the state of (out of date) public opinion

I’d like to take a look at the general state of public opinion in Taiwan these days. I’m not really interested in exactly what it is right this moment. After all, the next election is still more than a year away, and by then no one will care whether Tsai Ing-wen had a 27% or 29% approval rating 14 months ago. I’m more interested in taking stock of the general trends over Tsai’s first year and a half in office. Over the past few months, we have seen headlines screaming that Tsai has a lower approval rating than Donald Trump or that her administration is sinking fast as it loses popularity. On the opposite end of the spectrum, there are people who assume that the KMT is doomed and that continued DPP rule is inevitable, unless the New Power Party replaces it.

Public opinion is always a bit murky, but right now it is murkier than usual. To put it bluntly, reasonable people can see whatever they want to see in the data. None of the trends are sharp or clear enough that you can’t easily explain them away with other readily available numbers. It’s like staring at clouds: maybe it’s a duck, and maybe it’s a train. Not only that, but the person who thinks it is a duck can see the duck clearly, while the moron who thinks it is a train can’t see how anyone could look at that train and see a duck. I think that one interpretation of public opinion is more correct, and I’ll argue for that one. However, keep in mind that the other interpretations aren’t necessarily stupid or misguided.

I’m relying on data from the Taiwan Election and Democratization Studies (TEDS). TEDS is the main academic project for political surveys. The surveys are done by academic institutions (mostly the Election Study Center, NCCU) for scholarly purposes. They are not contracted out to private survey companies, nor are they designed to make a splash in the public discourse. Questions are designed by committees of scholars with all political stripes, so a leading question designed to make one side look good or the other side look bad will simply not make it onto the final questionnaire. We all have lots of questions that we want to ask, and space is severely limited. Only items that can be rigorously defended from an academic perspective make it through. You might wonder why you have never seen a media report trumpeting the latest finding from a TEDS survey. The reason is simple. TEDS doesn’t hold press conferences to announce its newest results. In fact, it doesn’t release results immediately at all. Data usually are only released several months after the interviews are completed. For example, I am using quarterly surveys in this post. These surveys are only released when interviewing for the next survey begins. For example, TEDS just released the data for the June 2017 survey, but we currently have a small army of students doing calls for the September 2017 survey. Who cares about the June 2017 results? We want up-to-the minute information! When the June survey was done, pension reform hadn’t passed, the brouhahas over the infrastructure plan were still in the future, Mayor Ko was still planning for the university games, and Lin Chuan still had a couple more months to go as premier. Everything is different now! Well, it is precisely because the media doesn’t pay attention to TEDS results that they are so trustworthy and valuable. Unlike all the other data you see, you can be confident that these results weren’t produced with the goal of manipulating your opinions. And if you want to know what the state of public opinion is today, I guess I’ll be able to tell you that in three months, even if by then you will no longer care what people thought before Trump’s nuclear attack on North Korea, the Bangladeshi refugee crisis following the massive cyclone, or the upheavals in China following Xi Jinping’s aborted attempt to name his housekeeper’s mentally deranged son as Crown Prince of the CCP.

 

Let’s start with the headline number. Everyone has been talking about President Tsai’s low approval rating. How far did she sink over her first year in office?

請問您對她擔任總統以來的整體表現滿不滿意?

“How satisfied are you with Tsai Ing-wen’s overall performance as president?”

(Chinese question wordings are from the TEDS website: teds.nccu.edu.tw. Some of the English translations are from that website, and some are my own.)

Tsai approval

As you can see from the chart, Tsai’s satisfaction rating dropped quite a bit from June 2016 to June 2017. In the first survey, she was over 50%, the next three surveys were in the mid-30s, and the latest survey was in the upper 20s. That looks pretty terrible. If you want to see the electorate as unhappy with her performance and primed for a change, you certainly can.

It gets worse. TEDS asks about Tsai’s performance in four specific policy areas, cross-straits relations, foreign affairs, the economy, and national defense. These all mirror her overall satisfaction rating fairly closely, except that the trend lines for economy and cross-straits relations are 5-10% lower than the overall trend line. That is, Tsai’s general satisfaction is the optimistic number. As you drill down into specifics, people are even less satisfied.

