Thursday, October 13, 2011

Global Views Shuts Down Polling! =UPDATED=

Huge news: Global Views has shut down its polling center for the duration of the election, apparently:
The Global Views Survey Research Center’s sudden announcement on Tuesday that it would no longer conduct polls on elections or political issues has triggered intense media coverage and allegations that it caved in to political pressure.

The surveys the center had been conducting on January’s presidential election, as well as its monthly public support trend investigation and political party inclination investigation will all be halted.

The center’s last survey was conducted on Sept. 21 and found that if President Ma Ying-jeou (馬英九) ran just against Democratic Progressive Party (DPP) Chairperson Tsai Ing-wen (蔡英文), support for Ma was 39.2 percent to Tsai’s 38.3. percent.

If People First Party Chairman James Soong (宋楚瑜) joined the race, Soong would garner 10 percent of the votes, and Tsai would lead Ma with 36 percent to his 35.8 percent, the survey found.
If you look at the poll right below this one, you'll Ma has a huge lead. That's in the pro-KMT China Times. Other pro-KMT polls are similar. But GV constantly interrupted this narrative of inevitable DPP loss with the sad news that the race is neck and neck -- even the international media was reporting it as a neck and neck race. It's hard not to read this shutdown of the one semi-reliable poll as an act of political pressure.

Remember too that if the KMT loses they have now prepared the ground for the claiming that the vote was rigged: "Look, the polls had Ma up by 10 or more, but he lost by 2. How is this possible? There must have been cheating!!"

UPDATE: SY in the comments says:
The owners and the head of the GV group are close to KMT, most are members of KMT.

I was told:

The above-mentioned geo-demo-adjusted projections ("analysis") have been showing Tsai leading for the last two months (the Oct data is said to be of the same trend.) The "bad news" has been spreading from ear to ear.

I believe this was the reason why the Ma regime decided to kill it. Personally, I still believe Ma will win the election. Money does talk.
Thanks also to the commenter who pointed to the Lib Times piece: GV currently has Tsai up over Ma 4-6%. The head of GV resigned in the wake of this mess.

UPDATE II: Read Taiwan Echo's post
UPDATE III: Global Views polling center director interview summary in the TT
_______________________
Don't miss the comments below! And check out my blog and its sidebars for events, links to previous posts and picture posts, and scores of links to other Taiwan blogs and forums! Delenda est, baby.

35 comments:

OzSoapbox said...

If this has never happened before then yeah, it's a bit suspicious.

I note that no official explanation was given, what's up with that?

Who puts out a press release announcing abornmal behaviour without an explanation??

Michael Turton said...

The press release is an obvious form of subtle public protest. Nothing stopped them from simply... quietly not polling..... but by making a PR moment out of it, they called attention to their action.

Anonymous said...

Aren't the poll from the Election Study Center of National Chengchi University really more useful? I'd be more worried if these types of polls disappeared.

Michael Turton said...

Chengchih U's polls were spammed by Ma supporters a while back and rendered useless. TT reported on it. I don't know whether the election market is still functioning, and it can't tell you what the poll results are, only who people think will win. Different information altogether.

Michael

Rust said...

Chengchih University uses a model similar to that of the stock market to predict the possibility of the election. Before the spamming, Tsai lead Ma in elected probability (not vote %) by something like 55 v.s. 45. After the spamming is the other way around. They still have the website though.

Anonymous said...

In an interview published today in LibertyTimes.com.tw, the director of Global Views's past polling center, Mr. Dai (戴立安), who resigned after the close of the center, said that currently, Tsai has an advantage of 4-6% advantage over Ma.

http://www.libertytimes.com.tw/2011/new/oct/13/today-p5.htm

SY said...

>The press release is an obvious form of subtle public protest. Nothing stopped them from simply... quietly not polling..... but by making a PR moment out of it, they called attention to their action.<


Not true, Michael. They have subscribing "members" (customers) who have paid for a period of time (e.g. per year) to access their monthly poll raw data, analysis (including geo-, demographically adjusted projections), etc. The public announcement was necessary.

The center is currently busy closing accounts and refunding subscribers ("members".)

The owners and the head of the GV group are close to KMT, most are members of KMT.

I was told:
The above-mentioned geo-demo-adjusted projections ("analysis") have been showing Tsai leading for the last two months (the Oct data is said to be of the same trend.) The "bad news" has been spreading from ear to ear.

I believe this was the reason why the Ma regime decided to kill it. Personally, I still believe Ma will win the election. Money does talk.


NOTE:
The following is taken directly from their publications on how to pay to open a "member" account:

-----
C. How to obtain information about the brief analyses on the surveys, raw data and cross-tabulations?

The GVSRC online database provides brief analyses on surveys for our members with a valid account.

Members with a valid account can log on to 專案資料庫 ("Project/Case Database") to check related information. To download raw data, weighted data, and crosstabs in the GVSRC database, an additional fee will be charged.

For full access to the database, please go to the link: 網路申請 ("Apply On-Line") to register; if you are a registered member with an expired or suspended account, please link to 選購使用期 ("Purchase a Subcription Period") to finish the purchasing process for valid membership. Currently, the database and survey results are in traditional Chinese characters.

-----

Tommy said...

I too find the China Times poll a bit suspicious. Taiwan Thinktank released their own poll earlier this week that showed Tsai down by a little over one point. The question remains over whether green polls (I know that GV was not green, but it did tell a different story than the obviously blue polls) are any more reliable than blue polls.

For a description of what is wrong with a lot of the blue polls, look here: http://blog.udn.com/amlink/5712104. It contains the text of a Wikileaks cable from 2007, written ostensibly by the acting AIT Director. Wikileaks may not really provide any news that we have never heard before, but still, some of the info can be useful.

The gist: Most blue polls contain structural sampling error. Due to the fact that they concentrate on cities, they get a larger than average share of either white collar city residents, who tend to vote KMT. They also tend to rely on telephone surveying in the early evening, at a time when working class folk, who tend to prefer DPP, have not gotten home from work. Since many of the polls do not weight properly, the data that is reported is often very wide of the mark. One of the cable's observations:

"While media polls were relatively accurate in predicting the 2005 county and city elections, more recently, their forecasts for the 2006 Taipei and Kaohsiung mayoral elections were far off the mark. Taipei DPP candidate Frank Hsieh garnered 41 percent of the vote despite polls in the preceding weeks that showed him in just the low 20s, and in Kaohsiung, DPP candidate Chen Chu squeezed out a narrow victory over her KMT rival although she trailed by double-digits in most polls (Ref B)."

My own hunch is that Tsai is still behind, but not by as much as the blue polls say. Of course I could certainly be wrong. Certainly the character assassination of Su over that stupid farmhouse thing isn't helping the matter.

Michael Turton said...

Ah, thanks SY. I forgot the paid subscriptions.

Michael

John S said...

Then there is the built-in, structural bias of both polls AND actual elections due to the fact that voting is unfairly made inconvenient for MANY workers in Taipei from the South of Taiwan.

These people (if they want to vote) have to devote two whole days to exercise their right to vote.

The elections would have a completely different outcome if people were allowed to vote in the area where they actually reside, rather than have to spend a whole day going back to their hometowns in order to stand in line for two hours to place a vote, and then drive another whole day back to their residence in Taipei.

It is obvious which party gains an advantage from this inconvenience. And if it only makes a 2-3% difference in the votes, then that is enough to decide the future of Taiwan.

M said...

Tommy - actually the "pro-blue" polls were fairly accurate in the most recent presidential and municipal elections. They only race where they were way out was in Taiching City. In Taipei and Tainan they actually underestimated blue support last time round.

M said...

John S. - it is not at all obvious who gains.
Do you have any figures on party support for people based in Taipei but with household registration in the south?
The KMT would also point to the many military personnel and China based Taishang who have to travel even further to vote. These groups are generally believed to lean blue.

Anonymous said...

I think how you question and count polls. Whenever I received calls like that, I basically answer with an open attitude that could sway either way. Most people with a higher education probably will also answer in such a flexible manner. I never trust polls that I don't know how they asked question no matter how well known the organization is. Sorry for putting on my auditor's hat here, but the devil is in the detail of these as with anything else.

George

Okami said...

In Changhua County, Tsai Yingwen is everywhere. It's kind of odd. Ma Yingjiu has no pics nor advertising up.

Anonymous said...

The "editorial" by Lincoln Millstein of the Hearst papers and now this. What seems to be shaping up is that the multinational corporations seem to be calling the shots, along with KMT who also have considerable vested interests in China. This isn't politics, its Big Business. Damn, lift up a few stones in this election and you find all sorts of disgusting critters.

Tommy said...

M,

Nobody is saying that blue polls are never accurate. Of course they can be accurate. The point is that it is very difficult to tell whether they are accurate or not until after the election. And it is very difficult to exonerate all of the polls based on some accurate responses. It is possible, for example, that the polls to which you refer were more accurate due to lukewarm voting by Green voters. Take the presidential election. In this case, the structural bias might not matter as much because the people who were not counted would not turn out to vote in high numbers.

If you read the cable in question, the writer says that polls are to be used at your own risk, not that we should never pay attention to polls.

John S said...

No, I don't have numbers to cite, because I don't think any reliable figures have ever been published. Getting relaible numbers would be very difficult.

But I think it is a deliberate obfuscation to try to equate the number of Taisheng with the number of workers in the north, who must vote in their hometowns in the south.

There are many hundreds of thousands of workers residing in the north, but voting (or not voting) in the south. NO WAY there is anywhere near a similar number of Taisheng working in China.

Anyway, the Taisheng get several days off and special flights from China. Normal workers don't get days off or transportation to go vote.

The real reason there has been no system put in place to allow people to vote at any location in Taiwan on election day is because the KMT knows that would give an advantage to the pro-Taiwan parties.

The first reason one often hears for not reforming these outdated voting rules is that changing the rules would result in more voting fraud. But that is not the real reason, it is just a cop out.

M said...

Nobody is saying that blue polls are never accurate. Of course they can be accurate. The point is that it is very difficult to tell whether they are accurate or not until after the election. And it is very difficult to exonerate all of the polls based on some accurate responses.

Isn't that stating the obvious? Of course we can never tell if any poll, blue or green, is accurate until after the election.

And it is very difficult to exonerate all of the polls based on some accurate responses. It is possible, for example, that the polls to which you refer were more accurate due to lukewarm voting by Green voters. Take the presidential election. In this case, the structural bias might not matter as much because the people who were not counted would not turn out to vote in high numbers.

So you are insisting that the polls still have a "structural bias" despite them being largely accurate. Many greens were indeed lukewarm in 2008 - and the polls would have captured that.

M said...

There are many hundreds of thousands of workers residing in the north, but voting (or not voting) in the south. NO WAY there is anywhere near a similar number of Taisheng working in China.

There are at least 750,000 Taiwanese residing in China. Some estimates put the figure even higher.


Anyway, the Taisheng get several days off and special flights from China. Normal workers don't get days off or transportation to go vote.

There might be more flights and charters to cope with the demand, but they have to buy their own ticket! They are not provided with "transportation to go and vote". Even with discounts, it will come to a lot more than a train ticket from anywhere in Taiwan. There is also no rule saying that they get time off for voting.

For Taipei based voters from the south, it should be much easier to get back. With the HSR you can be in KHH in less than 2 hours. Also the vote is on a Saturday, which is also designated as a public holiday. An easy day trip for many I would think.

The real reason there has been no system put in place to allow people to vote at any location in Taiwan on election day is because the KMT knows that would give an advantage to the pro-Taiwan parties.

The first reason one often hears for not reforming these outdated voting rules is that changing the rules would result in more voting fraud. But that is not the real reason, it is just a cop out.


There are definitely a lot of technical issues to sort out, such as organizing registration and preventing double voting. However, the Minister of the Interior has been pushing for this for quite some time. It will happen sooner or later (but only for voters within Taiwan, overseas voting will still not be allowed).

les said...

@Okami: In most areas, Matong is completely absent from the campaigns of local candidates. He's so deeply unpopular in most quarters they consider him a liability, not an asset. Only in the Hengchun area have I seen him on posters, or Wu Dunyi for that matter.

@John S: There are well over 100,000 Taiwanese living and working in Shanghai alone. I think the number for the whole country is close to a million. KMT is going to give them a buy one, get one free deal if they come home and vote blue. Guess who will pay that bill!

Taiwan Echo said...

Hi Michael,

Trying to say something but it gets too long, so I made it a post:

The Shutting Down of a Reputable Poll Center

I provided some background info so we can understand what's going on better.

Anonymous said...

I have a friend that would come back and help KMT polls analysis back in the earlier days. Based on my understanding of her personality, I trust they truly do good work at it. What is finally announced might be different, but I think we really should know what kinds of questions are asked during the polls, and the exact sequence in which they were asked regardless who conducts these. Personally, I think Soong is a good administrator, but not a good leader. Tsai is still in a changing mode, it's hard to tell what she will do. But currently, she does seem not have the vision nor administrative capability. Ma needs a devoted team to carry out his vision; how strong his team can get really is going to be hard to tell, probably only God knows what will happen.

George

John S said...

Yes, every paper and publication throws around guesstimated figures regarding how many Taiwanese reside and work in China or elsewhere, but AFAIK there is no single institution actually counting these people, or even giving a definition for who is and is not "Taisheng".

Are we talking about people who reside there year-round? Or who have to go there 2-3 times a year for a couple of weeks at a time?

Are we also counting the recently-retired Taiwanese military personnel who are now paid consultants for the PLA in Beijing? There are a lot of them, according to Chinese who love to cite them as examples of brotherhood and reconcilliation.

At any rate, re. the topic of "inconvenienced voter groups", you just can't convince me that the number of Taisheng (however defined) is more than 1/10 the number of people (in fact, whole families) who work in companies in Taipei, Hsinchu, Neihu, etc. who must go south or east to vote.

Those people are living and needing services in the north, but have no representation there. If it is difficult for them to travel long distances to vote, then they are, in fact, not being adequately represented in the south or east, either.

Tommy said...

"So you are insisting that the polls still have a "structural bias" despite them being largely accurate. Many greens were indeed lukewarm in 2008 - and the polls would have captured that."

Essentially, yes, that is what I am "implying could be the case" (your choice of the word "insisting" is a bit wide of the mark since I never said that blue polls have not improved in the last few years. We don't know that yet.).

Imagine a situation where a population has people who only like pears and people who only like apples. A polling agency wishes to find out what people like in advance of a fruit election. They only contact people who like pears. They will then conclude that 100 percent of the population likes pears. Then, when election day rolls around, something happens that keeps all of the apple lovers at home. The pear wins the election with 100 percent of the vote, and the polling agency concludes that their work was flawless. Clearly, the coincidental matching of the two numbers says little about the agency's ability to measure public opinion, which is the point of polls after all, right?

Your comment misses the point about sampling. The issue isn't the number of greens who feel this way or that. The issue is the number of greens who are asked to be a part of the sample.

The memo in question draws attention to a key problem in the results of several blue polling agencies -- namely that they chronically contain structural sampling error. This means that they are more likely to be wide of the mark. This does not mean that they will never hit the right numbers.

Based on the simplistic logic in the apple-pear example, we see that even the most biased sample can yield results that are right on the money under the right conditions. Such is the nature of polling. But, if polls are properly done, they should ideally produce, over time, results that vary randomly around a mean that represents the "real" opinion of the population... with a few outliers to boot.

Regardless, my comment about not knowing about blue polls until after the election was not referring to any one poll. Rather, it was referring to the "correctness" of the mean that the polls were varying around before the election.

Incidentally, this is also what the writer of that memo was referring to. Due to the fact that Taiwan polling experts (the one in question was precisely from Global Views) say that blue polls frequently have structural sampling error, we necessarily find it difficult to rely on any of the polls to tell us what will happen in a race, especially when the race can be close.

Anonymous said...

I would be interested in how the polls are revealed during US elections. Certainly there are lots of people that do not reside in the US. Additionally, it is quite common to consult in other countries in this global village. I believe US personnel consult internationally on a much larger scale in various ways. Having had friends and relatives working in the news media,it is a wonder how people view things in ways far from reality. A current Daimond distributor in the Amway business told us how they would dream us stories based on the bits and pieces of information they could put together.

Nobody cares about reality, even iF Design. Recently I attended a meeting explaining the Taiwan Bicycle Design and Innovation Awards procedures organized by iF design. In one of the example they presented was a description of a bicycle that talked about "increased lateral stability", to I to the opportunity to ask how could they compare against other bikes how good the lateral stability is? Based on their response, it was clear that performance issues were not of real concern. The real things are just what you talk about and how you present it. Where what is said a truth or not was really of no concern, but rather how the message was delivered, and whether it hits a sweet spot in the reviewers view or not.

George

M said...

Yes, every paper and publication throws around guesstimated figures regarding how many Taiwanese reside and work in China or elsewhere, but AFAIK there is no single institution actually counting these people, or even giving a definition for who is and is not "Taisheng".

Are we talking about people who reside there year-round? Or who have to go there 2-3 times a year for a couple of weeks at a time?

Are we also counting the recently-retired Taiwanese military personnel who are now paid consultants for the PLA in Beijing? There are a lot of them, according to Chinese who love to cite them as examples of brotherhood and reconcilliation.

At any rate, re. the topic of "inconvenienced voter groups", you just can't convince me that the number of Taisheng (however defined) is more than 1/10 the number of people (in fact, whole families) who work in companies in Taipei, Hsinchu, Neihu, etc. who must go south or east to vote.

Those people are living and needing services in the north, but have no representation there. If it is difficult for them to travel long distances to vote, then they are, in fact, not being adequately represented in the south or east, either.


Yes, but you ignore the fact that people can and do change their household registration easily. For those who choose to not to move their household registrations, all the main population centres in the west of the island are less than 2 hours from Taipei, and the election is held on a public holiday. Furthermore, the turnout figures do not support your claim that the DPP is at a disadvantage due to the fact that you can only vote where your household registration is:
2008:
Little difference in turnout between DPP and KMT areas in the western half of the island, despite high levels of dissatisfaction among many DPP supporters. KMT areas in the east and the outlying islands show the lowest turnout.
http://db.cec.gov.tw/vote212.asp?pass1=A2008A9900000000
2004:
Very high turnout in both north and south. KHH city turnout the highest in the country. KMT dominated east coast and outlying islands again showed lowest turnout.
http://db.cec.gov.tw/vote212.asp?pass1=A2004A9900000000
2000:
As above, but highest turnout in Taichung County this time.
http://db.cec.gov.tw/vote212.asp?pass1=A2000A9900000000

M said...

Tommy- if there are sampling errors, then this will be reflected in inaccurate poll figures. The memo you refer to is from 2007. Since then the polls have improved in reliability.
They were basically accurate in the last 2 elections. This would not have happened had they been consistently under-sampling green voters.

Taiwan Echo said...

M:Since then the polls have improved in reliability. They were basically accurate in the last 2 elections.

Don't forget there was a shooting right before the election last year.

If the shooting changed the result, and the result thus matched what the pre-election polls said, then it means that the polls reflect a post-shooting outcome, thus not accurate in nature.

So to argue that the polls were accurate, a condition has to be true: the shooting didn't change the result.

Can we make that argument ?

Anonymous said...

No matter how accurate polls are, too much exposure to it can dangerously create a phenomena of expectation. This really defeats the purpose of letting people make decisions in a more educated manner. I would vote to have these polls made available after selections instead of before.

George

Tommy said...

M,

You have missed the point again. No matter how many times you tell me that the most recent elections happened to be in line with polling data, you still are no closer to showing that any improvements have been made. Remember the apples and pears? Improvements can only be verified when we have more information to work with. That means another election or two must go by.

Polling agencies only gain their reputations over time precisely because, without the passage of time, their record is unknown. But, by all means, be the parrot and say, "Oh but the most recent municipal and presidential elections once more." News flash: 2007 to 2010 is not really a long time.

M said...

Tommy - 2007 to 2010 is a long time if they have changed their sampling methodology. We will have to see how accurate the polls are this time, but the general view at the moment is that Tsai is behind. DPP people I have spoken to think that "she doesn't stand a chance" and "needs a miracle".
Why is she likely to lose?
Along with the KMT's usual money advantages, there is a powerful incumbency effect in presidential elections (which is why Chen won in 2004), Tsai is an inexperienced campaigner, the attacks on Su Jia-chyuan have worked so far (interestingly the KMT's attempts to attack Tsai herself have failed abysmally), and Su Tseng-chang is waiting for 2016.

M said...

Taiwan Echo - the shooting might have been a factor. In fact the KMT polled 4% higher that the TVBS prediction in Taipei and Tainan. The one big anomaly was Taichung where Su Jia-chuyuan ran Hu much closer than the polls predicted.
Most commentators think this is that many supporters of Liao Liao-yi either didn't vote or defected to the DPP. Although Liao declared his support for Hu, a defeat for Hu would give him a shot in 2014. Therefore, he didn't do anything to mobilize support for Hu beyond what was necessary for appearances sake.

Taiwan Echo said...

@M,

How do we access the accuracy of a poll so far ahead? You have to assume that nothing changes before the election, and no one changes minds during the period of 3 months. I don't think it is a reasonable argument.

DPP people I have spoken to think that "she doesn't stand a chance" and "needs a miracle"

I think that depends on whom you spoke to. There are DPP people who want to see Tsai lose. Tsai's primary opponent, Su Tsen-chang (蘇貞昌), never seems to put himself under Tsai. In many campaign trips where he should have helped local campaigners, he walked into the place (from behind the audience) like he is the candidate. He spent most of his speech bragging about his achievement and left small time frame to talk about whom he is supposed to promote. There's always doubt about whom on earth he is campaigning for. A simple logic deduction to this situation: a winning in 2012 will give Tsai a huge advantage in 2016, which would mean Su's dream of president will be forever sealed.

In fact, before Chou Yi's mud-throwing attack, articles written by Su Su Tsen-chang's followers, aiming at attacking Tsai, were the main fuel the blue media picked up to attack Tsai. As you pointed out, they failed. I believe that's why Chou Yi took back the stage and change the target to Su Chia-Chyuan. But that doesn't mean Su Tseng-chang's followers stop their attacks on Tsai. In fact, by adding fuel into it, Su Tsen-chang's followers play a significant role in making an impression that the green camp also think that Chou's mud attack works.

There are also TI extremists who might not want to see Tsai lose, but are blind to Tsai's messages and to the tide Tsai created. They would tell people that Tsai has no chance at all, too.

the attacks on Su Jia-chyuan have worked

Is that so? If you believe that, you might want to give yourself a better explanation on why the political poll center is Global Views is suddenly shut down in the first place.

M said...

Taiwan Echo - yes of course a lot can change between now and the election. A poll can only give an indication of the position now. Everyone seems to think that Ma is ahead, which suggests that the Global Views poll is an anomaly.
As for Su Tseng-chang's failure to fully support Tsai - this could have an adverse impact. Unlike blue supporters, green voters are very loyal to the party - they vote for the party rather than the candidate. But any divisions in the green camp could hamper Tsai's efforts to win over floating voters.
By attacking Su Jia-chyuan, the KMT is trying to link the Tsai-Su ticket to A-Bian in voters' minds. It remains to be seen how effective this tactic will be.
But I think Ma will win because: (1) the KMT have more money, (2) he is the incumbent

Dave said...

Michael- thanks for the article.

However, I think the discussions here have missed the point so far. It's not only just about the accuracy of the polls; it's about freedom of expression. No matter Global Views is accurate or not, it doesn't deserve to be shut down this way.

From what I understand, it is suspected that senior management of GV and KMT political forces jointedly killed someone (Mr. 戴) who just tried to do his job, professionally, for their political gains. If so, is there any way to ask international press freedom organizations to investigate this?

M- Just want to point out two things: 1) so-called blue polls still badly underestimated DPP's support in 2009's local elections and in 2010, as someone already pointed out, there was the bullet factor 2) the "every poll" here you are talking about are actually just blue polls; in fact, the recent Liberty Times showed a small lead for Tsai, for whatever is worth.