Perceiving Freedom: Civil Liberties and COVID-19 Vaccinations

Abstract

Why have some countries been more successful in their COVID-19 vaccine rollouts than others? Despite efforts by governments to vaccinate their adult populations against COVID-19, vaccination rates remain irregularly low in some countries. We suggest that a crucial piece of this puzzle lies in resistance against government directives from the public due to civil liberty protections. Countries with greater protections for civil liberties can be expected to have lower vaccinations administered than countries with fewer protections, as the public enjoys a sense of freedom regarding their private lives. In such countries, de jure constraints on government policies are complemented by the fear of public backlash, even in crises; consequently, beyond structural limitations, governments with high levels of civil liberty protections face an additional hurdle in managing the COVID-19 crisis. Evidence for this hypothesis is presented for 153 countries by combining civil liberty scores with newly available data on COVID-19 vaccinations.

Even though the novel COVID-19 coronavirus is less lethal than other viral diseases such as Ebola or Tuberculosis (Callaway et al., 2020), its higher infection rate has caused catastrophe for individuals and governments all around the globe. While initial research on the impacts of COVID-19 has been dominated by epidemiology and economics, there are additional puzzles emerging now whose answers lay squarely within the political domain.

One such puzzle is the asymmetric success rate enjoyed by governments globally for their COVID-19 vaccination mandates. After the initial uncertainty around the disease, the availability of various COVID-19 vaccines has given countries an opportunity to combat the spread and deadliness of the virus. Yet, in some countries, we observe lower levels of vaccinations than expected, while controlling for structural factors such as availability of the vaccine and national wealth. For example, the entire US adult population had access to COVID-19 vaccines since 19 April 2021 (Reuters, 2021a) but even in October 2021, only slightly over 60% of the US population had been fully vaccinated against COVID-19 (Centers for Disease Control and Prevention, 2021a). In contrast, in China—the country with the biggest population—vaccination rates were hovering around 80% (Reuters, 2021c) in the same period.

What explains this asymmetry in government success regarding COVID-19 vaccine rollouts? Factors such as government resources, population size, demographic characteristics, and so on surely play a role in terms of how well a government can respond to this crisis, but we suggest that there is a piece of the puzzle here that remains unexplored. In particular, we contend that protections for civil liberties could explain government success—or lack thereof—in implementing COVID-19 vaccine mandates and subsequent low vaccination rates. Crisis situations, such as the COVID-19 pandemic, are characterized by threat, uncertainty, and time pressure (Lipscy, 2020). In crises, governments would ideally like to have more discretion available to them to mitigate these three factors. However, in countries with more civil liberty protections, governments face greater constraints. In some instances, they are bound by law (i.e. a government cannot violate a civil liberty guarantee) but they may also be constrained in a de facto manner. These de facto constraints come from the overwhelming individualism and the sense of freedom that permeates such societies. In the presence of both formal protections and informal expectations for individual freedoms, governments cannot readily engage in ad hoc behavior without facing significant public backlash.

These constraints are crucial in the context of the COVID-19 pandemic, as they make managing the crisis vastly different in states that have high levels of civil liberty protections as compared to states that have low levels of such protections. In states with fewer civil liberty protections, governments can easily issue vaccine directives on an ad hoc basis without fearing significant public backlash, as formal and informal protections for civil liberties are rare. While governments without constraints are undesirable, such flexibility is also uniquely useful in crisis situations, as governments can act quickly, decisively, and unilaterally. Consequently, in the case of COVID-19 vaccines, several states with fewer civil liberty protections—such as Bangladesh, Pakistan, and Malaysia—have successfully instituted vaccine mandates that have gone unchallenged by the public.

On the contrary, states with robust civil liberty protections find it more challenging to manage crisis situations like the COVID-19 pandemic. Therefore, instituting vaccine mandates on a large scale becomes difficult for such countries. This is the case because government initiatives run into individualism and perceived constitutional protections, which creates public backlash. For instance, in Germany, Australia, and the United States, COVID-19 vaccination requirements have been criticized as government overreach, thus resulting in the absence of universal vaccine mandates. For this reason, we expect countries with more civil liberty protections to have lower vaccinations administered overall, while holding other factors constant. Evidence for this hypothesis is presented for 153 countries by utilizing newly available vaccination data for the first 11 months of the year 2021, which is used as the dependent variable in ordinary least squares (OLS) regression against civil liberty protection scores. The results suggest that governments with greater protections for civil liberties may continue to face hurdles in their plans for higher COVID-19 vaccination rates, even in situations when more vaccinations may be desirable from a public health perspective.

Argument

The COVID-19 pandemic has given political scientists an opportunity to explore crises in a different context. This article contextualizes the debate surrounding this pandemic in terms of political behavior, and in doing so, highlights the utility of political explanations that have remained unexplored. The argument developed here is a novel, first-of-its-kind attempt to explain differences in COVID-19 vaccination rates, an important public health and policy phenomenon, through political variables, such as civil liberty protections. The implications from our research are therefore multi-faceted, including lessons for the future regarding managing crisis situations, the dual-edged sword of government checks-and-balances, as well as the persistent nature of individual attitudes regarding the safeguard of civil liberties.

