Wildly divergent polls in the Mayor's race

Kari Chisholm FacebookTwitterWebsite

Well, it's hard to figure out what to make of this. This evening, two polls in the Mayor's race were released. They're so divergent, they might as well be from parallel universes.

One was conducted by SurveyUSA on behalf of KATU-2. The other was conducted by Elway Research on behalf of KGW-8 and the Oregonian.

DatePollBradyHalesSmithUndOther
2012-05-07SurveyUSA282527155 - "another candidate"
2012-05-06Elway Research162928186 - "other", 4 - "won't say"

That's right - one has Eileen Brady leading and Charlie Hales losing a statistical three-way tie, and the other has Hales winning a statistical two-way tie and Brady way behind. Go figure. I'm guessing that the Jefferson Smith camp is thrilled - since they're in second place in both.

A few additional data points for you to mull over:

My take? I think it's pretty hard to look at that 16% for Brady in the Elway poll and consider it anything but an extreme outlier. Six consecutive three-candidate polls over six months have had her between 21 and 26%. And with SurveyUSA pegging the already-voted group as being at 36% for Brady, the Elway number just doesn't make any sense. (And one more thing - I've heard a number of reports of some polling firm employing live callers who ask about an "Ee-leen Brady". No idea who it is, but a consistent mispronunciation would tend to skew the numbers in that poll.)

Here's one final data point for you: Today, Multnomah County elections reports that voter turnout is at just 8.77%, relatively low for this point of the election.

So, regardless of what you think of the polls, this much is true: The winning campaign is going to be the one that gets the most communication out to voters in the closing weekend - over the air, in the mailboxes, on the phone, and on the doorstep. If you've got a rooting interest, get to work and make a big donation. It'll all come down to the wire.

Here's the full rundown of all polling released in the race:

DatePollBradyHalesSmithUndOther
2012-05-07SurveyUSA282527155 - "another candidate"
2012-05-06Elway Research162928186 - "other", 4 - "won't say"
2012-04-28DHM Research232520286
2012-04-13SurveyUSA342215219 - "another candidate"
2012-03-26Hales for Mayor21211139
2012-02-27SurveyUSA251610287 - Arrow, 3 - Dant, 2 - Brumm, 8 - "another candidate"
2011-11-07SurveyUSA23191444
2011-11-07SurveyUSA1913113820 - Reese
2011-10-30Brady for Mayor26161048
2011-10-02Portland Business Alliance151394222 - "one of the other candidates"
  • (Show?)

    Full disclosure: My firm built Eileen Brady's campaign website. I speak only for myself.

  • (Show?)

    Here are the two big news stories on the latest polling from KGW and The Oregonian.

    • (Show?)

      Hey Spencer, Wouldn't that be one big news story since KGW and the O have the same poll? If you want to share two stories, which is always a good idea, then shouldn't you share the KATU poll? Thanks.

  • (Show?)

    Wow! What a difference. I'm inclined to go with Survey this time. Larger sample and in the current election season they've been fairly close. Their count on returned ballots bodes well for Brady. Looks to me like a Brady-Smith runoff for the fall.

    • (Show?)

      Stephanie, I'm still waiting for any substance to come from Jefferson. Please let me know what his plans are. So far I've heard opposing the CRC, vaguely talking about growing small businesses, and including East Portland. Sounds great, but that wouldn't qualify as substance.

  • (Show?)

    Graphed: https://docs.google.com/spreadsheet/oimg?key=0AnnQYxO_nUTWdGI0Y0VKV0NyS2NwQTdjbTVXZmwzZWc&oid=2&zx=fc0265snu83s

    • (Show?)

      For people who want that graph hotlinked.

        • (Show?)

          So we're picking our Mayor solely based on debates? That should work out well.

          • (Show?)

            No, the debates brought out her positions more clearly. It was not her debating style, which was professional. I think people disagreed with a number of her positions when the contrast was easier.

  • (Show?)

    you can't really compare the 36% from one survey with the 16% from another survey. It's 36%/28% in one. That's sort of interesting. If 22% have voted and Brady has gotten 36% of those it means, according to the SUSA poll, that she's favored by about 26% of those who haven't voted yet. That's quite a bit higher than 16% but just as much lower than 36%. In other words, it seems she's gotten her most enthusiastic voters to send in ballots early and that her support isn't quite as strong from here on in.

