An open access publication of the American Academy of Arts & Sciences
Summer 2020

Democracy & Religion: Some Variations & Hard Questions

Kent Greenawalt
View PDF

The ideas sketched here concern the nonestablishment and free exercise norms expressed in the U.S. Constitution, their application to governmental institutions from legislatures to prisons and the military, the place of religion in the curricula of public schools, and the proper role of religious convictions in lawmaking. A major concern of the essay is the problem of achieving an appropriate balance between governmental neutrality toward religion, as required by the nonestablishment norm, and governmental accommodation of religious practices that would otherwise violate ordinary laws, as required by the free exercise norm. A recurring theme is the complexity of the issues and the variability of possible solutions given differences in the history and culture of democratic societies.

Kent Greenawalt, a Fellow of the American Academy since 1980, is University Professor at Columbia Law School and the Department of Philosophy at Columbia University. He has served as Deputy Solicitor General in the U.S. Department of Justice, Editor-in-Chief of the Columbia Law Review, and President of the American Society for Political and Legal Philosophy. He is the author of, most recently, Realms of Legal Interpretation: Core Elements and Critical Variations (2018), When Free Exercise and Nonestablishment Conflict (2017), and Exemptions: Necessary, Justified, or Misguided? (2016).

When one asks about the relation between democracy and religion, we have some answers that seem fairly obvious and others that do not. My basic claims are that there are important variations within democracies, that these may affect aspects of the proper treatment of religion, and that even within a modern, liberal democracy like that of the United States, we have some hard questions that lack simple answers. Certain answers to these questions do seem true across the board; others do not. The latter require a more particular focus.1

What does democracy in general entail? Perhaps we have no precise definition, but we can take democracy as a system of government in which all adult citizens have a right to vote. Assuming we are not talking about a minuscule political order in which ordinary people would directly determine prevailing law, citizens elect legislators, and the highest executive officials are either also subject to citizen votes or are chosen by legislatures. I think we can say that if it is a genuine democracy–that is, a country that recognizes the political rights of all citizens–it will allow people to choose whether or not to worship and essentially what form of worship to engage in. Of course, there can be some limitations if a form of worship is obviously harmful for those engaging in it or for others. 

What people see now as counting as a genuine democracy has developed over time. The original United States may have been conceived as a democracy, although racial slavery existed in many states and women rarely had a right to vote. Under contemporary conceptions, a political order with either of these factors might not be seen to be a genuine democracy. In respect to freedom of worship, one can imagine an exception if a particular religion and most of its followers are committed to violent acts against other citizens or overthrowing the basic political system.

The United States, like many other modern democratic states, has no established church. What does this nonestablishment norm imply regarding governmental favoring or endorsing some particular religion? Suppose members of a particular religion basically form a society. This was true for certain sections of the British Colony in America that were created by religious groups, some of which maintained influence in the early states. And to note something often forgotten, the original First Amendment instructed “Congress shall make no law respecting an establishment of religion.” This meant partly that Congress could not interfere with state establishments. If we consider those states to have been genuinely democratic, we would not see nonestablishment as required for democracies in general. To put this a bit differently, if the vast majority of people are members of a particular faith, government support for that faith does not seem at odds with basic principles of democracy, at least as long as nonadherents are both free to worship in a different way, or not to worship, and do not have their fundamental political rights, such as voting and running for office, denied because they do not adhere to the dominant religion. The fact that a particular religion is established might have little effect on the fundamental rights concerning a liberal democracy, although it can be in some tension with a maximum sense of religious freedom, having a tendency to yield some preferential treatment for those who are members of the established church. England, for example, for many years had both an established religion in the Church of England and been essentially a democracy, although it maintained its monarchy.

The free exercise principle is an important aspect of the general liberties afforded to citizens in modern liberal democracies. Exactly how special it is turns out to be a complex topic on which I will offer a few brief observations. One can ask about both how human perceptions figure and what our law now provides. For seriously religious persons, religious convictions and priorities can be central in their lives; they may care deeply about whether the government is interfering with these in any way.2 In a diverse society, even people who do not themselves possess such feelings do well to recognize them in others. It follows that the government should be taking these convictions and sentiments into account, at least if a significant percentage of the population possesses them. 

When one asks about existing law, matters are factually complex. In the case of Employment Division v. Smith in 1990, the Supreme Court decided that for most general laws not directed at religion, those with religious objections had no constitutional free exercise right to special treatment.3 This has led to questions of whether the free exercise clause has become redundant, swallowed up by freedom of speech and association.4 But a great deal remains in the special status of religious exercise. Here are five aspects. Employment Division v. Smith does not cover all religious practices. Churches and other religious practitioners retain the right to limit their clergy to men and to those who are not homosexual.5  Employment Division also indicates explicitly that legislators can make concessions to religious practices.6 We now have the Federal Religious Freedom Restoration Act, the Religious Land Use and Institutionalized Persons Act, and many similar state provisions that do just that.7 A subtler point concerns circumstances in which the government must treat nonreligious claims the same way it treats similar religious ones. Even here, if the religious claims help to provoke the basis for what equal treatment is required, free exercise remains important. 

Two further aspects of significance concern the relation between free exercise and nonestablishment. The most obvious is that both clauses are designed to promote government noninterference and freedom of religious belief and practice. Free exercise bears on how one should see the basic notion of nonestablishment. And sometimes the values of the two clauses do seem to come into conflict, as with prayers to begin legislative sessions. If this content is suitably neutral regarding the issues on which the legislative body must vote, those wanting the prayers may claim that allowing them is a free exercise right. Then the question is how far free exercise qualifies the coverage of nonestablishment, or is itself qualified by the conflict.8 In all five of these ways, free exercise remains a special liberty that has not become redundant.

When one considers how religions should be treated, one recognizes that given the diversity of populations, nonestablishment, at least in some form, is needed. In an important sense, the two basic concepts of nonestablishment and free exercise work together. If the government favors one religion over others, that will enhance the actual practices of that religion, while possibly interfering with what other religions do. Also important, favoritism is bound to encourage some people to get involved with that religion; this impairs the basic idea that people should choose freely whether to join a particular form of religion, without being pushed by the government. Some obvious examples are these. If the government promotes strong financial support for and endorses a particular religion, involvement in that religion may seem more attractive to someone not already dedicated to another faith. And if favoring includes teaching of a particular religion within public schools, that could incline students to believe that it is the true religion. Of course, teaching about a religion is not the same as teaching or implying its truth, but that distinction may not be simple for teachers or students. Nevertheless, omitting reference to religion in human history would convey a nonobjective, unrealistic account of all that has mattered.9

For the values underlying both free exercise and nonestablishment, the government should not favor some religion over others. This key to the basic idea of nonestablishment is strongly supported by the core value of free exercise, since people will feel more free about religion if they understand that the government will not favor or disfavor them based on their convictions or the groups to which they are joined. 

A modest exception to the no-disfavoring occurs if a religious group, or a segment of that group, supports violence against others. An existing controversial example of this concerns Islam. So long as a significantly large proportion of Muslims support violence against non-Muslims or Muslims of different denominations, it may be appropriate to do a more careful screening of Muslims–at least Muslim adult males who are not elderly–who seek entry into the United States. This cautionary policy differs from objectionable “racial profiling” in deciding who to admit to our country. 

Although free exercise and nonestablishment basically fit with one another, we do have, as mentioned, certain tensions between them. For some of these, it is not easy to say what are the right approaches within a liberal democracy. Perhaps the most obvious example is government engagement in religious practices and messages, at least if these do not promote some particular religious beliefs and groups over others. Is it appropriate for legislative sessions to begin with non-denominational prayers and for presidents to end formal addresses with an appreciation of God and a request for God’s help? Presidents, like ordinary citizens, are free to have their own religious convictions, but when they reference those convictions in an official speech, such as a yearly address to Congress, their comments amount to something beyond a simple personal expression. If most officials, as well as most citizens, have religious beliefs, free exercise can support their expressions for such occasions. For the most part, what the nonestablishment clause requires does not depend on the religious outlook of citizens and officials, but the extent to which free exercise concerns may qualify likely applications could depend on it. Of course, what is generally relevant is the content and context of the religious element in a public speech.

One way to view some of the apparent religious references is to see them as merely “ceremonial deism,” referring to the culture and history of the country. This was suggested by Justice Sandra Day O’Connor regarding the use of “under God” in the Pledge of Allegiance.10 Although this perception may be accurate as a representation of how a great many citizens regard the pledge, I am skeptical that these are the dominant understandings of either aliens who say the pledge before becoming citizens or students in public schools who are called upon to do so. I think many in these categories, as well as some others, will perceive the pledge as including an acknowledgment about the place of God, or at least references to an actual God, in the United States. 

The military and prisons, coercive institutions in which citizens lose many of their rights, are two special government domains. For both of these, the government should in some form provide religious exercise for those whose overall freedom is constrained. For military members stationed abroad in combat zones or aboard navy ships, the government may need to provide clerics themselves. For prisoners, it may manage by bringing clerics from outside to enter and provide services. The free exercise clause should here be taken to require, or at least authorize, reasonable efforts by the government to provide actual opportunities for typical exercise for those not free to go where they can worship as they choose. 

An interesting question connected to all this is whether nonreligious activities and convictions should be treated equally. A believer in absolute “neutrality” might think the right answer is “yes”; but I believe, as noted earlier, that this view is an oversimplification. If soldiers and prisoners are given time to pray or an opportunity to have their dietary needs satisfied, allowing others a time to reflect or satisfy their genuine convictions about acceptable food makes sense, but the provision of clerics is different. Despite some decline, religion remains very important in the lives of many Americans, and for most religions, the role of clerics is central to worship. One might imagine some nonreligious analogue, in which a leader is central to gatherings organized around basic values and experience, but actual examples are few or nonexistent. For something like actual military chaplains, we cannot expect a government accession to a nonreligious analogue. Here religion will appropriately be given special status. However, apart from special cases, government need not provide for clerics in all denominations in every military situation in which there is a need for a chaplain. In some cases, nondenominational chaplains might have the appropriate skills.

Prisons present harder questions still, such as whether religion should count about judgments concerning parole and, if so, what the role of clerics should be. Although this consideration could produce concern about dishonest affiliations, if it is true that religious involvement makes subsequent criminal acts less likely, parole boards should be able to take that involvement into account. They should, however, probably avoid making these determinations vary depending on precise statistics about particular denominations. An obvious exception to equal treatment concerns attachment to religions that themselves promote criminal acts. An interesting special example here concerns a religion that encourages polygamy. One might conclude that its members are more likely to commit what counts as a particular violation of law, but no more likely or even less likely to commit other crimes. 

Determining the proper role of clerics in parole board decisions is itself not simple. If religious practices and convictions are to be taken into account, clerics need to be able to testify about individual applicants for parole, although this constitutes religious personnel seriously affecting a certain kind of official determination. An important distinction here is between clerics contributing to information bearing on the problem of recidivism and their describing the specifically religious character of the prisoner, which might or might not have such a bearing.

In a number of states, clerics actually serve on parole boards. That may well be too great an involvement of clergy in government decisions, an involvement especially likely to encourage prisoners to get involved with the particular religions of those clerics. I believe this practice should be regarded as at odds with the values of nonestablishment and free exercise.

Public schools in democratic societies generate their own problems. As a matter of principle, schools should teach about the place of religion in human history, but not the truth or falsity of a particular faith. They should also not teach more general points, such as that a loving God genuinely exists, or that atheism is actually true. The distinction between teaching about religion and teaching a religious claim as true may be difficult for teachers to draw and for schoolchildren to perceive. This may lead some to conclude that it is desirable for those subjects simply to be omitted. But doing that would yield an incomplete account of what has mattered historically and would do so in a way that minimizes the actual place of religion. One could see this as a form of establishment of nonreligion conceived as presupposing atheism or at least as minimizing the actual place of religious views and practices in human life. This would implicitly encourage a kind of minimization of the importance of religion in students’ perspectives. Despite the complexities about distinguishing between an “objective” account of various beliefs and practices and an apparent implication of their likely truth and intrinsic value, to totally disregard the place of religion in human life and in our culture is much worse. Public schools properly include religious topics in what they cover, while teachers should at the same time try hard not to endorse any particular religious conviction.

When it comes to teaching subjects like evolution that are well established by science but conflict with the religious beliefs of those that take certain biblical passages about creation as literally true, should teachers delve into the competing version? I believe not, although teachers may tell students that some people have a strikingly conflicting religious view. If a topic is subject to rational analysis and does not depend on any particular religious outlook, it is appropriately taught for itself in public schools. This would be true of mathematics and science among others. Concerns that are raised by a subject, such as worries about climate change, are appropriately covered. When it comes to competing views that are based on entirely different premises about reality, such as a biblical account of when God created human beings, it is fine for a teacher to mention these, but not appropriate to explore them in analytical detail.11

Matters are more complicated when it comes to moral issues. Some moral questions are answerable on rational grounds. For example, parents should provide care for their children, and no one should kill another due to slight irritation. But we do not have simple rational answers about when abortions are not a matter of moral concern and what laws and public policies should thereby be instituted. Similar concerns exist for whether adoption by intergender couples should be preferred over gay couples because it is desirable for a child to have parents of both genders. I am assuming here that gay couples should have the right to marry they were accorded by the Supreme Court in Obergefell v. Hodges, and that this includes a right to adopt children.12 But it does not necessarily follow that parental genders are irrelevant to who might be favored for a specific adoption. When it comes to such issues, it may be best for teachers briefly to explain opposing views, including religious ones, but without getting into details. Something similar may be appropriate for some political issues, although Donald Trump’s presidency has led many to believe those involve certain moral concerns that have correct answers, such as whether political leaders should be basically honest.

What of religious convictions in lawmaking? Should laws and policies in our liberal democracy, or any democracy, be based exclusively on grounds that are not religious or anti-religious? If so, both legislators and citizens with relevant religious convictions about an issue should make every effort to disregard them in their political stances. How persuasive or realistic this position is turns out to be quite complicated. We need to distinguish among kinds of issues, between legislators versus ordinary citizens, and between actual reliance versus articulated bases for a stance, as well as how much courts should be involved in all this in constitutional and statutory interpretation. One may think it is healthy for judges constitutionally to protect the exercise of religion from ordinary laws that impair it. But with a few exceptions, the Supreme Court decided, in Employment Division v. Smith, that no such right exists. That leaves it in principle up to legislators to decide about the range of special treatment for religion. Ironically, legislators may decide to adopt a flexible standard that reinstates the range of judgments left to judges. This is what Congress did with the Religious Freedom Restoration Act and the Religious Land Use and Institutionalized Persons Act.13

Political decisions that actually favor one religion over others are not of general concern: under a basic nonestablishment principle, such laws and executive practices should not be adopted and citizens should not support them. More deserving of our concentration are essentially nonreligious questions, about which religious teachings may take particular positions. An obvious example that has existed throughout time is how much aid the government should give to people who are poor and disadvantaged. A different illustration concerns a modern controversy in the United States: is it or is it not acceptable to separate children from their parents in immigration detention? 

How simple is it for someone to distinguish the weight of nonreligious convictions from religious ones and to assess their degree of personal influence over time? Suppose that someone believes she should make a decision on non-religious grounds: the government should give substantial aid to the poor and disadvantaged because it is the just thing to do. Yet she also believes that a loving God strongly wishes people to provide that kind of help to others. For most such people, it would be very hard or nearly impossible to discount completely their religious beliefs in arriving at an attempted religious-independent position. 

Let us consider a more complicated example. Suppose someone was raised in a religion that takes a strong position on a particular issue, such as aid to the poor, abortion, or same-sex marriage. In his early years, he was a devoted follower and embraced these positions. As he grew older, his religious convictions disappeared, and he came to regard religious bases as irrelevant. But when our subject thinks about contentious issues in nonreligious terms, can he really discount the influence of his earlier views? In a straightforward way, his past perspective could lead him to believe that what he long accepted as sound positions on crucial social issues remain so. But we can also imagine a kind of reverse influence. If a person now believes the religion itself is foolish, he might conceivably discount the force of nonreligious reasons that support political positions the religious group has taken. 

When we put all this together we can see how hard it could be for many people to genuinely rely only on nonreligious thinking.14 This counts strongly against telling citizens that they should rely only on nonreligious reasons. More directly, assuming many in the country do have religious convictions, when it comes to issues that do not directly concern religion, such as public aid for the poor, I do not think it should be seen as wrong, nor as a kind of establishment, for them to rely self-consciously on the religious truth in which they believe. 

Given that legislators represent many kinds of citizens, the more powerful argument is that they should rely as far as possible on nonreligious reasons, reasons of a kind that can be shared by rational citizens independent of any religious convictions they may have.15 Legislators, however, like the rest of us, may have some difficulty figuring out how far religion has influenced their positions. Of course, one nonreligious factor for legislators is a need to satisfy the desires and convictions of those they represent. And that could well include giving a degree of weight to the religiously based positions of members of that group. 

When we turn to public articulations defending positions, as in open legislative sessions, political platforms, and campaign speeches, we can expect legislators to rely on nonreligious bases that are widely accepted. And in a liberal democracy, it makes good sense for advocating citizens to act similarly. If this is right, then the public arguments for positions may be more nonreligious than the complete balance of influential bases. 

Does religion merit special treatment in a liberal democracy? To approach this complicated and sometimes highly controversial question, I begin with three important generalizations. The first is that the appropriate answers may well not be the same for all types of liberal democracy. The beliefs and practices of most citizens will shift over time and will be quite different in different countries. The best answers for a given country depend partly on the cultures of the country at the time. Here I focus on the present-day United States. The second generalization is that even in the context of a single democratic country, we should not assume that there is one decisive answer to apply across the board. It may well be that religious convictions and practices will warrant special treatments in some parts of a country but not others. The third point is that, for this discourse, one should not rely directly on an individual religious conviction itself but rather reasons that have wide acceptance.

Among the issues of concern here are non-favoritism of some groups or individuals over others, concessions to beliefs and practices, and specific privileges for groups.

A core idea of nonestablishment that contributes to free exercise is that the government should not favor some particular religious bodies and organizations over others. Is this special for religion or does it have broader application? There is no simple answer. We can certainly understand that the Equal Protection Clause precludes favoring white groups or African-American groups, and the Free Speech Clause may similarly bar certain categorizations, but at least in our present culture, the constraint concerning treatment of religious groups is taken as more absolute. To this degree, the free exercise and nonestablishment clauses do exercise a greater constraint against differential treatment than do other constitutional provisions.

If the government does not favor a particular religious group over others, may it grant some privilege to religious groups that does not exist for nonreligious groups? Of course, concessions should not allow religious groups to directly harm others or to receive privileges that have nothing to do with their religious practices. But that leaves us with questions about religious practices that may be at odds with general legal requirements. Two notable examples here are hiring decisions and the consumption of substances.

Suppose a religion holds that God has instructed us that only men should be priests. Precluding women from the position is at odds with established law prohibiting gender discrimination. But to tell members of a religion that they must accept as clergy those they believe are ineligible would be a substantial restraint on their free exercise. Not surprisingly, the Supreme Court has accepted the practice by churches, including the Roman Catholic Church, of limiting clerics to men.16  

When it comes to controlled substances, what is generally forbidden by law may be part of a core practice of a religion. Two examples here involve communion wine and peyote as an ingredient for a religious gathering. Since no state now bans the sale and drinking of alcohol, the wine example is no longer a practical concern; but such bans did exist in the United States in the past. Some Christians believe God has instructed the ritual consumption of wine as a representation of the blood of Jesus, and many others think this use is at least symbolically valid. Given the small amount of wine taken by those participating in communion, which itself does not elicit typical concerns about the consumption of alcohol, an exception here was obviously favorable (even if a few consumers might have been encouraged by the experience to go home and drink more).

More difficult is the case of religious use of peyote, since the basic effects resulting from religious medicinal use are not so different from those generally regarded as harmful or dangerous enough to warrant a broad prohibition. Whether the use in a religious service is enough to warrant an exemption is a more nuanced question, with complicating factors of sovereignty and history, among many others.

Should individuals be excused from ordinary legal requirements, such as military conscription, because of religious convictions and, if so, when? Should nonreligious convictions get the same treatment? Obviously, if the legal requirement offers citizens protection from substantial harm, such as criminal laws prohibiting battery, no special exemption should go to religious individuals and groups. It may, however, be acceptable for religious groups to discipline and treat their own members in more subtly negative ways that could be subject to tort liability in other contexts. 

What if the privilege does not cause direct harm to anyone? Shall a religious objector be excused from jury duty or a military draft? The draft situation has invoked a specific statutory exception, prompting the key question of whether nonreligious claims should be treated similarly. Very briefly, given that a genuine pacifist will not engage in military efforts, a broader exemption clearly makes sense, especially if some form of alternate service is required. Congress sought to limit the exemption to religious claimants, but the Supreme Court responded by reading “religion” in the statute so broadly that it included those whose pacifist convictions were not religious in an ordinary sense.17 (Justice Harlan voted with the plurality to make a majority, but his basis was that restricting the privilege to religious convictions in this context was unconstitutional.)

Readers may disagree with some or many of my actual positions on these complex and controversial issues. But my overarching point is that the right relations of democracy and religion can depend on cultural settings; and even within a particular setting, like the present liberal democracy of the United States, we have a number of less-than-simple questions about what is called for. These lack complete and indisputable answers. Like much of our lives, what is right is both complex and disputable.

author’s note

This essay was written for a research seminar held in Rome in March 2019 and sponsored by the Australian Catholic University. For helpful questions and comments, I thank David McCabe, who was my respondent in the seminar. I benefited greatly from suggestions from Robert Audi.


  • 1What follows summarizes matters I have explained and positions I have taken in various writings. Among these are the following books, which can be a source for readers who want to explore some of the issues in more depth. Kent Greenawalt, Religions and the Constitution: Free Exercise and Fairness (Princeton, N.J.: Princeton University Press, 2006); Religion and the Constitution: Establishment and Fairness (Princeton, N.J.: Princeton University Press, 2008); Does God Belong in Public Schools? (Princeton, N.J.: Princeton University Press, 2005); Statutory Interpretation: Twenty Questions (New York: Foundation Press, 1999); Conflicts of Law and Morality (Oxford: Oxford University Press, 1987); Exemptions: Necessary, Justified, or Misguided? (Cambridge, Mass.: Harvard University Press, 2016); From the Bottom Up (Oxford: Oxford University Press, 2016); When Free Exercise and Nonestablishment Conflict (Cambridge, Mass.: Harvard University Press, 2017); and Religious Convictions and Political Choice (Oxford: Oxford University Press, 1988).
  • 2See Robert Audi, Democratic Authority and the Separation of Church and State (Oxford: Oxford University Press, 2013).
  • 3Employment Division v. Smith 494 U.S. 872 (1990).
  • 4I am presently working on a book about the nonredundancy of free exercise (it has not yet been submitted for actual publication).
  • 5See Hosanna-Tabor Evangelical Lutheran Church v. Equal Employment Opportunity Commission 132 S. Ct. 694 (2012).
  • 6Employment Division v. Smith.
  • 742 U.S.C. §2000bb-1 (1993); and 42 U.S.C. §2000cc–§2000cc-5 (2000).
  • 8Among my writings to address the subject is Greenawalt, When Free Exercise and Nonestablishment Conflict.
  • 9See Greenawalt, Does God Belong in Public Schools?
  • 10See her individual opinion (concurring in judgment) in Elk Grove Unified School District v. Newdow 542 U.S. 1, 36–37 (2004).
  • 11Although I have explored much of this in earlier work, Robert Audi’s essay “Religion and Politics of Science: Can Evolutionary Biology Be Religious Neutral?” Philosophy and Social Criticism 35 (1–2) (2009): 23–50, provides a superbly detailed account of various considerations and the weight they should carry that far exceeds what my own work contains.
  • 12Obergefell v. Hodges 576 U.S. ___ (2015).
  • 13See 42 U.S.C. §2000bb-1 (1993); and 42 U.S.C. §2000cc–2000cc-5 (2000).
  • 14See Greenawalt, Religious Convictions and Political Choice.
  • 15Ibid.
  • 16See generally Hosanna-Tabor Evangelical Lutheran Church v. Equal Employment Opportunity Commission 132 S. Ct. 694 (2012).
  • 17Welsh v. United Sates 398 U.S. 333 (1970).