A Borderline Case: The Establishment Clause Implications of Religious Questioning by Government Officials

Introduction

Imagine you go to Toronto for a weekend trip with your family. While driving home to Detroit, a border agent pulls you aside, brings you into an isolated room, and asks you, seemingly out of nowhere, “How many times a day do you pray?”

Now imagine you are an Arab Muslim man named Ali Suleiman Ali.1 When you are asked this question, do you feel the same way you do when the government requests that you report your religious affiliation on a census form? Or do you feel insulted, believing that this agent, operating in the shadow of September 11 and statements by President Donald Trump like “Islam hates us” and “we’re having problems with the Muslims,”2 has equated the religious practices he assumes you hold with some sort of threat?

The implications of this type of religious questioning by government officials have been considered in two recent district-court cases: Cherri v Mueller3 (brought by Ali and other Muslim Americans stopped at the Canadian border)4 and Isakhanova v Muniz5 (brought by a Muslim American who was questioned by prison guards while visiting her inmate son).6 After facing a line of questions in this vein, the plaintiffs in each case argued that questions from government officials about plaintiffs’ religious practices violated the Establishment Clause of the US Constitution.7 Prior legal cases provided little guidance about whether the plaintiffs had a cognizable Establishment Clause claim. Although the Establishment Clause prohibits both government endorsement and disapproval of religion,8 Establishment Clause jurisprudence has historically focused on cases of endorsement. It has matured through cases centered on nativity scenes, statues of the Ten Commandments, and school prayer. And even this canonical Establishment Clause case law is infamously muddled. Neither the Cherri nor the Isakhanova court grappled with the complicated state of Establishment Clause jurisprudence. They did not reason through which test they ought to apply to religious-questioning policies or take the chance to provide guidance to other courts on how to analyze such a policy. Instead, they both provided cursory holdings on their respective plaintiffs’ Establishment Clause claims.

Courts should not treat religious-questioning policies so mechanically. These policies raise novel questions about the proper scope and application of the Establishment Clause that have recently gained urgency. Reports that border agents questioned travelers about their religious practices during the implementation of Trump’s “travel ban” in January 2017 trigger the same concerns as the practices considered in Cherri and Isakhanova.9 Senators Dianne Feinstein and Dick Durbin both raised similar red flags when they questioned then–judicial nominee Amy Coney Barrett about her Catholic beliefs during her September 2017 confirmation hearing before the Senate Judiciary Committee.10 This Comment therefore uses the Cherri and Isakhanova decisions to start a broader discussion about the validity of religious questioning by government officials. Part I describes the tests used by the Supreme Court when analyzing Establishment Clause cases. It then explores a recent line of cases that provides guidance on how courts should analyze government policies that disapprove of, rather than endorse, religious beliefs. Part II introduces the legal question: whether religious questioning by government officials violates the Establishment Clause. Part III determines that this legal question is properly analyzed under a novel “Lemon-Larson” test. This approach combines two of the Supreme Court’s Establishment Clause tests to create a modified analysis that is tailored to consider the countervailing government interests in religious-questioning cases.11 Part III applies the Lemon-Larson framework to religious-questioning scenarios. It concludes that the questioning policies in both Cherri and Isakhanova would be unconstitutional under the Lemon-Larson test, as would most religious questioning. However, the Lemon-Larson test accommodates the government interests that truly require entanglement with religion in a way that current Establishment Clause jurisprudence does not, adapting the Establishment Clause to a new manifestation of disapproval-of-religion cases.

I. The Establishment Clause, as Interpreted by the Courts

In contrast to the large volume of academic literature that considers when the Establishment Clause has been violated, the clause itself is to the point. The Establishment Clause reads, “Congress shall make no law respecting an establishment of religion.”12 Part I.A examines how the Supreme Court has interpreted those ten words, focusing on the two tests most relevant to the religious-questioning cases: the Lemon test and the Larson test. Part I.B then presents a recent line of cases that demonstrate that courts analyze disapproval-of-religion cases under the same Establishment Clause tests as in the more common endorsement cases. Part I.C concludes by examining how other constitutional and statutory claims that could apply to religious-questioning cases interact with the Establishment Clause.

A. The Background of the Establishment Clause and Its Tests

The foundational statement of what the Establishment Clause prohibits comes from Everson v Board of Education of Ewing Township,13 a 1947 Supreme Court decision:

Neither a state nor the Federal Government can set up a church. Neither can pass laws which aid one religion . . . or prefer one religion over another. Neither can force nor influence a person to go to or to remain away from church against his will or force him to profess a belief or disbelief in any religion. No person can be punished for entertaining or professing religious beliefs or disbeliefs.14

This passage added meat to the bare bones of the Establishment Clause’s text. It also left open many unanswered questions. Everson did not, for example, say anything about whether other kinds of state actions could violate the Establishment Clause. Nor did it specify the test that courts should use to identify when the Establishment Clause had been violated.

The threshold question of what additional acts violate the Establishment Clause, as well as the contours of the Everson prohibitions, have been the subject of voluminous case law in the decades since Everson. This case law can readily answer some questions about Establishment Clause cases. For example, a government policy is judged primarily on its objective effect rather than the subjective intent behind it.15 The Supreme Court has also explicitly stated that a government policy can violate the Establishment Clause even if the policy does not directly compel the exercise of religion.16 There is little remaining debate on this type of elementary Establishment Clause question.

Substantial debate remains, however, on the best way to determine whether a specific government policy violates the Establishment Clause. This Section provides an overview of the various tests adopted by the Court in analyzing potential Establishment Clause violations. It does so with the caveat that the Supreme Court treats none of these tests as talismanic. The Court has been inconsistent about the tests used to analyze Establishment Clause cases. It has characterized the tests it does employ merely as “helpful signposts.”17 Cases apply multiple tests, with little discussion on whether any one is dispositive.18 Some justices have strongly criticized this “unintelligib[le]” approach to the interpretation of such an important constitutional protection.19 Nevertheless, the Court’s Lemon and Larson tests both illuminate how courts should treat religious-questioning policies and are therefore detailed in this Section.

1. The Lemon test.

The Lemon test is generally considered the guiding, if flickering, light in Establishment Clause jurisprudence.20 The Court has applied the Lemon test in a broad array of traditional Establishment Clause cases, including those considering whether school activity or overt government symbolism violates the Establishment Clause.21 Taking its name from Lemon v Kurtzman,22 the Lemon test was initially described as a three-factor test.23 Although the Court eventually collapsed the third prong of the analysis into the second,24 the original test included (1) whether the government policy has a legitimate secular purpose;25 (2) whether the policy’s primary effect is one of advancing or inhibiting religion; and (3) whether the policy creates excessive government entanglement with religion.26 A government policy violates the Establishment Clause if it “fails to satisfy any of these prongs.”27

The first prong, or the “purpose test,” concerns the “actual purpose” of the policy.28 It seeks to expose pretextual purposes that obfuscate a policy’s actual religious purpose.29 The object of this prong is to “prevent[ ] [the] government from abandoning neutrality and acting with the intent of promoting a particular point of view in religious matters.”30 In general, “no consideration of the second or third criteria is necessary if a statute does not have a clearly secular purpose.”31 The Court often finds policies to have invalid, religious purposes, even when the government has provided what purports to be a secular purpose. For example, the Court concluded that a policy requiring the posting of the Ten Commandments in a school was religious, even though the Government argued that it was intended to promote “the fundamental legal code of Western Civilization.”32 In another case, the Court rejected as insincere a school’s claim that it required the teaching of creationism alongside evolution to “protect academic freedom.”33 It found, in contrast, that a voucher program meant to provide assistance to poor children in a failing school district furthered a valid secular purpose even though students could use the vouchers at religious schools.34

The second prong, the “effects test,” or “endorsement test,” is often the crux of the analysis. It considers, from an objective viewpoint, “whether the government action has the purpose or effect” of either endorsing or disapproving of religion.35 The Court has always understood endorsement and disapproval of religion as opposite sides of the same coin. It prohibits both because they reach beyond the governmental powers as “circumscribed by the Constitution.”36 “[C]rucial” to this analysis is that “a government practice not have the effect of communicating a message of government endorsement or disapproval of religion.”37 A governmental policy does not need to succeed in promoting or degrading a religion to violate the Establishment Clause. What matters is whether a reasonable observer would understand the policy to be motivated by a desire to endorse or disapprove of religion.38 Courts take a fact-intensive and context-sensitive approach to this analysis.

To invalidate a policy under the effects test, courts must find that endorsement or disapproval of religion is the primary effect of a policy. For example, the Supreme Court concluded that a city’s public display of a nativity scene did not violate the Establishment Clause because the promotion of religion was not its primary effect. Instead, the Court found that the primary effect of the crèche was to promote a “significant historical religious event . . . long recognized as a National Holiday.”39 The Court also noted that:

[T]o conclude that the primary effect of including the crèche is to advance religion in violation of the Establishment Clause would require that we view it as more beneficial to and more an endorsement of religion, for example, than expenditure of large sums of public money for textbooks supplied throughout the country to students attending church-sponsored schools.40

This holding shows the Court’s willingness to look beyond the overtly religious nature of a nativity scene in its Establishment Clause analysis. It simultaneously hints at the Court’s desire to avoid holdings that will invalidate a significant number of other government policies. This is not to say that the Court is never willing to find endorsement. Examples of government actions that courts have struck down because of their unconstitutional effects include pregame prayers at football games41 and an “editorialized” version of the Ten Commandments in an Ohio state court that “exhort[ed] a return to ‘moral absolutes.’”42

Both the purpose and the effects prongs of the Lemon test are objective, meaning they are based on how a reasonable observer would understand the policies rather than on the government’s subjective intent.43 The concern is that a government policy of religious endorsement or disapproval will impermissibly “send[ ] a message to nonadherents that they are outsiders, not full members of the political community.”44 This is why the core of the analysis is not what the government intends the policy to do but what reasonable observers would understand the policy to do.45 Circuit courts have understood this guidance as militating against policies that treat certain religious participants as “second-class citizens,”46 “leav[e] members of minority faiths unwilling participants” in public activities,47 or require average citizens to “burrow into a difficult-to-access legislative record for evidence to assure themselves that the government is not endorsing a religious view.”48

The Court has provided little guidance on exactly what this “reasonable observer” knows, with justices acknowledging (almost apologetically) that there is considerable judicial discretion to determine exactly what a reasonable observer would know in any situation.49 When deciding how the reasonable observer would react to a government policy, lower courts assume that the reasonable observer has knowledge of the history and context of the community in question. This is not “the everyday casual gawker.”50 Circuit-court cases suggest that courts generally assume the reasonable observer adapts to the times and has fairly extensive familiarity with the precise community in question (down to the local county history).51

The third prong of the Lemon test, assessing government entanglement with religion, has largely been subsumed into the analysis of the second prong. In the 1997 case Agostini v Felton,52 the Supreme Court explicitly folded the third prong into the second.53 It has since argued that “[t]his made sense because both inquiries rely on the same evidence and the degree of entanglement has implications for whether a statute advances or inhibits religion.”54 Therefore, courts now consider ongoing and excessive government entanglement with religion as evidence that a policy fails the effects test.55

As with Establishment Clause jurisprudence generally, courts do not have bright-line rules to determine when entanglement violates the Establishment Clause. Courts acknowledge that the line between religion and the government resembles a “blurred, indistinct, and variable barrier,” rather than a wall.56 “Fire inspections [and] building and zoning regulations . . . are examples of necessary and permissible contacts” between religion and the government.57 “Entanglement” becomes unconstitutional when these “contacts” morph into unnecessary “intrusion.”58 Impermissible entanglement occurred in Lemon. The Court considered statutes in Pennsylvania and Rhode Island that used state money to fund religious elementary and middle schools, provided those funds supported secular education within those schools akin to that offered in public schools.59 The Court held that “comprehensive, discriminating, and continuing state surveillance” would have been required to ensure that secular teachers in parochial schools abided by the requirements for teachers in public schools.60 The Lemon decision shows that the entanglement inquiry, now used only as supporting evidence in the effects test, requires courts to interrogate the “character,” “nature,” and “resulting relationship” of any government interaction with religion in order to gauge whether the interaction crosses the line into unconstitutional intrusion.61

2. The Larson test.

Although the Lemon test remains the primary test that courts use to determine when state actions violate the Establishment Clause, the Court has developed another Establishment Clause test for cases involving government actions that discriminate among religions rather than endorse or disparage religion as a whole. This test was first articulated in Larson v Valente.62

In Larson, the Supreme Court considered government-imposed reporting and registration requirements that applied to only a subset of religions (those that solicited more than fifty percent of their funds from nonmembers).63 Larson held that strict scrutiny should be applied in cases in which a government policy suggests a “denominational preference” between religions.64 When reviewing such a policy, the Lemon principles can offer helpful guidance. However, Lemon is not the proper test to use to analyze the policy.65

