Public Opinion Polls and Election Outcomes in Taiwan
Taipei mayoral election (December 9, 2006)
Public opinion polls:
TVBS (12/4-5): Hau Lung-hai (52%), Frank Hsieh (21%), James Soong (7%), Undecided (14%)
China Times (12/5): Hau Lung-hai (44%), Frank Hsieh (17%), James Soong (5%), Undecided (22%)
United Daily News (12/2): Hau Lung-hai (48%), Frank Hsieh (17%), James Soong (8%), Undecided (26%)
ERA TV (11/27): Hau Lung-hai (41%), Frank Hsieh (20%), James Soong (8%), Undecided (26%)These poll results contain "undecideds" which we will eliminate by re-percentaging the 'decided' results to add up to 100%. This assumes that the undecided will distribute in the same proportions as the decided. This assumption may be wrong, but it is needed for the comparisons in the absence of further information. Here are the percentages.
TVBS: Hau Lung-hai (60%), Frank Hsieh (24%), James Soong (8%)
China Times: Hau Lung-hai (56%), Frank Hsieh (28%), James Soong (6%)
United Daily Times: Hau Lung-hai (65%), Frank Hsieh (23%), James Soong (11%)
ERA TV: Hau Lung-hai (64%), Frank Hsieh (27%), James Soong (11%)And here are the actual election results:
Hau Lung-hai (692,085; 53.8%), Frank Hsieh (525,869; 40.9%), James Soong (53,281, 4.1%)
Meanwhile, the gambling syndicates had Hau Lung-hai favored by between 150,000 to 180,000 votes. The actual outcome was that Hau Lung-hai won by 166,216.
Kaohsiung mayoral election (December 9, 2006)
Public opinion polls:
Liberty Times (12/5): Huang Jun-yin (34.38%), Chen Chu (34.09%), Undecided (24.72%)
China Times (12/4): Huang Jun-yin (43%), Chen Chu (29%), Undecided (27%)
TVBS (12/3-12/4): Huang Jun-yin (51%), Chen Chu (31%), Undecided (11%)
United Daily News (12/2): Huang Jun-yin (39%), Chen Chu (27%), Undecided (32%)
Era TV (11/26-11/27): Huang Jun-yin (41%), Chen Chu (26%), Undecided (30%)These poll results contain "undecideds" which we will eliminate by re-percentaging the 'decided' results to add up to 100%. This assumes that the undecided will distribute in the same proportions as the decided. This assumption may be wrong, but it is needed for the comparisons in the absence of further information. Here are the percentages.
Liberty Times: Huang Jun-yin (45.66%), Chen Chu (45.38%)
China Times: Huang Jun-yin (59%), Chen Chu (40%)
TVBS: Huang Jun-yin (57%), Chen Chu (35%)
United Daily News: Huang Jun-yin (57%), Chen Chu (40%)
Era TV: Huang Jun-yin (59%), Chen Chu (37%)And here are the actual election results:
Chen Chu (379,417, 49.41%), Huang Jun-yin (378,303, 49.27%).
Meanwhile, the gambling syndicates had Huang Jun-yin favored by between 10,000 to 20,000 votes until Friday afternoon, when they suddenly switched to even odds. The actual outcome was that Huang Jun-yin lost by 1,114. (TVBS) However most of the NT$5 billion bets (for both Taipei and Kaohsiung) had already been placed at the 15,000-20,000 handicap in Kaohsiung. So what happens is that a lot of the gambling syndicate heads have "left town" because they could not cover the bets.
Final scorecard: Hail to Liberty Times and the gambing syndicates.
(Apple Daily) Public Opinion Polls, Gambling Odds and Democracy Gone Astray. December 9, 2006.
[in translation]
Public opinion polls are manipulated in Taiwan elections. Public opinion polls were supposed to have objectivity as the highest goal. Instead, they have become the tools for political parties and biased media. This has destroyed the scientific nature of public opinion polls and make them lies and jokes in Taiwan.
Apart from buying votes, the gambling syndicates also contaminate elections. When gambling brings in the monetary consideration, the original intent of democracy is subverted. When a voter casts a ballot, he/she ought to be considering his/her true interests and not gambling interests. Candidate A is the best fit in terms of interests and ideas but the odds are that he will lose. Thus, the gambler may place the bet on Candidate B instead, even though Candidate B is not his favorite. Not only that, the gambler will maximize the chances of winning by pressuring his friends and relatives to vote for Candidate B. As a result, the election results does not match the logic of democracy. In the long run, this will corrode the quality of democracy and disrupt political stability.
As for public opinion polls, they have been manipulated since the first day of the democratization of Taiwan. This is the reason why in each election, the ultimate outcome deviates largely from pre-election public opinion polls. If public opinion polls were accurate in Taiwan, the president today would be Lien Chan and not Chen Shui-bian as early as year 2000.
Using this election as an example, the biased media cooperated with the KMT to protect Hau Lung-hai and forsake James Soong. The published public opinion polls deliberately indicate that Hau Lung-hai was far ahead (45% or above) and push down James Soong (only 5% to 7%). This creates the impression that there is no need to waste the vote on Soong. But furthermore, they cannot afford to "make" Frank Hsieh's support too low because their voters may get complacent and don't bother to show up. Therefore, Hsieh's support level is always around 18%. But the election outcomes today may be very different from the aforementioned numbers.
Taiwan voters are also "public opinion poll" savvy. When they receive a telephone call from a pollster, they will either cooperate by lying to the interviewer according to their own needs or else they will just hang up. This is another reason why public opinion polls are inaccurate. Compared to the accurate coorespondence between polls and election results in western countries, Taiwan public opinion polls are so peculiar. Western scholars who tried to study Taiwan elections based upon public opinion polls are confounded.
It is bad enough for political parties to manipulate public opinion polls. The media ought to be above the fray and stay neutral, but they are in the middle of making or altering public opinion polls in order to mislead the audience and influence the election. This is even more shameless. Making up fake public opinion polls or doing unserious work has serious social consequences -- the candidates, the political parties and the voters might believe that victory was certain, but the outcome was the opposite; they find it unacceptable; they suspect fraud; disturbances and riots occur; social stability is disrupted.
When the lies deceive yourself and society has to pay the price, that is despicable. Therefore, the voters should calmly and rationally cast the vote that represents their own dignity and personal decisions, and not be influeced by public opinion polls and gambling odds.
(United Daily News) Inaccurate Public Opinion Polls? There are more green recessive voters. December 10, 2006.
[in translation]
In the Kaohsiung mayoral election, the front-leading Huang Jun-yin lost by over 1,000 voters. Does the difference between the pre-election public opinion polls and the election outcome mean that the public opinion polls are inaccurate?
In the public opinion polls, the voters can be classified as "dominant" and "recessive" (note: terms from the theory of genes). "Dominant" voters will publicly show that they support specific candidates, whereas "recessive" voters will not. The key issue about the accuracy of public opinion polls is the assessment of the "recessive" voters.
Ordinarily speaking, the public opinion polling organizations will publish the supports of the survey respondents. For example, in our newspaper's polls on the Kaohsiung election, Huang Jin-yin led comfortably Chen Chu comfortably by 4% to 12%. While Huang Jin-yin appeared to be leading, the people who could really affect the election outcome is the 30% or so who refused to state their preferences in the public opinon poll. If these recessive voters are light-green supporters, then the apparent lead of Huang Jun-yin is misleading.
So who can we determine the number of blues and greens among the recessive voters in the public opinion poll. The historical data show that the Kaohsiung recessive voters are more green than blue but the ratio will shift with the political climate.
In 2002, Frank Hsieh and Huang Jun-yin were vying for the position of Kaohsiung mayor. In the final public opinion poll conducted by United Daily News, Frank Hsieh had 34% support and Huang Jun-yin had 35% support. The two were within the margin of error of each other. The election outcome was that Hsieh received 50% of the votes and Huang 47%. From this, it can be inferred backwards that the people who did not indicate their preference in the poll voted 57% for Hsieh.
After the 'horse fart culture' and the Kaohsiung BOT scandals broke open last September, there were more pan-green supporters turning into recessives. In the public opinion polls, the number of people willing to admit that their political position leaned towards the Democratic Progressive Party and their candidates slid down. Based upon our newspaper's pre-election polls, about 75% of the recessive voters in Kaohsiung are pan-green supporters.
Since the recessive voters in Kaohsiung are mostly green supporters, the pre-election poll may show the Huang was leading Chen but after considering the structure of the recessive voters, both the KMT and DPP internal assessments were that this was an even race. This is a case in which the insiders had completely different readings from the outsiders.
When considering the public opinon poll results, it is not enough to look at the "support levels." It is also necessary to consider the structure and leanings of those voters who have not expressed their preferences, and then you can come up with a more accurate assessment.
Related Links: