Global Intelligence Series

Support for Restricting Liberty for Safety: Evidence during the Covid-19 pandemic from China, Israel, Japan, South Korea, Taiwan, and the United States

Date October 23, 2020
Speaker Charles CRABTREE (Assistant Professor, Department of Government, Dartmouth College)
Commentator ONO Yoshikuni (Faculty Fellow, RIETI / Professor of Political Science, Faculty of Political Science and Economics, Waseda University)
Moderator WATANABE Tetsuya (Vice President, RIETI)

Democratic governments around the world have taken dramatic steps to halt the spread of COVID-19. These steps have prevented new infections and deaths, but they have also entailed unprecedented restrictions on civil liberties. Navigating this tradeoff between security and liberty is particularly difficult for democracies because they need to maintain public support for their policies and are constrained by their constitutions. We administered surveys to nationally representative samples in five economically advanced democracies—Israel, Japan, South Korea, Taiwan, and the United States—and one autocracy--China--to assess the extent to which the public supports liberty restrictions designed to combat COVID-19. We found consistent and widespread support for policies restricting civil liberties across all three countries. We also experimentally manipulated information about (1) the constitutionality of these policies and (2) the infections they would prevent, finding evidence that respondents' support for restrictions on civil liberties may depend more on their effectiveness than their legality.


Public support for unconstitutional policies in response to COVID-19

The COVID-19 pandemic has caused the largest simultaneous suspension of civil rights in recent history. Some important aspects of the government response to COVID-19 remain unknown. One question is whether the public supported these efforts. Governments around the world made serious efforts to protect their citizens by limiting their freedoms. The U.S. is an exception. We don't know what the public thinks about government actions that may have violated national law or otherwise exceeded a government's powers. I have the results of a large project that has tried to answer these questions.

Effect of constitutions in limiting government action

Our biggest question is how constitutions limited governments. Most governments have written constitutions. Constitutional limits are somewhat counter-intuitive because governments are supposed to be constrained by them while they are also charged with enforcing them. They are asked to be honest in restricting their own violations of the law, because fundamentally there is no supranational force that can restrict governments if they violate their own constitutions.

Most of the literature suggests that public opinion is the biggest check on constitutional violations. This puts pressure on governments and regulates what they can do: increased public opposition to constitutional violations leads to increased state compliance with the constitution. This is basically the only way in which constitutions actually restrain action. Unfortunately, we don't know whether or under what circumstances ordinary citizens will actually oppose a government or punish it for violations of their constitutional rights. Anecdotal evidence suggests that this rarely happens in practice. The empirical evidence shows mixed results on the extent to which the public is willing to sanction governments for overstepping constitutions or violating national law.

Little work has been done on this, particularly in times of emergency like the COVID-19 pandemic. One would think that public support for restrictions on civil rights should actually be at its highest during an emergency. We are interested in trying to understand public opposition to constitutional violations in a period of profound emergency. To what extent does the public support rights restrictions that are unconstitutional? How does this support vary internationally? Almost all of the research about the binding power of constitutional law focuses on the U.S. and Canada and we know very little about how the public in other countries might respond. We know that the U.S. is a very strong constitutional state compared to other countries, while in Japan, for example, the judiciary often defers to the executive and typically does not deem things to be unconstitutional, even if they arguably are.


These questions have broad implications for our understanding of the importance of constitutions and public law, public support for human rights and more broadly, interactions between the public and the state and how the public actually perceives radical policy actions by governments. We surveyed the public in the U.S., Japan, Israel, South Korea, Taiwan and China during the COVID-19 pandemic in various ways after registering our design, our expectations and our analytical strategy at the Open Science Foundation.

Nationally representative samples were polled. We had 3,012 participants in the U.S., 2,488 in Japan, a little more than 2,000 in Israel, 1,400 in South Korea, 800 in Taiwan and about 2,000 in China. These samples are broadly nationally representative and quota-based. We had quotas on age, gender, education and region and our samples match these key characteristics of national populations.

