global economy with the threat to use weapons of mass destruction; now, therefore, Be It Resolved by the House of Representatives of the State of Oregon: That we, the members of the House of Representatives of the Seventy-second Legislative Assembly: (1) Acknowledge the courage of President George W. Bush, the President's cabinet and the men and women of the Armed Forces of the United States, and express our support for the victorious removal of Saddam Hussein from power; and (2) Praise the courage, dedication, professionalism and sacrifices of the men and women of the Armed Forces of the United States and their families in the defense of freedom.

Oh, and I can't leave this alone either.

Merkley staffers - zero allegations of ethical malfeasance of any sort. Merkley - has not attempted to criticize, let alone smear, Steve Novick in any way.

I know a citation for reckless driving (albeit with the candidate in the car) is not normally what we think of as an ethics violation, but it does suggest a kind of "the rules don't apply to us" approach to life and campaigning.

As for Merkley's personal expressions, they are pablum for the most part, for sure. But he has Mitch Greenlick, Mary Nolan, Russ Kelley, Kari Chisholm, and a whole echo chamber of insta-blog surrogates to do his dirty work for him. It's pretty transparent, really.

Further, his silent departure from fundraising on ActBlue (which is good enough for countless other Democratic candidates) suggests that the real reason was not the credit card processing fees but rather the degree of transparency to which his campaign was being subjected with every dollar it raised (or failed to raise) online, and the ready comparisons with the Novick campaign and other Senate campaigns.

Once again, beware of any two way polling on this race because of the Frohnmayer factor.

That's a valid point, Becky. However, I don't believe that Frohnmayer will be a factor. His fundraising has been, to put it very generously, anemic. He can't siphon off votes from folks who don't even know he's running.

for the poll geeks, at what point does a statistically small difference become meaningful when it is continually ahead on the side of one cadidate, and behind on behalf of the other?

Good questions, Petrichor.

The Riley Research "omnibus survey" (his own characterization of it) isn't worth the effort it took to type this sentence for methodological reasons. So that leaves this Rasmussen poll and the much earlier SurveyUSA poll. Both polls show no change in Smith's numbers regardless of who he's pitted against. Honestly, I don't see any other reasonable way to read that other than that both polls indicate very low name recognition for his challengers, to the point that both were essentially pitting Smith against a nameless Democrat.

Clearly whomever wins the Democratic Primary will absolutely have to get their name recognition up... way up in order to be competitive. It will be particularly crucial to get name recognition up within the swing voter demographic. Which will mean spending $$$. Which is why fundraising stats are always a central part of analyzing horse races.

Mapes point is quite valid there is no real difference between the two. Think about it. Especially since Novick has been spending bank on TV and Merkley hasn't, his slight numerical advantage is not surprising at all. Everyone that is paying attention at this point knows that Smith is bad for the state and are more highly informed voters than those who will Novick has the most appeal. Try handing a east Salem Democrat a voters pamphlet on May 2nd of someone who never has cast a public vote or held elected office. See if they support Novick. The polls so far don't capture Merkley's greatest advantage over Novick, his appeal to less than high information voters outside the peoples republic of portland. Furthermore, How can people claim that Merkley should have "put Novick away" by now when there hasn't been any legitimate polling head to head in a primary. Vrs smith polling has nothing to do with who will win the primary and considering the quite high level of undecideds, this poll still looks like generic D vs Smith to me.

Rank amateurs like Kevin and bdunn are exactly the WRONG people to ask about polling; they show a consistent lack of understanding and pure intent to distort and create facts that do not exist. For instance, they repeatedly alleged that the Riley poll didn't rotate candidate order, without explaining how they came to believe that, and on what basis. (Kevin also appears confused by the term "omnibus," which simply means they covered several different topics.) They don't appear to understand what a margin of error is, and how to use it, calling a larger MoE "sloppy" as if it's evidence of poor methodology or bad science--as opposed to a simple probability construct that tells you how wide a range brings you to 95% confidence that the results aren't random. And they (and Kari) continue to assail the Riley poll as "useless" or "flawed" without ever coming up with something within it that makes it flawed, opting instead for vague allegations based on analysis that Merkley's campaign manager doesn't like.

Bdunn continues the idoicy by saying that this result must be because of the ads, conveniently forgetting that Novick was similarly ahead in two prior polls done BEFORE any ads were produced or aired. So were voters in those polls responding to Novick based on ads they'd see a few months later? He further makes a wild claim that the polls don't show his similarly unsubstantiated "appeal with low information voters," as if those voters were somehow excised from the poll. If they're low information voters, either they're not going to vote--which isn't much help to Merkley--or they are, in which case they're included in the poll results just like everyone else. Same is true of "voters outside Portland"--it's a statewide poll, Brad.

Kevin makes one factual error--SUSA did show a different number for Smith when paired against Novick compared to Merkley--and one big analytical misstep: judging the relative appeal of the challengers based on the number for the incumbent. Now, in SUSA's poll the number for the CHALLENGER didn't change, and indeed one might suspect a level of "generic Dem" support at play--since voters don't seem to show any manifest difference in their evaluations of them.

But that's not the case here; Novick draws a bigger number against Smith than Merkley. He also is better liked, something which has nothing to do with Smith at all...and would not show up if the "generic" effect were dominant.

To answer petrichor's question, you can't make any direct comparisons between polls, or make any conclusive assumptions if you have a series of results where one candidate leads, but is within the MoE each time. Theoretically, the "truth" might be that the error goes the same way each time, and Candidate A is actually behind Candidate B in every poll. All MoE means, again, is that one can be 95% sure the results aren't random, within the parameter specified by the MoE.

That said, the probability of repeated errors in the same direction (or repeated invocations of that other 5%) drops with each successive poll. In other words, you can't say specifically that 4 polls showing Novick ahead means he must BE ahead...but with each additional poll showing the same candidate order in the same relative positioning, the probability that they are ALL wrong drops a little bit.

The bottom line is that the Lake polling, SUSA/Roll Call, Riley and Rasmussen show a stable race--Smith leads among those who know the candidates, most don't know the Democrats very well...but for those who do, the race is dead even, with a decent possibility that Novick is slightly ahead.

I think we can bring Smith's 48% approval number down. What do you think folks? The truth is both Merkley and Novick will represent Oregonians the way we all deserve to be represented. I have chosen to support Merkley because he's proved to me what kind of Senator he'll be by the progressive legislation he's pushed in the House. I do not think that Novick should have been put "away" by now, it's still pretty early in the race. I do believe that the Merkley Campaign is building a good, solid base to really take Smith on. The labor support, fundraising and messaging is spot on, and we're going to have to run a very strong campaign to unseat Smith.

Petrichor... We edit headlines all the time, after discussions amongst the editors - for lots of reasons. This time, it was because that second part of the headline didn't reflect Mapes's post accurately.

I think it's a reasonable characterization, but it wasn't Mapes's characterization.

It's not Mapes' headline Kari; it's yours. You really have to squint to not make the fairly reasonable claim that Novic k came out ahead in this poll.

So, Novick has been up on the air with two ads - one of them some people are calling the greatest ad in the history of politics - and this is all that he has to show for it?

You need to coordinate with the Merkley folks, who are trying to say the ads DID have an effect. :)

Somehow, as great as the ads are, I have a hard time believing a couple of points laid down on airtime (I'd be shocked if it cost even a third of what the DSCC has given Merkley so far) would be enough to show up in the results. The fact that they are generally the same as results published prior to the ads also puts the proposition into question. NONE of it--endorsements, national party money, ad buys or viral exposure to them--has moved the race from where it was a couple months ago: dead even.

Posted by: torridjoe | Feb 21, 2008 11:56:27 AM

I humbly beg your highness's pardon but, while I aknowledge your superior to me in every conceivable way, your brilliantly shining intellect and vast wisdom in all things, the answers you claim don't exist are actually very easy to find and have actually been given to you multiple times.

How Riley Research did their survey is all explained in the crosstabs of the survey itself, here.

What an Omnibus Survey is can easily be understood simply by reading how the experts describe it:

Omnibus surveys provide those seeking information about markets and opinions with a means to get quick, relatively low cost answers to their questions without financing and organising a full market or opinion research survey themselves. - What is an Omnibus survey?

Even more interesting is the Wikipedia entry on it:

Usually, multiple research clients will provide proprietary content for the survey (paying to 'get on the omnibus'), while sharing the common demographic data collected from each respondent. - Omnibus (survey) I leave it to his supreme highness, torridjoe, to condescendingly explain the stunningly obvious to us all as if none of us grasp the very language we've used every day of our lives.

Kevin makes one factual error--SUSA did show a different number for Smith when paired against Novick compared to Merkley

'Tis true. I foolishly went from my notoriously faulty memory even though I was tempted to track it down and double-check before hitting "enter." My bad. As you point out, the constant number was for the challengers. See, my memory wasn't totally faulty, just a bit dislexic.

I'm curious who paid for and supplied the rigged question to Riley?

I humbly beg your highness's pardon but, while I aknowledge your superior to me in every conceivable way, your brilliantly shining intellect and vast wisdom in all things,

you're the one extrapolating to "all things" beyond polling science, which was the focus of my Master's thesis and the first ten years of my career in statistical analysis.

I understand perfectly how Riley conducted his poll. I've read the methodology. I'm asking what you base your claims on to say it's faulty, specifically. You haven't. You can start by explaining why you think the candidate names weren't rotated.

If you know what an omnibus survey is--what I said it was, a survey comprising several different topics--why were you so confused by Riley's use of the word?

"rigged question"

WTF?

TJ before calling others rank amateurs when was the last time you were employed by a polling firm? Or a political campaign?

You conflate the high margin of error with so many other reasons to look at anyone who relies on its authority as either high or having an agenda to push. 1 Reilly's polls have shown to not be accurate in the least like suggesting into October that Saxton would beat Ted despite Ted ending up with a comfortable win in 2006. Secondly Reilly didn't rotate the list names that he read off of always putting Novick first, a huge advantage. Third the margin of error was significant in showing a wide range of possible meanings for the poll. Similarly a poll showing that 3/4 voters are undecided isn't particularly reliable.

My point about lower information voters is that many of them will make their decision on "election day" based off of what info they recieve on the voters pamphlet where Merkley will have a distinct advantage with his resume. Does Novick have elected legislative experience? No. Foreign policy experience? Endorsements from other electeds voters trust? No. Merkley? Yes on all counts. The poll does not capture that dynamic.

Furthermore Merkley's bank account will allow him to define himself to using his distinct fundraising advantage during the primary which will raise his name id and percentage, something that Novick will be less able to do.

Petrichor... Those posts were over my own name. Much different standard applied there.

When our contributors write for themselves, they can write whatever headlines they want.

There's always an explanation, isn't there?

you're the one extrapolating to "all things" beyond polling science, which was the focus of my Master's thesis and the first ten years of my career in statistical analysis.

Oh my!! I am impressed!!

Seriously... WOW!!!

I hereby aknowledge you as THE expert in all things having to do with statisticulation.

There's not sufficient space on my monitor for me to properly grovel but this is me groveling abjectly at your perfect, statistically analyzed feet...

Oh, by the way... Have you ever read How To Lie With Statistics by Darrell Huff? What with me lacking your god-like mastery of statisticulation, I dare not comment on it but surely you can spare a crumb or two of enlightenment for us ignorant folk.

"TJ before calling others rank amateurs when was the last time you were employed by a polling firm? Or a political campaign?"

1. I spent a decade working for a survey research outfit in Virginia. I was hired by and mentored under the now-current Associate Director at Pew Research. One of my recurring duties was to take sections of the omnibus polls we would run, analyze the data and write the report on it for the client.

How about you?

Secondly Reilly didn't rotate the list names that he read off of always putting Novick first, a huge advantage.