那您對蔡英文在處理兩岸關係的表現滿不滿意?

那您對她在外交方面的表現滿不滿意?

那您對她在國防方面的表現滿不滿意?

那您對她在促進經濟發展的表現滿不滿意?

Tsai approval areas

We can go further. People are even less satisfied with the cabinet’s performance than with President Tsai’s. Tsai just changed premiers, but most of the unpopular cabinet is still in office. (Premier Lin did slightly better than the cabinet but significantly worse than Tsai.) Ick.

cabinet approval

On the surface, it doesn’t look good. However, I think Tsai’s approval ratings probably mean a lot less than the international media thinks. Comparison with American presidents is especially misleading. Taiwanese are simply more skeptical of their presidents. Unlike American voters, Taiwanese voters historically do not connect expressing satisfaction and intention to vote for a politician. To give a famous example, Mayor Chen Shui-bian had an approval rating in the 70s during his 1998 re-election bid but only got 46% and lost. Looking at presidents, perhaps Ma Ying-jeou’s experience is instructive. Ma had fairly pedestrian approval ratings during his first term, and yet he was comfortably re-elected. In fact, President Ma’s first term approval ratings look shockingly similar to Tsai’s thus far. At roughly this point (August 2009 – February 2010), Ma hit his nadir and then slowly recovered as the 2012 election approached and partisan loyalties reasserted themselves. It probably didn’t hurt that Ma appointed a professional politician who could communicate effectively with the public as premier at about this time. You will also notice that Ma started with a sky-high approval rating. Both he and Tsai had honeymoon periods that quickly evaporated. If you ignore those first data points, the declines for both don’t look quite so dramatic. Somewhere between 25% and 40% approval seems to be normal for Taiwanese presidents. (In contrast, anything below 40% is a disaster for an American president.)

Anyway, I increasingly don’t believe in the predictive value of satisfaction. Lots of the people who are dissatisfied are unhappy that Tsai Ing-wen has been too timid. These people wanted more of her program, not less. Someone who is angry that transitional justice has been too slow or that pension reform wasn’t drastic enough is not itching to vote for the KMT in future elections.

 

So let’s turn to party ID, a much more important and accurate indicator of which way the winds are blowing. Here are the five quarterly surveys from Tsai’s first term:

在國內的政黨之中,請問您認為您比較支持哪一個政黨? (if none, follow-up) 那請問您有沒有比較偏向哪一個政黨?

“Among the political parties in our country, which do you support most?” (if none, follow-up) “Which party do you lean toward?”

party id recent.png

The initial impression is that the DPP is bleeding support and the KMT is gaining. The gap between the two parties has shrunk from about 20% in June 2016 to about 3% a year later. Wow!

Again, some caution and some longer-term perspective is useful. That first survey showing the DPP with 39% party ID is wildly out of line with historical patterns. The DPP has never been near that high in any other survey. Let’s chalk that up as a fleeting honeymoon effect and discard it. Still, the last few surveys show the DPP falling from about 30% to about 25% and the gap between the two main parties narrowing. The KMT is making a comeback!

About that, maybe we should look at a longer time period. The TEDS quarterly surveys started in September 2012, so we don’t have the same sort of regular data before that. Nevertheless, party ID is always asked on every ESC and TEDS survey, so we have fairly reliable numbers going back to the mid-1990s. I’ve started this chart at the end of the Chen era to compare current party ID trends to the relatively stable period during Ma’s first term. As you can see, during Ma’s first term the KMT generally had a 10-15% advantage over the DPP in party ID. During his second term, the KMT hemorrhaged support, going from the mid-30s to the low 20s while the DPP gained slightly, going from about 25% to 30%. Put into that context, party ID in June 2017 doesn’t look anything like party ID in 2009. We are still in the post-Sunflower world in which the DPP is the more popular party and the KMT is in second place. In the chart with only five surveys, it looks like the KMT is making a comeback. In the longer perspective, it doesn’t look like that so much. The KMT is still mired in the low 20s, where it has been since late 2014.

party id since 2008.png

One interesting thing about this chart concerns the DPP’s honeymoon peak. The DPP’s peak comes in June 2016, after Tsai’s inauguration. The KMT also has a massive peak, when it hit 43% at the end of 2011. The difference between these two peaks is that one occurred right around election day. The KMT’s 2012 campaign apparently peaked perfectly, with a spike in support for the party just when the voters were going to the polls. This spike (and the 15% advantage in party ID) produced 51% of the votes in the presidential election. In 2016, the DPP’s spike was well after the election, so it did not translate into more votes. At the election, the DPP had an advantage of just a bit more than 10% in party ID, not the 20% it would have in June. The DPP’s party ID in the low-30s around election day produced 56% of the vote for Tsai.

Historically, the DPP has added more voters to its base of identifiers than the KMT has. If the DPP still has a narrow lead in party identifiers, it probably has a somewhat larger lead in actual votes.

The past two data points are not good for the DPP. However, I’m not ready to proclaim them as a tipping point or the start of a new trend. On party ID, it looks to me like we are still in the same post-Sunflower world. Nonetheless, it is something I’ll be keeping my eye on.

 

If I want to know what will happen in the near term, I look at party ID. If I want to know about the longer and deeper trends, I look at national identity. This is measured using a very simple yet telling question:

我們社會上,有人說自己是「臺灣人」,也有人說自己是「中國人」,也有人說都是。請問您認為自己是「臺灣人」、「中國人」,或者都是?

“In Taiwan, some people think they are Taiwanese. There are also some people who think that they are Chinese. Do you consider yourself as Taiwanese, Chinese or both?”

national id recent.png

Looking at the recent data, it doesn’t appear that there is any honeymoon peak. Rather, there is simply a decrease in exclusively Taiwanese identity and a commensurate increase in Chinese and dual identity. This is wrong. Actually, the June 2016 data point is higher than either 2015 or the rest of 2016.

TaiwanChinese.jpg

A look at the longer trend shows that the decline in Taiwanese identity is not merely something that has happened over the past year. The gap between the lines peaked in 2014, when exclusive Taiwan identifiers outnumbered people with some Chinese identity by about 60 to 35%. Over the past three years, that gap has slowly narrowed to roughly 56 to 40%. This is not merely a statistical blip. These data points in the ESC chart combine data from numerous surveys; each data point represents over 10,000 respondents. There is a real decline in Taiwanese identity over the past three years; the only question is how we should think about it.

Many smart people think that 2014 was a real inflection point, and the historical trend toward more and more Taiwanese identifiers has now reversed. They expect that Chinese identity will continue to increase over future surveys. I have not yet heard a convincing explanation for why this might happen, but then I don’t have a airtight explanation for the last three years either.

For now, my working hypothesis is that long-term drivers of Taiwanese identity are still in place. Younger people identify more strongly as Taiwanese than older people, and this is driven by education and real-world experiences in which China clumsily continually reminds Taiwanese that they are a different group of people. If the fast-growing China of a decade ago couldn’t attract Taiwanese youth, I don’t see how the slower-growing and more oppressive version of today will win over many hearts and minds.

For the time being, I am considering the peak in 2014 to be the outlier. My guess is that after the dramatic upheavals of the previous few years, many respondents who would have normally been on the fence were inspired to describe themselves as exclusively Taiwanese. As things calmed down, those people may have drifted back to their more “normal” dual identities. There is a rapid growth in Taiwanese identity from 2011 to 2014, and I suspect at least some of the people who changed their minds then have changed them back again. If you look in longer terms, the basic trend line over the past two decades still looks like it fits the current data.

It is also important to note that, even with the changes over the past three years, we are still not back in the world of 2008, when Taiwanese and Chinese identities were roughly equal. Exclusive Taiwanese identifiers still significantly outnumber people with Chinese identity, and the current trends will require several more years to close that gap. However, if my interpretation is correct, there may not be many more wafflers to convert back to Chinese identity. Closing the gap much further will require some fundamental change in the relationship between China and Taiwan to make Chineseness more appealing to Taiwanese youth. Perhaps that has already happened, and I am simply oblivious to it.

Regardless, this is an important indicator to keep an eye on. It is not at all an overstatement to say that Taiwan’s political future depends on the distribution of opinions about national identity.

 

The TEDS quarterly surveys mostly ask the same questions each time, but they also stick in one or two questions on topical issues each time. Some of these are illuminating.

In September 2016, respondents were asked a question on the recent Illicit Party Assets bill:

立法院在今年7月通過「不當黨產處理條例」(全名: 政黨及其附隨組織不當取得財產處理條例),請問您對政府在處理不當黨產的表現滿不滿意?

“In July the legislature passed the Illicit Party Assets Act. Are you satisfied with the government’s performance in handling illicit party assets?”

35.0% of respondents said they were either somewhat satisfied or very satisfied, while 42.0% said they were either somewhat dissatisfied or very dissatisfied with the government’s handling of this issue. The easy interpretation might be that the public sides with the KMT’s insistence that the DPP is conducting a vengeful, unjustified, and undemocratic witch hunt against it. However, digging a little deeper casts doubt on that interpretation. 26.0% of people who self-identified as DPP or NPP supporters also expressed dissatisfaction with the government’s handling of illicit KMT assets. Of course, we can’t know exactly what each individual was thinking and every individual thinks something slightly different, but it isn’t too difficult to imagine that the overwhelming majority of these people were dissatisfied because they thought the efforts to recover illicit KMT assets were not aggressive enough. If you want to know how much support the KMT has for its position, you probably need to subtract the vast majority of these people – 9.9% of the full sample – from the 42.0% who were dissatisfied. Similarly, a considerable chunk of non-identifiers were probably also dissatisfied for the same reason. What starts out looking like a good result for the KMT is probably actually nothing of the sort. People might be dissatisfied with the Tsai government, but this does not necessarily mean they are jumping over to the KMT.

 

Both the March and June 2017 surveys had a question on pension reform. Note that the June survey was conducted about two weeks before the legislature passed the pension reform bill.

In March, the survey asked about the preferential savings rate:

有人說: 「公教優惠存款(十八趴)的廢除對退休公教人員不公平」,請問您同不同意這種說法?

“Some people say, “Abolishing the preferential savings policy (18%) is unfair to retired public employees.” Do you agree or disagree?”

Agree Disagree
All 30.2 58.7
KMT identifiers 45.3 44.2
DPP + NPP identifiers 21.8 72.7
Public employees 42.7 47.6

This question wording is a particularly strong one for proponents of pension reform. This focuses attention on the most easily understood aspects of a very complex topic. The preferential savings rate has been the horse that advocates have loved to beat for years, as a guaranteed 18% interest rate on savings deposits is far out of line with anything a normal person could hope to obtain. In fact, nearly twice as many people disagree with the statement as agree with it. Even among KMT identifiers and public employees, the two groups most hostile to pension reforms, nearly as many people disagreed as agreed with this statement.

In June, TEDS asked a very different question:

整體而言,請問您對政府處理公教人員年金改革的表現滿不滿意?

“Overall, are you satisfied with the government’s performance in handling public employees’ pension reform?”

satisfied Dissatisfied
All 31.7 56.4
KMT identifiers 17.6 77.5
DPP + NPP identifiers 49.9 43.0
Public employees 20.1 75.4

(Remember, this is before the legislature passed the bill.)

By now, you should know how I feel about these satisfied/dissatisfied questions. Taiwan’s population is highly critical. If they don’t get their ideal outcome, they do not hesitate to express dissatisfaction. In fact, the Tsai government moved deliberately and cautiously on pension reform, angering a lot of green supporters who wanted a more radical approach. In this survey, not even half of DPP and NPP supporters were satisfied.

On this topic, I have a little private data. I’m doing a project on fighting in the legislature, and I did an internet survey before and after the pension reform was passed. Now, internet surveys have to be interpreted with extreme caution since they are not representative samples. Our goal was to study how attitudes changed after a brawl, not to make statements describing the Taiwanese population. However, that is what Nathan Batto, rigorous scholar, does with the data. Frozen Garlic, the irresponsible blogger, is going to throw caution to the wind and give you some results that you shouldn’t take at face value.

First, let me tell you that my data are biased. My respondents are extremely highly educated, have too many middle aged people, too many northerners, not enough farmers or homemakers, and too many public employees. KMT and NPP identifiers are overrepresented, while DPP identifiers are underrepresented. In general, all the results in my survey skew much bluer than those of a representative telephone sample.

Unlike the TEDS questions, we framed our question in terms of partisan positions:

在年金改革的議題上,您比較支持國民黨的立場還是民進黨的立場?

“On pension reform, do you support the KMT position or the DPP position more?”

Wave 1 Wave 2
KMT position 20.8 21.1
DPP position 39.9 42.1
Neither/no opinion 39.3 36.8

In this sample, roughly twice as many people preferred the DPP position as the KMT position. Remember, this sample is too blue, so in the actual population it was probably more than a two to one ratio.

Think about this. The general population was generally dissatisfied with the government’s handling of the pension reform, even though it preferred the DPP’s position over the KMT’s position by a large margin. Taiwanese people are not easily satisfied! Nevertheless, even if the general public isn’t happy with the DPP’s performance, the pension reform issue is definitely not a winner for the KMT.

 

How about marriage equality? TEDS has questions from March and June 2017:

有人說: 「應該修改民法讓同性可以結婚組成家庭」,請問您同不同意這種說法?

“Some people say, “The Civil Code should be amended to allow gays to marry and form a family.” Do you agree?”

 

This question wording presents a stricter test for support of marriage equality than a less specific question, such as “Do you agree that gay should be allowed to marry?” Amending the Civil Code is the strongest version of marriage equality. There are people who support a weaker version of marriage equality, such as enacting a special law but not amending the Civil Code. Nonetheless, the degree of opposition to amending the Civil Code is striking.

agree Disagree
March 2017 39.1 52.1
June 2017 33.7 57.0
Breakdowns of June sample
KMT identifiers 21.9 69.4
DPP identifiers 37.5 54.3
NPP identifiers 64.6 32.3
Taipei City 40.3 44.8
Age 20-29 71.1 21.1
Age 30-39 50.0 40.5
Age 40 and up 18.6 72.1

KMT identifiers are overwhelmingly against amending the Civil Code, while DPP identifiers also have a clear majority against it. NPP supporters are clearly the outliers. Geographically, Taipei City sticks out. While Taipei residents are split evenly, every other place has 60-70% against amending the Civil Code. When Taiwanese sarcastically talk about Taipei residents living in a bubble (天龍國), maybe this is part of what they are talking about. There are dramatic differences by age. People in their 20s are overwhelmingly for marriage equality, while people in their 30s are somewhat for it. However, most eligible voters are over 40, and these people are overwhelmingly against amending the Civil Code.

In May, the Council of Grand Justices ruled that the current law is unconstitutional and gave the government two years to change it. Looking at these numbers, you can see why the Tsai government is not eager to push through an amendment to the Civil Code, regardless of President Tsai’s personal sympathies. I don’t think the very vocal supporters of marriage equality have yet realized that the government is on their side. (Don’t forget, Tsai appointed most of the Grand Justices.) With these numbers, the only realistic action is a special law, which the activists don’t want. Instead, the government has chosen a third path: wait for the two year period to expire and then simply consider the Civil Code to allow gays to marry. At the cost of a two year wait, the marriage equality activists will get their most favored outcome while not inflicting enormous political costs on a sympathetic government.

 

The final item to consider echoes newly elected KMT chair Wu Den-yi’s proposal that Taiwan should revert to the 1992 Consensus. This was asked in the June 2017 survey.

在處理兩岸關係上,有人主張我們應該使用九二共識與中國大陸協商,也有人主張我們不應該再使用九二共識,請問您比較支持哪一種看法?