Each country was surveyed once. In the U.S., the survey took place in the middle of March. We surveyed Japan at the end of March with Nikkei Research. In Israel, we fielded the survey in the first week of April. China, South Korea and Taiwan were simultaneously surveyed at the end of May. These surveys were done at different times and also crucially during different periods of the COVID-19 pandemic within each country. In the U.S. in mid-March, there was a sense that Americans were going to have to go under a lockdown but it had not happened yet. In Japan, the survey predates Abe's state of emergency declaration on April 8. In Israel, the government actually had already radically limited human rights in various ways. South Korea, Taiwan and China were coming out of the first wave and things were easing up.

We recruited a total sample of about 11,603 respondents. We told respondents that the government—and we kept this kind of general, so it could be a subnational government like a city government or the national government—was considering several policies aimed at curbing, preventing and limiting COVID-19 infections. These policies would save a random, reasonable number of lives. We provided actual statistics from epidemiological forecasts and randomized the number people were shown. We said, "If the government adopts this policy, it'll save this many lives." We did this to ground the policy proposals in a real-life context, because there is always a tradeoff between rights restrictions and saving lives. That is the fundamental tension in all states. Each respondent was eventually asked to rate their support for nine different policy options.

Before respondents went on to actually rate the proposals, we told them that many legal experts believe that each of these policies may not comply with the country's constitution by, for instance, violating civil liberties or not providing sufficient due process. This experimental treatment was randomly assigned to half of the respondents in each survey. In the US, a random group of 1,500 of the 3,000-some total respondents received this additional information about how the proposals may be unconstitutional or illegal. This allows us to determine whether people respond differently when they are told that policies may be unconstitutional.

Policies surveyed

The nine different policies that we presented to people were all being actively considered by governments as ways of limiting the spread of COVID-19. One is prohibiting all nonessential movement of people outside their homes and banning all large gatherings. Another is prohibiting all people, including citizens currently outside the country, from re-entering the country. A third is prohibiting all non-citizens except legal permanent residents currently outside the country from entering the country. The fourth is suspending all nonessential public or government services, including all primary and secondary schools. The fifth is prohibiting all people from publicly spreading information about the virus that could have adverse health consequences, including downplaying the virus, or inducing panic; potentially a free speech violation. The sixth is taking over businesses and seizing physical property, including pharmaceutical companies or spaces typically used for government gatherings. This could violate several constitutional provisions depending on the country, including prohibitions on illegal search and seizure. The seventh is suspending all religious services or gatherings, including regular church services, which could infringe basic religious rights and liberty. The eighth was about the government requiring people to work for the government performing health care tests when there are labor shortages for things like assisting in hospitals or distributing medical supplies; a kind of medical conscription. Finally, the ninth policy is detaining any individuals exhibiting coronavirus-like symptoms and quarantining them in a government facility for at least two weeks.

All nine of these policies are reasonably unconstitutional according to some interpretations of the constitutions in each of the countries surveyed and also represent going above and beyond existing medical guidance and advice at the time that we fielded the survey. According to the best medical guidance that was available from places like the CDC in March, April and May, all of these policies would have been excessive and unnecessary.

Broad findings on support or opposition

The outcome of our statistical analysis is a four-point measure of support for the various policies, ranging from strongly opposed, which we give a one, to strongly supportive, which we give a four. I am first going to show you some basic descriptive results which simplify the four-level ordered measure of support into a binary one: whether people support a policy or not.

The existing literature believes that the public would be unwilling to support measures that strictly run against the constitution or that are really restrictive of their rights. We found the opposite: all of the policies are supported by the majority of respondents. A majority of people in every country that we surveyed believe that their rights should be restricted in all of these different ways in order to battle the spread of COVID-19. We find that of the nine different rights, 75% of respondents support the restriction of approximately seven of them on average. Rights restrictions in the face of a pandemic are strongly supported.

However, there is a lot of variation in support across countries for specific rights violations. Support for the government taking over property and banning all people from entering the country ranges from somewhere around 67% to 92% for banning all people from the country. Support for taking over businesses and property ranges from a little over 50% in the U.S. to about 92% in China. Support for suspending religious services, by contrast, ranges from 75% to 90%. Almost everyone across countries supports limiting religious liberty and freedom in the middle of the pandemic in order to save lives.