And you make this claim based on what, exactly? (It's also not a "huge" advantage; typically it's been figured as worth 2-3 pts at best, and is usually exacerbated by a large gap between name IDs; there's no problem with that here).

Similarly a poll showing that 3/4 voters are undecided isn't particularly reliable.

"Reliability" meaning whether the results are repeatable, I'm not sure what you think you're saying here, since the results are manifestly repeatable...having been repeated a couple of times already.

Endorsements from other electeds voters trust?

I think Erik Sten, Randy Leonard, and Pete Sorenson (among others) are likely to be fairly insulted by your characterization, that voters don't trust them.

Yes, petrichor, if Charlie had written a post chock-full of his own commentary explaining how he thinks this poll shows Novick ahead, a headline that says, "Novick Leads!" would be perfectly reasonable.

But Mapes's post didn't say that. In fact, it said the exact opposite.

And yes, BCM, I should probably disclaim again here: My company built Jeff Merkley's website, but I speak here only for myself.

I'm done with the meta-chatter on this one.

If you think this poll shows Novick leading, by all means, make the case!

The trick to closing tags here is to type in the closing tag first and then type whatever you wanna say.

Adding a space to nullify the closing tag, here's what I typed above:

< / i >The trick to closing tags here is to type in the closing tag first and then type whatever you wanna say.

Oh my!! I am impressed!!

Seriously... WOW!!!

I hereby aknowledge you as THE expert in all things having to do with statisticulation.

"he thinks this poll shows Novick ahead"

Not only does he think that, that's exactly what it DOES show, numerically. Whether you think the result is statistically significant or not, the result does indeed show Novick ahead. It's not a matter of opinion, but subtraction (35-30=5).

Especially if they're doing so with a political axe to grind.

::: Irony Meter explodes into a million tiny pieces :::

It's now obvious that Kari is so deep in the tank for the Merkley that he eschews even the most minor of points favoring Steve Novick.

Well, in all seriousness, that's been obvious for many months.

Having said that, I don't think that Kari is some kind of evil mastermind.

I do think that Kari, a little like John McCain in the news today, is so certain of his personal rectitude that he just can't see that certain words and actions create an adverse impression of said personal rectitude.

Unfortunately, he's going to have this problem a lot as long as he maintains his dual role as candidate partisan and blog owner.

I believe the answer is the former, but I'm guessing. I'm sure TJ will be along in a little while to answer your question based on, you know, actual professional knowledge of how polling works.

I'm not sure I agree that the result is "not statistically significant." It may or may not be politically significant at this point, since it's still February, but this is the first poll I've seen that showed either candidate with a lead that was wholly outside the margin of error.

With the poll under discussion here, my understanding would mean that (within the confidence of the poll) Novick's support is between 30.5% and 39.5% and Merkley's is between 25.5% and 34.5%--the difference is not statistically signficant.

This is correct--the MoE is applied to each data point in this situation. What I usually do is take each candidate's maximum potential lead within the MoE and make that the range. In this case, that'd be Novick +14 (39.5 - 25.5) to Merkley +4 (34.5 - 30.5).

In this fashion, you get a better sense of not only how wide the range gap really is--19 points overall, including a tie--but also where the tilt is, so to speak. Assuming an equal chance for the "true" lead to be anywhere on that line, just under 75% of the possible outcomes are Novick actually being ahead. (I think it's probably more like a curve than equal shares, but they're not different enough to make a difference here).

The best shorthand IMO is to take the MoE, double it, and compare it to the gap between the candidates. If the latter is larger than the former, you should be 95% confident that they really are ahead. Otherwise, take varying degrees of skepticism and speculation, depending on the difference between MoEx2 and the preference gap.

I'm sorry, I'm not trying to be Mr. Professor, and I've been out of the game for a while and never was one of the hardcore math geeks even then. But I hate, hate hate seeing people grossly misinterpret polling data, especially when they do it with a political purpose. If you come by your misunderstanding honestly, it's all cool. I don't think that's the case in some instances, here.

Rasmussen does good work--they NAILED the Governor's race in 2006--and their result fits in well with the rest of the polling.