Ahn Chulsu pulls out of merger talks

So, there has been an interesting development (article in English) in the exciting run up to the South Korean presidential election, namely that the independent candidate who had agreed to join forces with Moon Jaein (DUP, aka Minju dang) has pulled out.
We don’t know what rattled his feathers exactly, but according to Joongang(in Korean) some in the DUP (not Moon) might have been responsible for dirty tricks of spreading “false rumours” such as that Ahn is about yield.
In an interview with Hankyoreh Ahn also points to the fact that he feels Minju stuck in its old ways, and there is a communication block to Moon on what is going on within the party.

Moon has been trying to woo him back with apologies, but it’s not cutting it for the moment.

I think it’s a strategic move. I have been upset also with the way the polls are all over the show , when, in the end, it is the only tangible thing that Ahn can rely on as he is not of the party. If you try to follow the poll numbers (even with a huge pinch of salt) there are some funny, unexplained fluctuations depending on the case scenario. This seems to be prominent especially in the way Ahn always seems to be ahead of Moon in popularity in a three way between Ahn, Moon and Park, but Moon catches up, or becomes more popular in a case between just Ahn and Moon.
I agree in particular with this interpretation in the way that Moon may gain “false support” from Park supporters who fear Ahn as the rival.
So what do you think? A petulant child or a strategic time-out?
I know at least two votes for the former already.

Update:I found quite a good Joongang opinion piece on the issue

The contenders have to clear the fog they have created as soon as possible. There can’t be many countries in the world in which voters don’t know their presidential candidates 33 days before the election

and here is one of my first ever comments on this issue. Shall I set up with a 돗자리 totjari? (straw mat)? i.e. fortune telling.

  • yuna

    In case DLBarch(who likes Moon Jaein’s looks, amongst other things) drops by, I just remembered who Moon reminds me of and why I don’t like his 인상.

  • karl

    As I work on some of the polling cited in the Chosun story, I’ll add a few comments.

    First, our numbers have consistently shown a tight race between Ahn and Moon in a head-to-head. Almost every day has been within the margin of error. I’ve seen other polls that exclude Park supporters and have Moon leading by 7pp. That’s insane. I’ve seen polls that have Ahn up by 10pp over Park. And this despite the methodology being skewed toward the conservative. If they had taken a more even sample, Ahn would have been leading by 13pp or more. Obviously, that’s not realistic either. Some of the methodologies for these polls is shady, at best.

    Second, the Park effect that the Ahn camp is pushing is overstated and inconsistent. While the Ahn camp is right that more Park supporters (as identified by supporting Park in a 3-way race) break for Moon rather than Ahn, it’s not as big as they imply. Our numbers consistently show that 35% of Park supporters opt for Moon versus 25% for Ahn. But, more importantly, 45% (a plurality) respond as ‘Don’t Know” or the refuse to answer the question. Thus, when Park supporters are excluded, the impact is not as big as is imagined. For example, in the poll on Nov. 15, Moon led Ahn by 6pp with Park supporters included. When Park supporters were excluded, that lead was 5pp. Over the past two weeks, with Park supporters excluded, the race remains a toss-up. For me, that passes the sniff test on the top level. To get any more information requires digging down a few more levels into the data.

    The greater concern for Ahn is this: on November 12 he was the preferred candidate in the head-to-head with Moon by 55% of those in their 20s. That was just before these merger talks began. Since then, he has dropped 5pp every day among this cohort. That makes a 15pp decline in three days. Alarming. He is now at 40% support among this cohort, and virtually tied with Moon. I’m sure the DUP’s alleged tricks were the motivator, but seeing an important part of his base shrivel so rapidly likely made the decision easier to suspend the talks.

    This also provides some insight into how the Park effect works. It only shows up when the base is secure. At other times, when Ahn’s base was secure, he received a 3-5pp bump when Park supporters were excluded. But now, with an important part of his base eroding, the Park effect is neutralized.

    I know there’s significant distrust in the polls around the Marmot’s Hole and I can’t disagree with that. That’s in part because of the polling companies themselves as well as the media. They just don’t do a good job of breaking down their data, and instead peel off the top-level numbers for consumption. Even digging down one more level provides important context that helps to make sense of what is actually happening in public opinion.

  • R. Elgin

    I would have never considered merging anything with “문제인사람”, considering his party – many of which are just as bad as the majority party. I would still vote for Ahn over the other two since it can not be worse, IMHO.

  • yuna

    Thanks much Karl for the insiders insight.

    I’ve also wondered if Obama vs Romney ever went as close as the polls suggested. It feels like poll manipulation (often through the media) is an untrustworthy game in a relatively close race.

  • karl

    Yuna: I agree that it can certainly feel like poll manipulation (and in some cases it is). What is released to the public is actually weighted data. That is, the polling company doesn’t call exactly 1,000 people and then just use those numbers. They call more than 1,000….say 1,200…and then apply a weighting formula to make the final results match the census numbers for age, sex, and population distribution. So, if there’s an 80 year old woman in Jeju, her single response might be “counted” 1.5 times. A 20-something male in Seoul would see his response count for less than 1. Clearly, if the weighting formula is flawed, everything can go to hell.

    I think a good rule of thumb for identifying a quality poll, and a good weight, is its consistency. We do daily polling, and I feel good about our data because it rarely has statistically significant changes over a three day period. Any poll that does is questionable. And if there is a significant change, there better be a damned good reason for it.

  • yuna

    #5 Yes, as I wrote poll manipulation, I hesitated, because that was clearly for want of a better word. It is not so much fiddling with the numbers I meant, but the way it is designed in the first place and/or presented in the media.
    Incidentally, I don’t think there was any agenda/motive for the main US media (though it is accused of being heavily biased for Obama by some) to necessarily present Romney as a real threat, but it was more like, when I watched the results as it came in real time I felt like “It’s over” straight away but felt the media kept trying to make it exciting and very close to keep us “Stay tuned”.

  • Yu Bum Suk

    Thanks so much for your insight, Karl.

  • R. Elgin

    Also, Seoul National University (SNU) yesterday declared it could find no evidence that independent presidential candidate Ahn Cheol-soo plagiarized any of his theses. As Karl observes, there is more than a little fear of ACS and manipulation of anything and everything that can be tricked.

    I hope ACS cuts off 문제인사람 since these people only use him for their gain.

  • yuna

    So my question is, how can they use the ratings (should they decide to) to decide who run between Moon and Ahn?
    As this Chosun article shows, it’s so very close and there exists much room for manipulation by the other Park camp, unless you do a tagged poll, i.e. you tag your individual throughout a period going way back.

    In the scenario of Moon winning between the two but due to *misrepresnted poll*, I wonder, if there are many Ahn supporters who would still vote for Moon.

  • yuna
  • yuna

    See, it’s like I said in #9.

    They are bickering over methodology over the tagged sample of the survey.