Support for policies indicated / not indicated to be unconstitutional

The big question we began with was to what extent the public is willing to support restrictions that are unconstitutional. Our secondary concern is the extent to which people who are informed that policies might be unconstitutional support them when compared with people who are not so informed. To do that, we estimated a series of ordinary least squared regression models. We take our four-point measure of support and we regress that on a dummy variable for whether or not people receive information about constitutionality, along with a set of pretreatment covariates about broad differences among individuals within our sample. We include in the model variables of the respondent's gender, age and education as well as how much they personally fear getting COVID-19 and how much they fear people that they know and love getting the virus.

In the case of Israel, telling people that policies might violate the constitution decreases support for prohibiting the spread of information. People are much less likely to support restrictions on free speech when told that they might violate the law and the same is true for taking over business and property, for conscripting people to work and for detaining people in government facilities. We can see a reasonable backlash from the public. What you see in Israel is how people process information about policies being unconstitutional. What you see in other countries is quite different.

In the U.S., you find that people are largely indifferent to whether you tell them that a policy might violate the constitution. That information does not cause them to indicate a different level of support for government policies. In Japan, people don't seem to care at all. The effect sizes across these different models are very, very small. They are all imprecise estimates. It seems that in Japan, telling people that policies violate the Japanese constitution does not decrease support in any measurable way for those policies.

In South Korea, we find roughly similar patterns to Japan. In some cases, there is almost a positive effect from telling people that a policy might violate the constitution; respondents actually occasionally seemed more likely to support policies that involve banning people from entering the country, non-citizens from entering the country and suspending religious services, although these findings are all statistically insignificant. Although we had a very small sample size, there is some evidence that telling people in Taiwan that policies actually violate the constitution leads to increased support for them. There's an unexpected backfire effect where telling people that something is actually unconstitutional leads to increased mobilization for those policies.

China has a very anomalous result. If you tell people in China that a civil rights restriction is unconstitutional, that the government is going to violate human rights in a way that is prohibited by the highest national law, this actually makes people more likely to support it. Nothing in our research design allows us to determine why this is the case, but our rough working theory at this point is that Chinese respondents interpret this to mean that if the policy is unconstitutional, it must mean that the government really thinks that it is necessary, making people more likely to support it.


Broadly, we find that the public across these six different places is very willing to support restrictions that are unconstitutional. We also find absolutely no evidence that some groups are more likely to respond to our constitutional treatment than others. The response to learning about constitutional infringement and how people process this information and whether or not they update or don't update their support for policies based on this information seems universal.

Some open questions remain. We really don't know to what extent Japanese residents support specific government policies. We have really good data on the extent to which they support the Prime Minister and the Cabinet, but we know very little about specific government policies. I think there is a lot of work that can be done to better understand that and make sure that the policies the Japanese government adopts are more representative of public preferences. We should also explore what the most effective arguments or framings are that the government can use to mobilize public support. We find that if the government provides information that some policies it is considering might be unconstitutional, this doesn't matter to the Japanese at all. However, other messaging frames may exist that might actually matter to the Japanese and figuring out a better way of framing policies and connecting with the population would be really useful. The way that you spell out certain policy proposals to the public really does matter in Japan. At the end of the day, there are many life-and-death policy issues in Japan on which we probably can bring additional insight, increased efficiency and better execution to prevent deaths and save lives.


Commentator: The first question I have is on the selection of countries. I am curious as to why you chose these particular regions and countries instead of choosing some European countries that have been similarly affected by COVID-19.

The second question is why you focused on comparison across countries in terms of research design. I think it might be a bit difficult to compare countries in this case because concepts of civil rights, for example, may differ as do government measures and responses to COVID-19. For example, the Japanese government didn't ban gatherings, which is different from the U.S. In Japan, government measures are mostly requests rather than compulsory. I think these things make it a bit difficult to explain the variations across countries.

The third question is about differences in responses. There may be significant differences between people depending on their partisanship and ideology. In the context of Japan, left-leaning people tend to be the ones asserting the unconstitutionality of government measures or policies .Such a tendency may explain the backlash effect you found in Taiwan. I don't know if you found any backlash effect in Japan but it may exist among LDP supporters. Political party support may be worth testing in that regard.

My next comment is that your presentation reminded me of the debate about the Patriot Act signed into law by the Bush Administration after September 11, 2001. The U.S. government restricted civil liberties under the Patriot Act and there were concerns that the government's "War on Terror" compromised civil liberties in favor of safety concerns. Do you think that your findings apply to other types of crises? Public health may differ from national security, but I am curious about the applicability of your findings to other domains.

Lastly, there are situations in which a government has to restrict civil rights. In those cases, how should the government communicate with its citizens effectively? The government needs to avoid overly frightening the public while at the same time giving them the right information.

Speaker: Going in reverse order: how does the government communicate the need for rights restrictions and radical policies to the public in a place like Japan? One takeaway potentially from our paper is that it doesn't hurt the government to explicitly acknowledge that these policies violate the constitution. In related work we find that the more the possible health benefits of policies are emphasized, the more likely people are to support them. I would suggest admitting to the possibility of unconstitutionality. We found that Japanese respondents were particularly supportive of policies that they are told will save lives. That seems to be specific to the Japanese population.

Your point about how public health may differ from security threats I think is probably accurate. I actually think that if all these countries were facing a similar security threat and we had done a study we would have found even less concern about the constitutionality of policies. People tend to underestimate the potential personal impacts of public health issues but tend to overestimate the likelihood that they're going to be hit by a terrorist attack. If you go back to the Patriot Act, the public in the U.S. was much more supportive of broad, sweeping policies that curtailed freedom in order to prevent something like 9/11, in which 3,000 people died, but now that we have 225,000 dead in America, many of which could have been prevented, the American public is shockingly resistant to supporting any kind of rights restrictions at all. A lot of this goes back to public perception.

On differences across respondents--we actually did investigate whether people that voted for the incumbent government respond differently to our constitution treatment. We do not find any differences. It is true that it is hard to make sense of the differences across countries and that many things differ across these countries, but we took great care to ensure that the survey instruments were as similar as possible in terms of meaning. In terms of language, we had everything double translated. We have reason to believe that the respondents across these countries correctly read these policies to be potential violations of freedom in the same ways.

On the final question about the countries selected: all of these countries are developed, economically successful countries or areas. All except China are democracies. The truth is that we did the study in the U.S. and we were going to do it in Israel because Israel has a really strong constitution and that would make an interesting comparison. And then opportunities became available through the Tokyo Foundation to field studies in Japan and across Asia. So, in many ways our sample selection was one of convenience. If we had it to do over again, we probably would have picked states that were in different regions in the world that faced different challenges when it came to COVID-19.


Q: Even if the constitution guarantees people's freedom, if people support their government's unconstitutional measures, does a constitution really function as an "iron cage" in preventing government violence?
Please let us know if you have any suggestions for the Japanese government related to this analysis.

A: On the first question, if people do not object to the government violating the constitution, then ultimately the constitution really does not have any effect in regulating the government and that should, I think, be alarming in light of broader geopolitical trends related to the erosion of civil rights around the world. This is in marked contrast to the flowering of freedom that we saw in the late 80s and in the early 90s. It is something that should concern anyone who cares about the spread of democracy around the globe. According to Freedom House and every other human rights organization, there has been a marked pushback on human freedom in the past several years and it seems like the public is willing to go along with this. They are not holding governments to the promises made in their constitutions, to the fundamental pact between the citizenry and the state, and that is deeply concerning for the future. It is a fascinating problem: the people that are most affected by a lack of liberty don't seem to care much about it and seem to lack an appetite for freedom.

As for suggestions for the Japanese government, in many ways Japanese law is relatively weak. In order for the Japanese government to accomplish something, they have to rely upon citizens volunteering to do it. We saw this acutely in the public response to COVID-19. The Japanese government was worried about overstepping legal boundaries and in trying to restrict rights in the same way that they were being restricted elsewhere, so they relied on the public to do the right thing. Many governments around the world have data labs that test different frames and messaging techniques to determine which ones actually work best at getting the public to do what the government thinks is necessary. The Japanese government is the only OECD country that does not have one of these labs. I think that is a glaring omission and something that probably should be rectified sooner rather than later. The costs of doing this are vastly exceeded by the economic gains from more effective policy pronouncements and engagement with the public.

*This summary was compiled by RIETI Editorial staff.