In the past week, both parties have finalized their nominations for the Tainan mayoral race. The DPP has nominated Lai Qingde 賴清德, while the KMT will nominate Guo Tiancai 郭添財. The early polls (and all other indicators) make Lai a prohibitive favorite to win the general election in November.
The DPP released the results of its nomination surveys last Friday. Here are the results:
| Lai Qingde | Xu Tiancai | Su Huanzhi | Li Junyi | Ye Yijin | |
| 賴清德 | 許添財 | 蘇煥智 | 李俊毅 | 葉宜津 | |
| Pollster | |||||
| Shanshui | 46.27 | 29.84 | 19.60 | 2.62 | 1.66 |
| Guanchajia | 43.52 | 33.17 | 18.63 | 2.78 | 1.90 |
| Jingzhan | 43.05 | 34.80 | 18.70 | 1.65 | 1.80 |
| Average | 44.28 | 32.60 | 18.98 | 2.35 | 1.79 |
As you can see, Lai won by a comfortable 12% margin, well above the margin of error for a single poll. Moreover, the three polls yielded nearly the same results. While the Shanshui poll was a bit more favorable to Lai, the difference was only about 3%, not an unreasonable result for a random sample. Like the Kaohsiung surveys, the DPP got a clear and clean result here: Lai was clearly the winner.
It’s probably a good thing that Lai did win by so much, because comments from Xu’s camp lead me to believe that he was looking for an excuse not to accept the results. On the one hand, he doubted the fairness of the polls because he didn’t think he should have lost by that much. (Had he lost by less, you can bet that he would have thought he should have won; politicians are like that.) On the other hand, he doubted the fairness of the polls because he wondered if there was some sort of sampling problem. His argument was roughly as follows. He was particularly strong in two townships in Tainan County, Dongshan 東山鄉 and Madou 麻豆鎮, and his support organizations (後援會) there reported that none of them had received a telephone call. He concluded that there must have been something wrong with the sampling.
Ok, lets do a little math to see how likely that is. In the 2008 presidential election, there were 1432399 eligible voters in Tainan County and City. Of those, 55353 lived in Dongshan and Madou. How many of those could be in Xu’s support organization? Let’s assume (optimistically) that 80% of the eligible voters are politically active enough to potentially participate. Then assume (again, optimistically) that 65% of those lean green, and 50% of the green leaning voters supported Xu. Finally (and most unreasonably), let’s assume every single one of those people is in Xu’s support organization. That’s 14392 people, or roughly 1% of the total electorate. (In fact, I’d be shocked if his support organization in those two towns were even one tenth that size.) Assuming the sample sizes were 1000 for each of the surveys, or 3000 total, about 30 of those people should have been sampled. If (and here we assume perfect reporting from every member of the organization back to the top) none of them were in fact sampled, we might be a bit suspicious. On the other hand, if it turns out that Xu’s support organization is only one twentieth that size, roughly 700 people, then only 1.5 of them should have been sampled. There are lots of sets of random samples that would not include any of these people. In short, Xu’s complaint is only reasonable under the most heroic of assumptions. Frankly, if Xu Tiancai, a supposed expert in financial matters, doesn’t understand basic notions of probability, (I probably shouldn’t finish this sentence).
In the day or two after the results were announce, Xu succumbed to better judgment and indicated he would not split the party by running as an independent.
Why did Xu and Su lose? I have heard two good explanations. The first is that both Xu and Su were two-term incumbents, and voters wanted a change. Lai certainly hammered this point throughout his campaign. If he didn’t think it was working, he probably would have stopped saying it. However, I’m not completely convinced that this was the real key. Voters in Taichung City seem to have few qualms about giving Jason Hu (another two-term incumbent) four more years. One might argue that Hu is a KMT member, and DPP voters take notions of rotation of power more seriously. Ok, but Taipei County voters didn’t seem too worried about sending Su Chenchang back into power which would have made it three of the past four years for him in that office. I think this probably helped Lai, but I doubt it was decisive.
The other explanation is that this primary (as well as the Kaohsiung primary) was a rejection of Chen Shuibian. Of all the candidates in the Tainan race, only Lai did not personally go and visit Chen in prison. In fact, Chen publicly complained that Lai had not visited him. Chen’s favored candidate was Xu, who was a member of Chen’s Justice Faction going back to the 1992 election. In fact, Chen’s support was a central theme of Xu’s campaign (though perhaps not as important as Xu’s record as mayor). It may be true that Lai’s victory really represents a desire to move away from the Chen era (and Lai is much closer to party chair Cai Yingwen). However, remember that this was a multi-candidate election. If we are going to interpret Lai’s support as a rejection of Chen, we must also remember that 55% of the Tainan electorate supported candidates who actively curried favor with Chen. So let’s not get carried away speculating on the national implications of this primary.
I don’t know much about Lai Qingde, and voters usually know more about their executives than their legislators. I have a feeling this was really more of a rejection of Xu and Su than a victory for Lai. My guess is that both of the above factors mattered, but that dissatisfaction with the performance of the two incumbents was probably the decisive factor. Of course, I have no evidence for this line of speculation.
Today’s newspaper brings reports of the KMT side of the race. After weeks of trying desperately to find someone else, the KMT finally admitted that it was stuck with the declared candidates (just as in Kaohsiung). Four waves of surveys all showed that Guo Tiancai 郭添財was slightly more popular than Li Quanjiao 李全教, by margins of between 1% and 3%. Now, 1-3% is not significantly different in a single poll, but if you find those differences again and again over several polls, eventually you can determine that one is, in fact, higher.
Li has accepted the decision and expressed support for Guo, though let’s not imagine he was swayed by my argument about repeated polls. He would probably be fighting a bit harder if this were a winnable race. However, the “prize” is to waste a lot of money and energy in what is almost certain to be a humiliating defeat.
The UDN has an interesting article on the decision between Guo and Li. (Well, interesting to me: it has the smallest headline and the least space on the entire page.) Guo, it seems, was the safe choice. Li had much more grassroots support and much better organization. However, Li also has a court case pending and is considered much more controversial and attackable. Guo, in contrast, is a former education bureaucrat who is now serving as vice-president of a technical college. He hasn’t cozied up to grassroots power brokers; in fact, they complain that he is a cold fish. However, this race is really about setting the table for 2012, and it wouldn’t do Ma Yingjiu any good to win a few more votes this year while sullying the party image.
So, how is the horse race looking? A UDN poll from last Friday says Lai 57, Guo 15, undecided 27. (The same poll had Lai 59, Li 15, undecided 26.) I don’t think Lai will win by a 4-1 margin, but a 2-1 victory is not out of the question.
Frozen Garlic 