Civil liberties are a hallmark feature of liberal democracies, but they are not exclusive to such regimes. In fact, research shows that hybrid or even authoritarian regimes sometimes provide their citizens with some basic protections, albeit infrequently (Møller and Skaaning, 2013). Moreover, even liberal democracies sometimes infringe on civil liberties, especially in times of threat and crisis (Apetroe, 2016; Davis and Silver, 2004). Civil liberties typically include several freedoms, namely, the freedom of speech, opinion, demonstration, and peaceful dissent. Other freedoms that may fall under the umbrella of civil liberties are freedom of movement and religion, freedom of association, and freedom from political killings and torture by the government. Individual countries may provide additional freedoms to their citizens, but the important point to note here is that all such protections are geared toward protecting individuals from governmental abuse. Hence, in the context of this article, civil liberties are considered as negative rights (Fried, 1973), that is, the right of an individual to not be interfered with in forbidden ways, as opposed to a positive right which is a claim to something (access to health care, for instance). The way in which civil liberty protections can be introduced and adopted by government is not the main focus of this article, but it is worth mentioning that such protections can emerge under pressure from the public, as doing so grants longevity to the incumbent regime and may also boost the government’s international reputation (Vreeland, 1999). Our argument therefore considers civil liberties protections as they are in various countries as given, and uses them as an explanatory factor toward another outcome, namely, the asymmetric success of governments in their vaccination rollouts.

As stated previously, the COVID-19 pandemic is a crisis situation, characterized by the triumvirate of threat, uncertainty, and time pressure (Lipscy, 2020). An additional, distinguishing characteristic of the COVID-19 outbreak is its exogenous nature, as most countries have had to make critical adjustments due to developments largely away from their own borders. Such abrupt disruptions exacerbate the threat, uncertainty, and time pressure dynamics of crises, and pandemics therefore provide incentives to policymakers for making ad hoc interventions in a centralized manner. Such interventions could frequently be desirable nonetheless, as ensuring public health and safeguarding economic well-being—two features of modern political life that are inexorably intertwined—is a priority for most regimes around the world.

Given this urgency toward managing the COVID-19 crisis, the lack of governmental success in some countries toward administering vaccines to their publics is puzzling. Research into the safety of COVID-19 vaccines has shown that they are overwhelmingly safe (Centers for Disease Control and Prevention, 2021b) and owing to uncharacteristic cooperation at the international level, they are also readily available (Bloomberg, 2021; British Broadcasting Corporation, 2021a). Nonetheless, vaccination rates for eligible populations remain irregularly low, even when accounting for structural differences. For example, in Singapore, over 82% of the population had been totally vaccinated in October 2021, even though the vaccine was only made available in July of the same year (The Straits Times, 2021). The higher vaccination rates in Singapore may be a function of the smaller population size, but in China—the country with the biggest population—vaccination rates reached around 80% (Reuters, 2021c) while countries like Australia (55%) (Australian Government Department of Health, 2021), Brazil (57%) (Reuters, 2021b), and Germany (64%) (Reuters, 2021d) lagged significantly behind in the same time frame; in this period, the United States also lagged behind at around 64% vaccination rates for the vaccine eligible population (Centers for Disease Control and Prevention, 2021a). Typically, these industrialized democracies would be expected to exceed vaccine projections, especially since the vaccine was made available early on in countries like the United States and Germany. Therefore, their lackluster success in COVID-19 administering vaccinations is puzzling.

We suggest that this anomalous behavior can be explained by resistance faced by governments in countries that accord high levels of civil liberty protections to their citizens. In countries such as the United States, Australia, and Germany, de facto and de jure protections for individual rights have resulted in hesitancy toward accepting governmental action, even in crisis situations when the potential for loss of life is high. This is the case because the public enjoys protections from government interventions in their private lives under normal circumstances. In many cases, de jure protections, such as laws and statutes, limit the government in its jurisdiction over private matters. In addition, a sense of individualism and personal freedom also permeates such societies, which not only results in de facto limitations on government actions—public backlash in case of unpopular actions by the government, for instance—but also provides an inventory to the public to assess government decisions in times of crises. In this way, both formal and informal protections provide the public with expectations for their private lives to be free from government encroachment.

Research in the field of cultural psychology supports this assertion: groups with relatively few ecological and historical challenges to their survival, such as high population density, external invasions, and disease outbreaks, evolve to have “loose” cultures—ones where social norms and rules are fairly lax and individual liberties are prized (Gelfand et al., 2011). Such research categorizes countries such as the United States and Brazil as having “looser” cultures. In contrast, “tight” cultures are characterized by few individual freedoms and greater societal restrictions; subsequently, countries such as China, Singapore, and Pakistan are considered to have “tight” cultures. We surmise that these loose–tight cultural frameworks have a direct analogue in Political Science discourse in the form of civil liberties and the degree of associated individual protections. To this end, we contend that having experienced greater protections for civil liberties in the past makes the public more likely to resist ad hoc decisions taken by the government. As such decisions are characteristic of crisis situations, we suggest that governments can face resistance from their constituents in taking unilateral actions, such as instituting vaccination mandates during this COVID-19 pandemic. Cultural psychology research shows that more protections for civil liberties directly relate to more resistance among the public in accepting government actions, especially in times of threat and uncertainty (Gelfand, 2018). Therefore, in countries where civil liberty protections are high, governments are relatively unsuccessful in implementing strict vaccination requirements. While a vaccine mandate does not necessarily constitute a violation of civil liberties, the robust sense of individualism and protection from government overreach in these countries make vaccine mandates extremely unpopular.

In contrast, countries with fewer civil liberty protections are typically characterized by government overreach, which can become particularly expansive in crises. In such countries, formal and informal protections for protecting civil liberties are low—fewer checks on executive and the threat of coercion in the face of protests, for example. This already implies that governments face less resistance in implementing their policies, as institutional checks-and-balances are ineffective, and informal avenues to protect civil liberties for the public are rare. Once a crisis situation replete with threat, uncertainty, and time pressure emerges, governments in such countries respond unilaterally, and we suggest that such actions—vaccination mandates, for instance—face relatively higher success in such countries. This is the case because the public generally has fewer freedoms and opportunities for individuality and expression. Therefore, they are more likely to comply with government mandates, as they have no other realistic option.

This proposed mechanism can be seen at play during the current COVID-19 crisis. For example, since gaining access to the vaccine in March 2021, Pakistan—a country with few civil liberty protections—had outlined a series of vaccine mandates and strict restrictions for unvaccinated people. In June 2020, the government required all teachers and administrative staff at academic institutions to be vaccinated (The Independent, 2021); unvaccinated personnel were suspended without pay unless they could provide proof of immunization. All students aged 18 and above were also required to get vaccinated, and unvaccinated students could not enter the university campus. The government also mandated vaccines for employees in the public and private sectors (Dawn News, 2021a), and all domestic and international air travel for the unvaccinated was banned (Dawn News, 2021b). Moreover, vaccine requirements have also been instituted to enter enclosed spaces or even for buying basic consumer goods, such as fuel for vehicles (Geo News, 2021). Whereas such curbs are quite overreaching in their scope, ¹ there are two important factors to note here: first, the crisis management arm of the Pakistani government, the National Command and Control Center (NCOC), has been criticized for aggregating authority unduly during this pandemic as its decisions are not open to review (Afzal, 2021). The NCOC also includes the Pakistani military, which provided its surveillance services to the government to keep tabs on citizens and their movements during this pandemic (Al Jazeera News, 2021). Second, there have been no reported protests or observable public backlash in response to these actions, to date. Both these factors point toward the ease with which countries with low civil liberty protections can institute curbs and vaccine mandates regarding COVID-19 without public backlash, thus making the management of the crisis easier.

In contrast, Australia, a country with more civil liberty protections, faced significant backlash and mass protests over lockdowns and vaccine mandates for workers in certain industries. The Victorian state government initially mandated the vaccine for workers in the construction industry and temporarily shut down the industry in an attempt to stop the spread (Victorian Building Authority, 2021). In response, construction workers and other groups who oppose mandatory vaccination clashed with the police in a multi-day violent protest event (NPR, 2021). These protests have since branched off into bigger movements that have turned violent (British Broadcasting Corporation, 2021b). Similarly, in the United States, government actions to impost vaccine mandates on public sector employees have faced significant resistance. For instance, police personnel in the US state of Massachusetts have resigned in protest over the state government’s decision to implement vaccine requirements for public employees (CBS News, 2021) while nursing staff in the state of New York are also resigning in protest against mandatory COVID-19 vaccinations (Reuters, 2021f). Similar public backlash emerged in Germany (Dawn News, 2021c) and Canada (Global News, 2021) as well. These patterns of public backlash against mandatory COVID-19 vaccinations in such countries speak to our mechanism at play, thereby providing answers to the puzzle of low COVID-19 vaccinations and also highlighting the difficulties that governments in countries with high civil liberty protections face in managing crises.

While research in Political Science is catching up, cultural psychology research provides more evidence in this regard. In particular, Gelfand et al. (2021) show that countries with greater cultural looseness—for example, Germany and the United States—have had a harder time managing the number of COVID-19 cases and subsequent deaths in comparison to countries with less individual liberties and strong social norms. ² Gelfand et al. (2021) attribute this finding to the lack of preparedness for facing disease outbreaks, as modern developed countries have evolved without significant external threats. But we suggest that this outcome is more informative than simply being reduced to arguments about preparedness to respond to pandemics. In particular, we suggest that more civil liberties necessarily imply that governments with formal and informal constraints are going to have a harder time implementing restrictions—such as the ones on public movements, as well as mask and vaccine mandates—which means that not only will such countries start out from a relatively disadvantageous position, but they will also have a harder time catching up with countries where governments can take unilateral actions without fearing public backlash. Keeping this in mind, our main hypothesis is stated below:

• Hypothesis. Countries with more civil liberty protections have fewer COVID-19 vaccines administered as compared to countries with fewer civil liberty protections.

It must be noted here that while democracies are likely to have more civil liberty protections overall, our argument is not simply about regime type. Civil liberty protections is one significant characteristic which can be used to classify regimes, but importantly, it allows for charting out a specific mechanism that produces changes in our dependent variable of choice, that is, vaccines administered. Other features used to classify regimes—such as electoral rules, executive authority, judicial constraints, and so on—may contribute here, but none of them provide a baseline for comparison among different countries in crises, as crisis management is bound to equalize government discretion. In other words, all governments—democratic or otherwise—would prefer to have more discretion in crises, but the fact that only some governments face resistance in their actions while others do not makes for an interesting puzzle. We contend that this puzzle can be resolved by looking at explanations grounded in civil liberty protections. Therefore, our approach leads to a more generalized explanation for understanding political behavior among different countries.

Another caveat worth mentioning is that we are not suggesting that vaccination requirements are a violation of civil liberties. In times of public health crises, governments rightly need to act in ways that seek to reduce the loss of life. Vaccinations are a useful tool in this regard, and even in countries with extremely high freedoms and individualism, such as the United States, the government’s right to require vaccinations has been accepted. For example, in 1905, in Jacobson v. Massachusetts, the US Supreme Court upheld Massachusetts’ mandatory vaccination statute, thus giving state governments the authority to require vaccines. Nonetheless, our theory suggests that the reason why vaccination mandates have been met with resistance in countries with significant civil liberty protections is due to the socio-political inventory that the public has with regard to protecting their individual freedoms at all costs. While the role of misinformation cannot be ignored, vaccination mandates becoming a position issue is a direct consequence of the sense of individualism that pervades such societies. Any government action is likely to be met with public resistance here, and resistance to COVID-19 vaccination mandates merely points to their contemporaneous nature. Other government actions in the same vein—mask mandates, vaccine requirements for government employees, and so on—are also likely to face the same resistance, as the US examples demonstrates. To this end, the following section outlines the research design undertaken to test our hypothesis.

Data and Methods

The unit of analysis is country year, and the sample includes 153 countries for which vaccination data are available for the year 2021, starting from January and ending in November; therefore, our analysis focuses on variation among units and not within. The sample contains a variety of states and can be considered reasonably representative. ³ The dependent variable is the natural log of total vaccinations administered in a given country from the day of their first availability in 2021 until November of the same year. These vaccination numbers are taken from the Our World in Data resource (Ritchie et al., 2020), which compiles data on COVID-19 vaccinations from publicly available government resources for all countries. ⁴ As the dependent variable is normally distributed, OLS regression is used with robust standard errors, clustered by countries.

The main independent variable is the Civil Liberties Index from the Varieties of Democracy dataset (Coppedge et al., 2021). This variable ranges from 0 (low) to 1 (high) and codes for liberal freedoms in a given country. The dependent variable (vaccinations administered) is sourced from data for the year 2021; however, the V-Dem dataset does not have civil liberty scores for the same year. Thus, the scores for the year 2020 are used. Despite the lack of data availability, we consider this choice to be reasonable for two reasons: first, civil liberty scores are unlikely to change radically from year-to-year without revolution or wholesale regime change, thus reducing the possibility of error. Even in extreme circumstances, the civil liberties within a country are bound to change incrementally. More importantly, using the civil liberty scores for 2020 is also useful as our argument rests on a public’s prior experience with individual freedoms. For our analysis, resistance to vaccination mandates is likely to come from a public’s recent experience, thus making this choice reasonable. ⁵

In addition, to account for other structural factors that can impact overall vaccination numbers, we use the following variables from the Our World in Data project (Ritchie et al., 2020). First, we use the total population (millions) in our analysis, as bigger populations are likely to produce higher vaccination numbers overall. In a similar vein, we use the population density estimates to account for the urban–rural divide, as higher population densities result in more disease cases, and thus require more vaccinations.

Second, as the starting date for vaccine availability for the public varies in the sample, early vaccine availability, coded as the total number of days from when the first vaccine doses are recorded in the data until the end of the sample time period (November 2021), is used as a control variable. Earlier availability of the vaccine—as was the case in United Kingdom, for instance—is likely to increase the number of overall vaccinations recorded in the data. Third, we use the life expectancy (years) for each country in our analysis to control for the possibility that a higher life expectancy results in an aging population, which is likely to require more vaccinations as COVID-19 disproportionately affects older citizens in an adverse manner.

Fourth, in order to control for differences in national wealth and standards of living, we use the gross domestic product (GDP) per capita (in US dollars, thousands) figures. Higher values for GDP per capita can be expected to reduce the number of vaccinations overall (negative beta coefficient) as they can be considered proxies for a higher standard of living. A higher standard of living, such as easy access to public health facilities, better sanitation, and higher educational achievements, is likely to result in lower COVID-19 cases, thus reducing the need for urgent vaccinations. Finally, we use the government stringency index (ranging from 0 to 100); this control variable codes for the severity of closures instituted by each country in the sample, and it serves as a proxy for potential disease spread in our sample. Table 1 provides the summary statistics for all variables in our analysis.

Table 1.

Summary Statistics.

Variables	Expected direction	N	Mean	SD	Min	Max
Total vaccines administered (ln)		153	15.81	1.99	8.93	21.73
Civil liberties index (V-Dem)	–	153	0.69	0.24	0.12	0.96
Total population (millions)	+	153	49.88	166.35	0.10	1444.22
Population density	+	153	0.20	0.67	0.00	7.92
Early vaccine availability	–	153	308.16	53.15	66.00	387.00
GDP per capita	–	153	18.75	19.86	0.66	116.94
Life expectancy (years)	+	153	72.68	7.56	53.28	84.63
Strict lockdown measures	+	153	69.87	17.87	11.11	97.22

SD: standard deviation; GDP: gross domestic product.

Results and Discussion

Our theory suggests that countries with more civil liberties will have lower COVID-19 vaccinations, all else constant. Table 2 contains the results for OLS regressions undertaken to test this hypothesis. For Models 1a and 1b, the dependent variable is the log of total vaccinations administered in a country, while the main independent variables are the 2020 civil liberties scores from the V-Dem dataset (Coppedge et al., 2021) and the Freedom House data (Freedom House, 2021), respectively. For Model 1c, the dependent variable is the log of total COVID-19 vaccinations administered in a country per 100 people (Ritchie et al., 2020). This dependent variable allows for additional control over the population differences worldwide. Finally, Model 1d has the log of total people vaccinated in the country as the dependent variable, as a robustness check. ⁶

Table 2.

Predicted Number of Vaccinations Given Civil Liberties (OLS).

Variables	Model 1a	Model 1b	Model 1c	Model 1d
	DV: Total vaccines	DV: Total vaccines	DV: Total vaccines/100	DV: Total people vaccinated
Civil liberties index (V-Dem)	–2.035***		–1.979***	–1.979***
	(0.464)		(0.473)	(0.473)
Civil liberties index 2020 (FH)		–0.032***
		(0.008)
Total population (millions)	0.004***	0.004***	0.004***	0.004***
	(0.001)	(0.001)	(0.001)	(0.001)
Population density	–0.075	–0.108	–0.059	–0.059
	(0.143)	(0.134)	(0.151)	(0.151)
Early vaccine availability	0.015***	0.015***	0.015***	0.015***
	(0.004)	(0.004)	(0.004)	(0.004)
GDP per capita	–0.016**	–0.015**	–0.018**	–0.018**
	(0.007)	(0.007)	(0.007)	(0.007)
Life expectancy (years)	0.080***	0.092***	0.064***	0.064***
	(0.024)	(0.024)	(0.024)	(0.024)
Strict lockdown measures	0.014**	0.013**	0.013*	0.013*
	(0.007)	(0.006)	(0.007)	(0.007)
Constant	5.795***	4.621***	6.610***	6.610***
	(1.229)	(1.279)	(1.228)	(1.228)
Observations	153	153	153	153
R2	0.567	0.566	0.525	0.525

Robust standard errors in parentheses. OLS: ordinary least squares; DV: dependent variable; FH: Freedom House; GDP: gross domestic product.

^*p < 0.1, **p < 0.05, ***p < 0.01.

As the results demonstrate, our hypothesis finds empirical support in our sample of 153 countries. Higher civil liberty scores reduce the number of overall COVID-19 vaccinations. In other words, countries with more civil liberty protections have fewer people vaccinated against COVID-19. Figure 1 charts the predicted vaccines administered over the values of the civil liberties index from Table 2. As can be seen, the plotted predictions demonstrate that countries with more civil liberties have lower COVID-19 vaccinations overall.

Figure 1.

Predicted Number of Vaccinations Given Civil Liberties (OLS).

Conclusion

Our theory suggests that various countries facing difficulty in their vaccine mandates could be due to the individual freedoms and sense of individualism that permeates societies with high civil liberty protections, both formal and informal. The empirical evidence presented here supports this hypothesis in a sample of 153 countries. As this is an early results paper, we recognize the need for further analysis to better understand the mechanism we outline here. We will continue to work on this project to develop a full paper, which will include two case studies that illustrate the relationship between civil liberty protections and vaccination rates. While our early results are robust, we also acknowledge that factors such as government effectiveness in communicating the threat from the disease could condition the relationship we discuss here; we will consider this in a full paper by including relevant data. We also hope to incorporate more data on subsequent waves of COVID-19 to see whether there is a point at which the public in states with few civil liberty protections are no longer willing to comply with government mandates and shutdowns.

Nonetheless, a few important policy implications follow from our analysis. First, a large swathe of governments in developed, institutionalized democracies are likely to keep facing resistance toward mandatory vaccinations against COVID-19 in the future, as the public continues to challenge government decisions involving their private lives. Second, the number of overall vaccinations is likely to increase with time, but public outcry will also persist. For instance, in Australia, public protests against lockdowns and mandatory vaccinations have increased, even when more and more people are getting vaccinated (British Broadcasting Corporation, 2021b).

On the contrary, governments in countries with fewer civil liberty protections will become more successful in instituting and implementing vaccine mandates as a response to this pandemic, but we must not ignore the human costs associated with such success. In other words, while decisive action in this case is not just desirable but also required, it must be balanced against the need to protect personal freedoms nonetheless. Finally, our theory does not doom countries with greater civil liberty protections to eternal failure regarding the management of the COVID-19 crisis. Historical evidence suggests that governments can manage crises more effectively if they can communicate the severity of the threat to the public in a coherent manner. For instance, the US government was able to successfully rally support for its policies after 9/11, even when those policies were likely to intrude on private liberties (Hetherington and Suhay, 2011). In addition, research on recent developments in Czech Republic shows that heightened threat perception can reduce public support for their own freedoms (Muzik and Serek, 2021). Therefore, the public can become a crucial ally toward making government actions succeed, but that can only be achieved once the severity of the threat has been established.

Author biographies

Hayley Munir is an Assistant Professor of Political Science at Albright College, USA. She specializes in law and courts. Syed Rashid Munir is an Assistant Professor of Political Science at Forman Christian College University, Pakistan, and specializes in domestic sources of foreign policy

Appendix 1

Countries in the Sample

Table A1 lists the 153 countries in the sample.

Table A1.

The 153 Countries Used in the Empirical Tests.

Afghanistan	Greece	Norway
Albania	Guatemala	Oman
Algeria	Guinea	Pakistan
Angola	Guyana	Panama
Argentina	Haiti	Papua New Guinea
Australia	Honduras	Paraguay
Austria	Hungary	Peru
Azerbaijan	Iceland	Philippines
Bahrain	India	Poland
Bangladesh	Indonesia	Portugal
Barbados	Iran	Qatar
Belarus	Iraq	Romania
Belgium	Ireland	Russian Federation
Benin	Israel	Rwanda
Bhutan	Italy	Saudi Arabia
Bolivia	Jamaica	Senegal
Bosnia and Herzegovina	Japan	Serbia
Botswana	Jordan	Seychelles
Brazil	Kazakhstan	Sierra Leone
Bulgaria	Kenya	Singapore
Burkina Faso	Korea, Rep.	Slovak Republic
Burundi	Kuwait	Slovenia
Cambodia	Kyrgyz Republic	Solomon Islands
Cameroon	Lao PDR	South Africa
Canada	Latvia	Spain
Central African Republic	Lebanon	Sri Lanka
Chad	Lesotho	Sudan
Chile	Liberia	Suriname
China	Libya	Sweden
Colombia	Lithuania	Switzerland
Congo Democratic Republic of	Luxembourg	Tajikistan
Congo, Rep.	Madagascar	Tanzania
Costa Rica	Malawi	Thailand
Croatia	Malaysia	Togo
Cyprus	Mali	Trinidad and Tobago
Côte d’Ivoire	Malta	Tunisia
Denmark	Mauritania	Turkey
Djibouti	Mauritius	Turkmenistan
Dominican Republic	Mexico	Uganda
Ecuador	Moldova	Ukraine
Egypt, Arab Rep.	Mongolia	United Arab Emirates
El Salvador	Morocco	United Kingdom
Estonia	Mozambique	United States
Ethiopia	Myanmar	Uruguay
Fiji	Namibia	Uzbekistan
Finland	Nepal	Vanuatu
France	Netherlands	Venezuela, RB
Gabon	New Zealand	Vietnam
Gambia	Nicaragua	Yemen, Rep.
Georgia	Niger	Zambia
Ghana	Nigeria	Zimbabwe

Discussion on Control Variables

This discussion analyzes the beta coefficients from Table 2 of the manuscript (Models 1a–1d). Keeping the results of Table 2 in mind, we can see that several control variables impact vaccination rates. First, the results shows that, consistent with our expectations, countries with more citizens have higher vaccination rates. Our model also confirms that the length of time vaccines have been available in a country increases vaccination rates. This is intuitive, as more time to administer vaccines creates more opportunities for vaccination and the normalization of vaccination within a society. In addition, the life expectancy variable also has a meaningful impact on vaccination rates. As life expectancy increases, vaccination rates also increase; this result is reasonable as we know that COVID-19 disproportionately impacts older populations. Gross domestic product (GDP) per capita is also significant in the negative direction. The variable shows that as GDP increases, vaccination rates decrease. This is consistent with our expectations, as Western industrialized democracies tend to have higher GDPs and significantly greater civil liberty protections. Finally, strict lockdown measures—acting as a proxy for disease incidence—is significant with a positive beta coefficient, as expected by our theory.

Table A2.

Predicted Number of Vaccinations Given Civil Liberties (OLS).

Variables	Model A1	Model A2	Model A3	Model A4
Civil liberties index (V-Dem)		–2.116***	–2.054***	–3.528***
		(0.461)	(0.464)	(1.061)
Civil liberties index 2021 (FH)	–0.033***
	(0.008)
Total population (millions)	0.004***	0.004***	0.004***	0.004***
	(0.001)	(0.001)	(0.001)	(0.001)
Population density	–0.119	–0.052	–0.171	0.034
	(0.134)	(0.139)	(0.131)	(0.154)
Early vaccine availability	0.015***	0.015***	0.013***	0.015***
	(0.004)	(0.004)	(0.003)	(0.004)
GDP per capita	–0.015**	–0.019***		–0.020**
	(0.007)	(0.006)		(0.008)
Life expectancy (years)	0.094***	0.083***	0.063***	0.088***
	(0.024)	(0.024)	(0.022)	(0.024)
Strict lockdown measures	0.013**	0.014**	0.017**	0.015**
	(0.006)	(0.007)	(0.007)	(0.006)
Regime type				0.132
				(0.217)
Secure economic rights				–0.006
				(0.009)
Civic strength				–0.725
				(1.129)
Media censorship				0.388**
				(0.158)
Constant	4.527***	5.662***	7.374***	6.567***
	(1.277)	(1.218)	(0.974)	(1.344)
Observations	153	152	155	152
R2	0.571	0.565	0.553	0.594

Robust standard errors in parentheses. OLS: ordinary least squares; FH: Freedom House; GDP: gross domestic product.

^*p < 0.1, **p < 0.05, ***p < 0.01.

Robustness Checks

Additional Controls and Alternative Explanatory Variable

Table A2 contains the results of OLS regression undertaken with different model specifications. Model A1 replaces the civil liberties index from the V-Dem dataset (Coppedge et al., 2021) with the values for civil liberties scores from the Freedom House data (Freedom House, 2021) for the year 2021. ⁷ Even though a public’s historical experience with civil liberties is important, the current state of civil liberty protections is also likely to inform our analysis as such protections seldom change from one year to another.

In addition, Model A2 uses the same model specification as Model 1a in Table 2 of the article, but excludes one country, the United States. Given the overwhelming tendency of the American public to vociferously resist any government decisions seen as intrusions on public liberties, this exclusion is reasonable. In the case of the COVID-19 crisis, resistance to vaccine mandates has been vocal and organized in the United States, thus necessitating this robustness check to ensure that our results are not driven by this outlier. Model A3 strives to eliminate the potentially confounding effect of one specific covariate—GDP per capita—as it is correlated with life expectancy. Finally, Model A4 introduces several additional control variables, such as regime type, civic strength (robustness of the civil society), media censorship from the government (Coppedge et al., 2021), and secure economic rights (Miller et al., 2020). These variables are added to account for additional factors responsible for contributing toward vaccination rates. These additional controls are nonetheless not included in the main manuscript as they are possible confounders, and additional testing is done in the “Endogeneity Concerns and Instrumental Variable Regression” section to account for this. Nonetheless, as Table A2 shows, our hypothesis still finds support when the parameters of the empirical model are changed.

Determinants of the Civil Liberty Index

The civil liberties index from the V-Dem data (Coppedge et al., 2021) is a composite index of three other variables, namely, the protections for physical, political, and private liberties of citizens. These variables are also available from the V-Dem dataset and are used in Table A3 as alternative independent variables. The motivation behind Models B1–B3 is to see if any of the constituent components of the civil liberties index are uniquely responsible for the significant beta coefficient of the composite index. As the results in Table A3 demonstrate, our results are not an artifact of any one component of the civil liberties index, as each of the three components have significant (negative) beta coefficients. These results provide an additional robustness check for our statistical analysis.

Table A3.

Predicted Number of Vaccinations for Components of Civil Liberties Index (OLS).

Variables	Model B1	Model B2	Model B3
Physical liberties	–2.039***
	(0.396)
Political liberties		–1.413***
		(0.405)
Private liberties			–1.831***
			(0.512)
Total population (millions)	0.004***	0.004***	0.004***
	(0.001)	(0.001)	(0.001)
Population density	–0.074	–0.089	–0.031
	(0.153)	(0.137)	(0.136)
Early vaccine availability	0.015***	0.015***	0.016***
	(0.004)	(0.004)	(0.004)
GDP per capita	–0.012*	–0.018**	–0.019**
	(0.007)	(0.007)	(0.007)
Life expectancy (years)	0.085***	0.073***	0.075***
	(0.023)	(0.024)	(0.024)
Strict lockdown measures	0.014**	0.014**	0.014**
	(0.006)	(0.007)	(0.007)
Constant	5.456***	6.020***	5.963***
	(1.200)	(1.254)	(1.233)
Observations	153	153	153
R2	0.581	0.553	0.553

Robust standard errors in parentheses. OLS: ordinary least squares; GDP: gross domestic product.

^*p < 0.1, **p < 0.05, ***p < 0.01.

Endogeneity Concerns and Instrumental Variable Regression

In order to rule out the effect of confounding variables, this section presents the results of instrumental variable (IV) regression undertaken in the empirical sample. The motivation for doing so is simple: to get at the causal effect of an explanatory variable, IV analysis is used to control for unmeasured confounding when perfectly randomized experiments are not available. Following Woolridge (2009), IV regressions require a valid instrument, which is a variable that (a) is independent of the unmeasured confounding and (b) affects the outcome only indirectly through its effect on the dependent variable.

Put simply, we need a variable that is highly correlated with our main explanatory variable—civil liberties—and its effect on our dependent variable—vaccines—is only evident through its effect on civil liberties (no or low correlation with the error term, therefore). The V-Dem dataset (Coppedge et al., 2021) provides just a variable, that is, regime type. As civil liberties vary significantly across different regime types—democracies and autocracies—we use this variable to account for such institutional differences. This variable is a good instrument as it satisfies the conditions outlined above: it is highly correlated with civil liberties, it is almost un-correlated with the error term, and it also conceivably only impacts vaccine uptake through our main explanatory variable (civil liberties). Put simply, as some regimes allow for more civil liberties, such governments are also faced with a greater resistance once they start implementing ad hoc decisions. Therefore, this IV enables us to control for any endogeneity concerns in the data through its use in a two-stage least squares (2SLS) setup. ⁸ Table A4 provides the results of this IV estimation (Model C1) along with several other IV regressions. As can be seen, our hypothesis finds support in this 2SLS estimations.

Table A4.

Predicted Vaccines Administered Given Civil Liberties (V-Dem)—2SLS Models.

Variables	Model C1	Model C2	Model C3
	IV: Regime type	IV: Judicial legitimacy	IV: 2012 civil liberties
Civil liberties index (V-Dem)	–1.518**	–1.847***	–1.792***
	(0.637)	(0.669)	(0.549)
Total population (millions)	0.004***	0.004***	0.004***
	(0.001)	(0.001)	(0.001)
Population density	–0.062	–0.070	–0.069
	(0.134)	(0.138)	(0.137)
Early vaccine availability	0.015***	0.015***	0.015***
	(0.004)	(0.004)	(0.004)
GDP per capita	–0.016**	–0.016**	–0.016**
	(0.007)	(0.007)	(0.007)
Life expectancy (years)	0.075***	0.078***	0.078***
	(0.023)	(0.023)	(0.023)
Strict lockdown measures	0.014**	0.014**	0.014**
	(0.006)	(0.006)	(0.006)
Constant	5.887***	5.829***	5.838***
	(1.190)	(1.190)	(1.194)
Observations	153	153	153
R2	0.564	0.567	0.567

Robust standard errors in parentheses. 2SLS: two-stage least squares; IV: instrumental variable; GDP: gross domestic product.

^*p < 0.1, **p < 0.05, ***p < 0.01.

In Table A4, Models C2 and C3 are of particular interest as well. Model C2 uses judicial legitimacy scores for the year 2020 from The World Bank (2020) data as a possible instrument. The inclusion of this variable as an additional instrument serves two purposes: it demonstrates that our theory has empirical evidence when data from other sources are used, and it also serves as a robustness check as judicial legitimacy can be reasonably expected to increase civil liberties but its impact on vaccinations is not direct, thus making it a good instrument. For Model C2 in Table A4, we find empirical evidence for our hypothesis.

Furthermore, following Lorentzen et al. (2014), Model C3 uses an 8-year lag of the main explanatory variable (civil liberties index) as an instrument. This approach is more econometric than theory-driven, but the instrument nonetheless qualifies for use as impact of the civil liberty score from such a long time ago on the dependent variable can only reasonably be through its current values. Model C3 presents the results of this estimation in Table A4, and we find empirical support for our hypothesis in this case. In the same vein, Table A5 estimates IV regression by substituting the civil liberties scores from the V-Dem data with its Freedom House counterpart. The two main instruments used earlier, regime type and judicial legitimacy, are used here, and Table A5 demonstrates that our statistical evidence corresponds with our hypothesis.

Table A5.

Predicted Vaccines Administered Given Civil Liberties (Freedom House)—2SLS Models.

Variables	Model D1	Model D2
	IV: Regime type	IV: Judicial legitimacy
Civil liberties index 2020 (FH)	–0.023**	–0.026***
	(0.010)	(0.009)
Total population (millions)	0.005***	0.005***
	(0.001)	(0.001)
Population density	–0.085	–0.092
	(0.127)	(0.128)
Early vaccine availability	0.015***	0.015***
	(0.003)	(0.003)
GDP per capita	–0.015**	–0.015**
	(0.007)	(0.007)
Life expectancy (years)	0.084***	0.086***
	(0.024)	(0.023)
Strict lockdown measures	0.013**	0.013**
	(0.006)	(0.006)
Constant	5.054***	4.922***
	(1.262)	(1.242)
Observations	153	153
R2	0.563	0.565

Robust standard errors in parentheses. 2SLS: two-stage least squares; IV: instrumental variable; GDP: gross domestic product.

^*p < 0.1, **p < 0.05, ***p < 0.01.

It must be noted that while the 2SLS models show evidence for our hypothesis, none of the post-estimation tests indicate that an IV regression is required over regular OLS or that civil liberties are behaving as an endogenous variable in the present sample. To check, we performed the Hausman test (Davidson and MacKinnon, 1993; Hausman, 1978) for checking the endogeneity of the civil liberties variable and to ascertain whether it is necessary to use an IV, that is, whether a set of estimates obtained by least squares is consistent or not. For each model presented in Table A4, we reach the conclusion that our main explanatory variable is not endogenous and that OLS is consistent.

Besides these models, Table A6 presents Models C4–C9 which use several control variables introduced in Model A4 and Models B1–B3 (in Tables A2 and A3, respectively) as instruments. These are economic rights, physical liberties, political liberties, private liberties, polyarchy index, ⁹ and civic strength. All of these variables are technically “good” instruments, that is, they all satisfy the conditions for instruments discussed above, but their inclusion here is purely to model their effect as potential endogenous covariates, whereas previously in Model A4 and in Models B1–B3, they were used as control variables in regular OLS. It is interesting to note however that both for physical liberties and the polyarchy index, the small p-value from the Hausman test indicates that the effect of the endogenous regressor is being correctly modeled through an IV regression, rather than regular OLS. For all other models in Table A6 however, the Hausman test indicates that OLS is consistent.

Table A6.

Predicted Vaccines Administered Given Civil Liberties (V-Dem)—2SLS Models.

Variables	Model C4	Model C5	Model C6	Model C7	Model C8	Model C9
	IV: Economic rights	IV: Physical liberties	IV: Political liberties	IV: Private liberties	IV: Polyarchy index	IV: Civic strength
Civil liberties index (V-Dem)	–2.169**	–2.511***	–1.780***	–1.794***	–1.563***	–1.894***
	(1.058)	(0.485)	(0.492)	(0.486)	(0.518)	(0.527)
Total population (millions)	0.004***	0.004***	0.004***	0.004***	0.004***	0.004***
	(0.001)	(0.001)	(0.001)	(0.001)	(0.001)	(0.001)
Population density	–0.077	–0.086	–0.069	–0.069	–0.063	–0.071
	(0.142)	(0.145)	(0.137)	(0.137)	(0.134)	(0.138)
Early vaccine availability	0.015***	0.016***	0.015***	0.015***	0.015***	0.015***
	(0.004)	(0.004)	(0.004)	(0.004)	(0.004)	(0.004)
GDP per capita	–0.017**	–0.017**	–0.016**	–0.016**	–0.016**	–0.016**
	(0.007)	(0.007)	(0.007)	(0.007)	(0.007)	(0.007)
Life expectancy (years)	0.085***	0.084***	0.077***	0.078***	0.076***	0.078***
	(0.024)	(0.023)	(0.023)	(0.023)	(0.023)	(0.023)
Strict lockdown measures	0.016**	0.014**	0.014**	0.014**	0.014**	0.014**
	(0.006)	(0.006)	(0.006)	(0.006)	(0.006)	(0.006)
Constant	5.527***	5.711***	5.840***	5.838***	5.879***	5.820***
	(1.224)	(1.214)	(1.191)	(1.191)	(1.191)	(1.190)
Observations	152	153	153	153	153	153
R2	0.571	0.565	0.567	0.567	0.565	0.567

Robust standard errors in parentheses. 2SLS: two-stage least squares; IV: instrumental variable; GDP: gross domestic product.

^*p < 0.1, **p < 0.05, ***p < 0.01.