“On cross straits relations, some people say the we should use the 1992 Consensus as a basis for negotiations with mainland China, other people say that we should not use the 1992 Consensus again. Which side do you support?”

use Don’t use Doesn’t exist
All 41.8 29.4 3.7
DPP + NPP identifiers 21.6 52.9 6.9
Exclusive Taiwan identity 26.0 41.4 5.6

I have to admit, I was quite surprised by this result. 42% of people were in favor of re-adopting the 1992 Consensus, while only 33% were against it. (I’m counting the 4% who refuse to admit the existence of the 1992 Consensus as being against using it.) I guess Wu Den-yi’s position is more popular than I thought.

Let’s take a minute to think about polling and the 1992 Consensus. For years, the Ma Ying-jeou government would shove reams of polling data showing a solid majority in support of the 1992 Consensus in the face of any journalist willing to look. Many eagerly and unskeptically repeated the government numbers in their stories. It wasn’t just journalists, though. I’ve heard academics reference the Mainland Affairs Council survey numbers. Here’s the problem. The MAC was producing survey results in order to justify – not to inform – its policies. The typical MAC question wording was both leading and confusing. (Is it possible to be both leading and misleading?) Here’s a footnote from one of my recent papers on this point:

“For example, a July 2014 survey asked, “The government’s position on the 1992 Consensus is that the ‘one China’ in ‘one China, each side with its own interpretation’ refers to the ROC. Do you support this position?” 52.3% expressed support, which was a fairly typical result. Sometimes the MAC preceded this question with other leading questions or employed even more loaded question wordings. For example, a May 2015 survey asked, “Some people say, ‘Since 2008, the important result of the government’s mainland policy has been to maintain cross-straits relations and a stable peace.’ Do you agree with this statement?” It then asked a loaded question on the 1992 Consensus: “Since 2008, on the foundation of the 1992 Consensus – One China, each side with its own interpretation, One China means the ROC, the government has steadily promoted cross-straits negotiations and exchanges. Do you support this position?” 53.9% expressed support.”

Note that in all of the MAC surveys, the formula is spelled out in its strongest version, emphasizing “each side with its own interpretation.” This matters a lot. The more you spell out the parts of the formula that the PRC doesn’t agree with, the more support there is. In these questions, they further emphasize the constitution and confusingly (at least to me) allow people to think they are agreeing with the statement that the government’s position is that One China refers to the ROC (and not the PRC), and so on. That’s how you get over 50% for this question. When TISR asked the questions in a more neutral manner two years ago, they got about 40% support for “one China, each side with its own interpretation”, 30% for “1992 Consensus,” and 20% for “one China, both sides with the same interpretation.”

Nearly two years later, support for the 1992 Consensus seems to have risen a bit. The 1992 Consensus gets 42% support, even though the wording does not include the phrase “each side with its own interpretation.” Moreover, this item has a response category for opposition, not just for support. 42% support turns out not to mean 58% opposition. In fact, 33% is not anywhere close to 58% opposition. A large chunk of the population is ambivalent on this question. Like many people, I interpreted the 2016 election result as a death sentence for the 1992 Consensus. I still think the chances of the Tsai government ever accepting it are between razor-thin and zero, but, in light of this result, I can see why Wu Den-yi and the KMT are holding out hope that the 1992 Consensus can still be the basis for a winning election campaign.

 

To sum up, I think these data suggest that the DPP is still on track to win another term in 2020. There are some encouraging numbers for the KMT, but they are easily exaggerated. Overall, I think these data are at least as discouraging for the KMT as for the DPP. I think we are most likely going to have something like the previous Japanese election in which a somewhat unpopular government easily beat an even more unpopular opposition.

PS: If these results trouble you in any way, don’t worry. They’re all horribly out of date. Everything is probably completely different now.

One Response to “the state of (out of date) public opinion”

  1. Fading Rainbows: How to Overcome Taiwan's Anti-LGBT Referendums - Ketagalan Media Says:

    […] Some criticize the Democratic Progressive Party (DPP) for failing to act. Others point out that there is no perfect path forward for the DPP: they don’t necessarily have the support to push through a change to the civil code, and so they […]

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Google+ photo

You are commenting using your Google+ account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s


%d bloggers like this: