Ahn Cheol-soo’s last stand

After three days of stand-off, Ahn Cheol-soo and Moon Jae-in met in person again and re-started the negotiation for unifying the ticket. The meeting was enabled thanks to the decision by the DUP’s leadership to completely resign. Particularly huge was the removal of Lee Hae-chan, a former Prime Minister of Roh Moo-hyun who acts as a Karl Rove-equivalent for the DUP. But the polls say that this round of disruption only helped Moon Jae-in. In the most recent RealMeter poll, Moon comfortably led Ahn as the preferred unified progressive candidate, at 44.6% to 36.1%. Even in the three-way survey, Moon comfortably led Ahn, 28.3% to 21.5%. (Park Geun-hye still very much leads the field that 44.7%.)

Moon Jae-in, in fact, played this negotiation picture-perfect. In the short term, Moon and the DUP essentially gave into virtually all demands from Ahn’s camp, including the resignation of its leadership. Moon also declared that he would defer to Ahn as to the methodology of unifying the tickets. But in the long term, Moon’s yielding on many fronts elevated him to a higher plane. In contrast, by insisting on so many demands and walking out of the negotiation, Ahn is looking more like the “old politics” that he is supposedly campaigning to abolish. It is not simply Ahn’s poll numbers that are sinking. Ahn’s campaign war chest, after the roaring start of collecting KRW 10 billion in the first 32 hours of fund-raising, has sputtered into collecting only KRW 170 million in the next 43 hours during which the negotiation was suspended. (In contrast, Moon Jae-in’s campaign collected KRW 20 billion in the first 56 hours.)

In short, we are potentially looking at the last week in which Mr. Ahn Cheol-soo holds any relevance in Korea’s presidential election.

Ahn Cheol-soo is not the only one who has suffered — the fact that the whole politics coverage is devoted to the drama on the progressive side is irritating Park Geun-hye’s camp. Park has been putting forth significant campaign promises in the last few days, but no one seems to care. Moon and Ahn is scheduled to have a televised debate on Nov. 21; in response, Park’s camp demanded that she be able to hold a televised solo debate on Nov. 23. However, like any intercourse, we all know that having two involved is far superior to having just one involved, both for the participants and for the viewers.

  • songyun

    they should just do 가위 바위 보

  • paulhewson

    a solo debate?

    sounds schizophrenic

  • brier

    Moon didn’t play these events picture perfectly. Only an unabashed partisan would say something like that. Moon had no choice but to accede to Ahn’s prescriptions for reform. If Moon didn’t, he would come across as protecting his own interests and not for the betterment of political culture in the Republic. Ahn wants change, and he looks like he might have to fall on his own sword to get it, though Moon’s camp certainly hasn’t played high minded the way Ahn would like them to play, and Ahn gets no help with the media spin either. Moon didn’t craft the position he is in now, he got lucky. Luck counts in politics sometimes as it does in life sometimes.

  • yuna

    The feeling I have is, when presented with a possibility of change in the Korean politics, and moreover it came from two possible channels it was exciting. However, as time progressed, it became clear that one of them appeared more of a false change.

    So, we have to stay resigned to little change close to no change.

  • karl

    I think the use of the polling here raises several problems.

    First off, I’ll say I don’t disagree with the final analysis that Ahn may be about to exit the national stage. I just think it’s much, much closer than the polls from RealMeter suggest, and that the debate tonight could be crucial.

    The first problem is with reporters handling the polling numbers. The numbers that are cited in the original post here are: Moon, 44.6%; Ahn, 36.1%. (Personally, I think those numbers are to far apart, but I’ll get to that.) It’s the numbers after that that raised my eyebrows. In the original story, it then goes on to some very, very curious math. It reports that excluding Park supporters creates a gain for Moon of 1.8pp, pushing Moon to 48.4%. But 44.6% plus 1.8pp is not 48.4%. It’s 46.4%.

    It then goes on to state that with Park supporters excluded, support for Ahn declines 3.2pp, and falls to 43.1%. That’s also problematic, as his original level of support was cited as 36.1%. 36.1% minus 3.2pp should equal 32.9%.

    The correct numbers should be, excluding Park supporters: Moon, 46.4%; Ahn, 32.9%. The original report said that there was a 5pp gap, but it is actually a 14pp gap. Sorry, but that’s just not believable.

    Here’s what the math tells me. The full sample is 1500. First, I’ll assume that the numbers reported for the three-way race are correct. I think that’s a fairly safe assumption. From that, using percentages from the three-way race, we can calculate how many respondents went for each candidate. There were 670.5 (I know…how do you get half a person?) Park supporters. That number has to the be deducted from the full sample for Moon vs. Ahn Excluding Park Supporters. That leaves a sample of 829.5. If 46.4% and 32.9% are right, then 385 people opted for Moon and 273 for Ahn. Those numbers add up up 658 people, slightly less than the missing Park supporters. But, that missing sample is likely short because of dealing with rounded numbers, and some shifting to the “I Don’t Know” category, which the original story does not report.

    Moreover, and again assuming the numbers for the three-way race are correct, using 48.4% and 43.1% (the incorrect numbers in the original story) would yield the missing sample (Park supporters) as 758. That would mean that in the three-way race Park would be at 50% support, which is too high. So, I am fairly confident that the correct numbers are 46.4% to 32.9%.

    Following through with that math you can also determine how Park supporters broke for Moon and Ahn. I won’t go into the details here, but back of the envelope calculations look like about 42.5% of Park supporters opted for Moon while 40.1% went for Ahn. That’s quite a change from the numbers I see on a daily basis, as those numbers consistently show 35% go for moon versus 25% for Ahn. What is counter-intuitive is that the more Park supporters that support Ahn, the worse it actually is for him. He has been advocating all along that Park supporters are biasing the polls. Now, if there has been a shift in the way that they break (and I don’t believe there has, as I don’t fully trust RealMeter), it actually works against him because he has a smaller sample size to begin with.

    Personally, I’ve never put much stock in the numbers put out by RealMeter. A 14pp gap between Moon and Ahn, with Park supporters excluded, doesn’t past the sniff test. Even a 7pp gap in a three-way is questionable. Moreover, this is not helped by the fact they they omit the mix of landline and mobile phones they use for the sampling. Without that, it’s difficult to run through any scenario to explain what accounts for the differences in breakdowns between their numbers and the numbers I see. In any case, it is certainly making me look at the results I see differently.

  • karl

    One small correction to the previous post: In accounting for the missing sample of Park supporters in the 6th paragraph: adding in the “I Don’t Knows” from the results that excludes Park supporters equals 671. The missing sample for Park supporters was 670.5.

  • R. Elgin

    Karl, I really wonder if statistical modeling could ever be an accurate indicator of how fickle Korean voters will vote. They really strike me as being more unpredictable than any group of voters back in the states or even the EU. Considering the differences between each generation, how could you possibly determine meaningful weighed variables for any population sample you collect?
    Because of this, and the love of inventing facts, I have never taken polls here seriously and I’m not even a mathematician.

  • karl

    R. Elgin,

    First, I have to take back a large part of my previous post. I should have run the numbers three times, instead of two. After talking with a friend, the original numbers in the story may have been correct, but their percentage point changes were wrong. After re-reading the Korean text, they may have also been referring to another poll. So….sucks to be me. The numbers I produced fit, but so would many others. I can discuss why I was wrong in detail in a private exchange if anyone is interested, but that requires a couple pages of by-hand arithmetic. In any case, I’ve learned my lesson: let STATA handle the math. That’s not my strong suit.

    But the larger point of reporters handling polling remians. I still think it was written confusingly. But doesn’t excuse my error.

    Fortunately, I don’t do the weighting myself, but I gather that it is not so difficult to do after establishing a formula. It’s just matching your sample to census data. In the case of Korea, that is gender, age, and region. That’s actually a little simpler than in the U.S because you don’t have to account for race. So in theory, a poll in Korea should be more accurate. Also, Korean has several features which makes polling easier, such as nearly complete saturation of mobile phones. In the U.S.,many elderly don’t have them.

    As for not taking polls seriously, I can understand that. I’ll say, as I said in other spots, that polls are only as good as those interpreting them. The top-level numbers are a start, but its much more important to get into lower levels of the data. From there, identifying patterns and trends is the game. Creating a meta-poll here is much more difficult because there simply aren’t enough polls.

    But I would draw a line between statistical modeling and polling. Statistical modeling, to me…and maybe I am misinterpreting this and would like to hear others thoughts…implies looking into and predicting the future. That would be something like a Likely Voter Model. That is a very popular tool in the U.S.but is just not done here. Matching a sample to census data is something different, and while difficult, is a known quantity.

    To be honest, I’ll have to think more on that and may reply with more later.

  • jkitchstk

    The Ahn vs. Moon debate will be tonight so I tuned in to Arirang TV to see what kind of coverage they’d be providing but was disappointed that they won’t be covering it live and translating the planned event(down to the most minute hollywood action Korean style detail). Arirang’s pathetic!

  • jkitchstk

    So scripted, they got a stack of papers before them on desk and stagehands keep fussing around with them hooking up mics. Moon was amused as they did so to Ahn. Moon was sucking on something big in his big mouth while camera’s were flashing everywhere. Wasn’t this thing suppose to start at 11:00 p.m.? JHC, get on with it already.

  • http://askakorean.blogspot.com thekorean

    It’s on now, and I am watching it via KBS 1TV Internet.

  • Pingback: Talks on the Left of a Unified Candidate | hangukpolitics()