    Still, having votes in is a very good thing. If Smith and Hales have more support among those who haven't voted, and if for some reason those ballots don't get in, that doesn't help either much.

  • (Show?)

    Came across something fascinating that didn't really fit into the post.

    This spring, Nate Silver noted that the period 7 to 14 days out are a "disaster" for accuracy, with accuracy improving both closer and farther out.

    He was talking about presidential polling (and, of course, Oregon has vote-by-mail), but I wouldn't be surprised if a similar effect exists.

    • (Show?)

      Very similar to what I've observed. For myself personally and a lot of my friends the more they've heard from Brady the more they tend to turn to Hales or Smith. My friends that really like experience turn to Hales but the people that like vision and forward thinking seem to turn to Smith, but both groups seems to really like Smith's message regardless of how they're voting. Brady has just seemed milquetoast to most people I've talked to about the race.

  • (Show?)

    The Survey USA poll results from its 4/13 poll ought to be included in your compilation of poll results: It shows Brady - Hales - Smith at 34 - 22 - 15% respectively; now (5/7) the Survey USA poll has them at 28 - 25 - 27%. So the fall by Brady and "surge" by Smith which appears also in the Oregonian data is quite real. And it indeed does "make sense" for the 8.77% (already voted) sample voting for Brady to be at 36% at the same time the Elway poll says her support (including the 91% not voted yet) to be at 16% (May 6th), so you obviously put an editorial "spin" on that comment. Sure, polls are polls; but spin is spin, ok? Here's the missing link to the 4/13 Survey USA poll you omitted: http://www.surveyusa.com/client/PollReport.aspx?g=9460b098-a757-4fb3-8d78-e5ca885338ab

    • (Show?)

      Hey, check that out! That 4/13 poll from SurveyUSA was news to me. I'm not finding any news story about it -- and I would have thought a 34% showing by Brady would have reported by someone.

      In any case, it's now on the chart. Thanks!

      • (Show?)

        It's really neither here nor there, but the more I look into this April 13 poll, the more curious I get.

        It appears that no one reported that poll on or around April 13. Not even KATU.

        It looks like they got an anomalous poll and sat on it. Which is perhaps something the Oregonian and KGW should have done here, too.

    • (Show?)

      Oh, and as for your math. The 36% "already voted" group for Brady comes from the 22% of respondents that told SurveyUSA that they had already voted.

      "Already voted" is a far different thing than "the ballot is at Multnomah County elections." Typically, after someone votes, the ballot sits on the dining room table for a while, then it goes in the mailbox for a while, then it's in transit, then it arrives and sits in a box at elections before getting barcoded.

      Perfectly reasonable that Multnomah County has 8.77% of the ballots while 22% of voters say they've voted.

  • (Show?)

    I can't claim knowledge what "already voted" means, Kari, but your theory seems a bit stretched, no worries, just another observation on the "spin" of your math. My ballot is sitting on my side table, mayor's race is marked and ballot is signed; some races aren't marked. If I were polled I wouldn't claim I have voted, and I think people understand the act of voting in Oregon is more than "I marked my ballot." Voting, to me, is the moment I commit the ballot to the mailbox or drop it off at Elections!

    • (Show?)

      I think people understand the act of voting in Oregon is more than "I marked my ballot."

      Well, clearly not. If 22% say they've vote, but only 9% of ballots are in and coded by the county, there's clearly a 13% gap there.

      And even if you're not counting the ballot sitting on the table, you're probably counting the ballot you've put in the mailbox. But that's still a 2-3 day gap before it gets delivered, coded and released in the daily count.

      • (Show?)

        There's not really a gap though because those are two different measurements. I presume the 9% returned ballot rate is out of all the ballots that were mailed. The SurveyUSA poll is surveying likely voters; a subset of the general population whom presumably are more likely to turn their ballots in early.

  • (Show?)

    unnoted throughout this discussion: the Survey USA poll started & completed AFTER the Elway poll. in other words, whatever Elway might have been measuring was already in the process of changing - which Survey USA showed.

    this is like the Tour de France: Smith made a breakaway but didn't have the power to sustain. Brady has reeled him back in. once again, it comes back to which team can pull their lead rider to the end.

    (disclaimer: i work part-time for the Brady campaign.)

  • (Show?)

    In this election there will be a winner and loser of pollsters as well. For myself, the best, most accurate, most consistent pollster around is PPP.

  • (Show?)

    there is no problem with the 9% vote in versus the 22% who said they've already voted.

    Simply, the poll with the 22% vote has an inordinate amount of early voters who are more enthusiastic about their candidates.

    There's no reason why a group of 500 should have returned their ballots at exactly the same rate as the general population.

    • (Show?)

      Sure there is. If the random sampling was done correctly, then the group of 500 should have returned their ballots at roughly (nothing is exact, this is all statistics) the same rate as the general population. If the random sampling wasn't done correctly, then the poll results are worthless.

  • (Show?)

    I already commented on the Survey USA poll for KATU and may give comments in a while on the dueling polls for KGW.

    I have a few comments from the analyst's perch. I would not disagree with the general sentiments expressed here:

    (1) Smith is clearly surging but is going to have to rely on a big GOTV effort for a tough portion of the electorate (2) Brady's support seems to be stalling or eroding but she could hold on (3) There remains a great deal of uncertainty, both in the poll estimates and I believe in the electorate.

    Some other disorganized thoughts:

    • I am generally a skeptic of robo-polls like Survey USA but they have the advantage of lots of repeated measurements, so while I may not bet on the specific numbers, comparisons between the two Survey USA polls can be made ASSUMING similar response / refusal rates. It's very difficult to compare Elway with Survey USA without knowing a LOT more about their methodology. Personally, I'd go with the local experts--DHM--because they have an investment in this state and in getting it right. It's encouraging that their numbers and Survey USA numbers are similar.

    • I am not sure how relevant Silver's comments regarding presidential primaries are to these results. The level of interest and coverage are simply light years apart. The one similarity is that these three candidates are not that different in my mind, and I suspect in many voters' minds, and that can lead to volatility. If you disagree with that statement, then the Silver commentary becomes less applicable.

    • Many have claimed that the undecideds are breaking for Smith. That is possible, but it is also possible that there is candidate switching going on--that's almost certain given Brady's early numbers.

    • Hopes for Brady: her supporters identify the "economy" and "schools" as important issues. These issues are strong motivators. She has the most "banked" votes (if you believe Survey USA).
      Worries for Brady: apparent erosion in support, particularly among women.

    • Hopes for Hales: strongest support among > 50 who turnout at high rates, similar issues to Brady.
      Worries for Hales: he's holding steady but doesn't seem to be picking up support. not sure what to make of the cell phone / landline number.

    • Hope for Smith: the trend line is very impressive.
      Worries for Smith: his core support is young people who tend not to vote, particularly in primaries. The core issue for his supporters is "transportation" but only 5% of the respondents identify that as the top issue. Has he taken too longto break for the front?

    • (Show?)

      Good notes, Paul.

      Only thing I'd note is that the only comment on this thread from Nate Silver about presidential primaries is the random aside I posted in a comment.

      Per Silver, his pollster accuracy ratings are a broad sample of a very narrow sort of poll:

      These ratings pertain to just one particular type of poll: those for which the field work was conducted within the 21 days preceding a public election, which surveyed people about their voting intention in that election, and which was released into the public domain in advance of the election.

      These ratings reflect polling for President (general and primary elections), U.S. Senate, U.S. House, and gubernatorial races since 1998. More recent cycles are weighted more heavily. This is a truly massive amount of data: roughly 4,700 polls.

      To be sure, a mayoral race isn't a presidential, Senate, House, or gubernatorial race.

      But it's also interesting to note that in the twelve years from 1998 to June 2010, Elway Research had released exactly ten polls in any of those kinds of races in the final three weeks of an election.

      And that's exactly the sort of poll that this one is.

  • (Show?)

    "any pole is a picture of an unfinished horse race"

    -Bill Clinton

  • (Show?)

    pole = poll, sorry Bill! :)

  • (Show?)

    Christian, I tried to use Kentucky Derby metaphors for both KGW and KATU but they thought someone would be offended!

    It is a great example of an early leader tiring out and the late kickers coming on strong. Problem is that in the Derby, the early leader won this year.

    I think I did say "did Jefferson break too late" on KGW but no, I did not compare hm to a horse! Absolutely not!! :-)

    • (Show?)

      Actually, Paul, the analogy does hold! Favorite Bodemeister led this year's Derby almost the entire race, with I'll Have Another hanging back in the pack until very late, coming on to win only in the final furlong.

      Even better, commentators noted that this was the first time in 138 runnings of the Derby that a horse coming out of the post from the 19th position won the race. Watch it here.

connect with blueoregon