To survive strict scrutiny under the Larson test, as in other contexts, the government must show that its policy furthers a “compelling governmental interest” and is “closely fitted to further that interest.”66 The government has a compelling interest only if it can prove that its policy addresses an “actual concrete problem”: “For an interest to be sufficiently compelling to justify a law that discriminates among religions, the interest must address an identified problem that the discrimination seeks to remedy.”67 After this interest is identified, the government must demonstrate that it has “closely fitted” its policy “to further that interest.”68 For example, in Awad v Ziriax,69 a recent Tenth Circuit case invalidating a proposed Oklahoma state constitutional ban on the invocation of Sharia law in court, the court reasoned that “[e]ven if the state could identify and support a reason to single out and restrict Sharia law in its courts, the amendment’s complete ban of Sharia law is hardly an exercise of narrow tailoring.”70

Perhaps surprisingly, given the frequency with which strict scrutiny is used in other constitutional contexts, courts rarely use the Larson test.71 In fact, before the Awad court analyzed the ban on Sharia law, it discussed whether the infrequent use of Larson had in fact rendered it bad law. It concluded that rare application of a doctrine did not invalidate a Supreme Court precedent that had never been explicitly overturned.72 There are several possible explanations for courts’ rare reliance on Larson. For one, the unpredictable application of Larson may simply represent a symptom of the general inconsistency in Establishment Clause jurisprudence. Another potential reason is that Larson is in fact obsolete. If so, the Tenth Circuit incorrectly concluded that Larson applied in Awad. The Supreme Court’s own actions, however, suggest that Larson remains a viable, if secondary, Establishment Clause test.73 While citations to Larson are rare, they exist: the Supreme Court has cited Larson in just under two dozen cases since its publication.74 Another explanation is that courts so rarely employ Larson because they are more likely to scrutinize government preferences among religions when the government is disapproving of a religion. They may be less wary of policies that endorse individual religions or less likely to view those as discriminating among religions. If that were the case, courts might be more likely to apply Larson in disapproval-of-religion cases. Both Larson and Awad can be distinguished from cases like Lynch based on the fact that they involved government practices that were critical—rather than approving—of a religion. Disapproval-of-religion cases are rarer than endorsement cases,75 which could explain why courts rarely apply Larson. However, as discussed in Part I.B, even when confronted with government actions disapproving of beliefs or practices associated with particular religious traditions, courts frequently apply the Lemon test rather than the Larson test.76

The most promising explanation for why courts rarely rely on Larson is probably that the Supreme Court’s narrow understanding of what it means for a policy to facially discriminate among religions limits the situations in which Larson may apply. In fact, many cases analyzed under the Lemon doctrine concern a practice that implicates a denominational preference. Consider the crèche case discussed earlier, Lynch v Donnelly.77 It is hard to argue that this case involved government endorsement of religion generally. If the government’s actions constituted an endorsement of anything, it was of Christianity. Nevertheless, the Court did not apply Larson on the grounds that the Larson test applies only when a policy is “patently discriminat[ing]” among religions.78 This holding suggests that a policy concerning “generalized Christianity” or theism does not count as facial discrimination.79 Thus, a policy that preferences Christianity as opposed to other religions is generally insufficient to trigger Larson strict scrutiny. A showing of a more specific denominational target is required. The Tenth Circuit in Awad based its unusual decision to apply the Larson strict scrutiny test on the fact that the law in question invalidated only Sharia law, as opposed to all religious laws. The court took this to mean that the law was truly discriminating among religions.80

3. Other tests.

Lemon and Larson are hardly the only tests courts use in Establishment Clause cases. However, other tests—including the endorsement test, neutrality principle, and coercion test—are less useful as analytic tools in the religious-questioning cases considered in this Comment.81 The endorsement test is itself just an elaboration of the Lemon test’s second prong, first articulated by Justice Sandra Day O’Connor in her influential concurrence in Lynch.82 The neutrality principle—asking whether a government practice is neutral toward religion—is embedded within the Establishment Clause and the other Establishment Clause tests, including the Lemon test. It is not its own independent test.83 Finally, the coercion test, which questions whether a government practice is coercive to individuals, is not the law84 —a majority of the Supreme Court has not embraced the coercion test. Instead, concurring or dissenting opinions occasionally recommend its adoption.85 Justice Clarence Thomas, for example, has used the coercion test in his arguments that the scope of the Establishment Clause should be narrowed.86

A recent Supreme Court trend, evident in Van Orden v Perry,87 is to ground Establishment Clause analyses in historical practice.88 In Van Orden, the Court explicitly stated that the Lemon test was not helpful when dealing with “passive monument[s],” such as a statue of the Ten Commandments on the grounds of the Texas state capitol.89 Similar to the neutrality principle, historical practice is not a stand-alone test. The Court instead simply decided to give significant deference to past practice: “[I]t is not necessary to define the precise boundary of the Establishment Clause where history shows that the specific practice is permitted . . . and has withstood the critical scrutiny of time and political change.”90 The Supreme Court, when it relies on evidence of longstanding practice, uses history as a proxy for the constitutionality of a government policy. The historical-practice trend has identified a kind of evidence particularly relevant in Establishment Clause cases but has not replaced the principles encoded in the Lemon test.91

B. Disapproving of, Rather Than Endorsing, Religion

Excepting Larson and Awad, the Establishment Clause violations discussed so far have involved government endorsement of religion. This is because the vast majority of Establishment Clause jurisprudence involves endorsement. There are only a few cases that apply the Establishment Clause to disapproval of a religion.92 However, government policies that disapprove of or express hostility toward a religion undeniably violate the Establishment Clause.93 In Everson, the Supreme Court clearly stated that the Establishment Clause not only proscribes endorsement but also “punish[ment]” of religious beliefs.94 This Comment focuses on the disapproval side of the Establishment Clause coin.

While the Supreme Court has offered little guidance in this area, recent circuit-court decisions provide a framework for understanding what type of government policy violates the Establishment Clause by disapproving of religion. They also demonstrate that courts approach such fact patterns in the same fact-specific, and occasionally doctrinally inconsistent, way as endorsement cases. Several benchmark cases applying the Lemon framework come from the Ninth Circuit. In Vasquez v Los Angeles County,95 the court held that a county government removing a cross from a county seal was appropriate, as it was not “motivated by hostility toward Christianity”—it was, in fact, motivated by the legitimate secular purpose of avoiding an Establishment Clause lawsuit.96 Similarly, in Vernon v City of Los Angeles,97 a government investigation into an assistant police chief’s religious practices did not violate the Establishment Clause.98 The court applied a Lemon analysis without mentioning Larson, even though the investigation focused on Robert Vernon’s involvement with a specific sect, the Grace Community Church.99 The court held that the investigation was appropriately motivated by Vernon’s erratic job performance.100 He had been quoted as depicting the police as “ministers of God,” ordering that no one was to be arrested at pro-life demonstrations, and pressuring police officers to attend church services.101 The investigation focused narrowly on whether his religious beliefs were impermissibly affecting his job duties, the questioning did not represent an ongoing policy, and the officers investigating him explicitly told him they were not telling him what his religious beliefs should be.102

The Tenth Circuit’s Awad decision, discussed in Part I.A.2, also explored disapproval of religion. The court decided that the Larson test was the best approach when scrutinizing a proposed state constitutional amendment that would outlaw Sharia law in Oklahoma courts.103 Unlike the Ninth Circuit cases, therefore, the Tenth Circuit eschewed the Lemon test in this disapproval-of-religion case. Under this alternative lens, the court found that the policy did not address a “concrete problem” or support a compelling interest and likely violated the Establishment Clause: “[T]o sacrifice First Amendment protections for so speculative a gain is not warranted.”104

A line of cases that sheds light on the specific concerns raised by religious-questioning policies involves policies of disparaging remarks toward a religion. In the 1983 case Marsh v Chambers,105 the Court noted that such policies can violate the Establishment Clause.106 A district court in California offered examples of statements that fall on both sides of the “disparaging” line in C.F. v Capistrano Unified School District.107 It found that a teacher’s comment that creationism was “religious, superstitious nonsense” violated the Establishment Clause by disapproving of a religion.108 Other comments the teacher made for educational purposes and not to demonstrate his own beliefs (such as, “What was it that Mark Twain said? ‘Religion was invented when the first con man met the first fool.’”) did not rise to the level of a violation.109 The court did not buy the argument that the “superstitious” statement was made for the secular purpose of education, concluding instead it was “unequivocal[ly]” driven by the belief that such religious beliefs actually were nonsense.110 The teacher could have easily taught the lesson without “disparaging those views.”111

Town of Greece v Galloway,112 which concerned a prayer program at monthly town board meetings, offered an additional standard for which statements rise to the level of an Establishment Clause violation. Town of Greece held that disparaging but one-off comments that are part of a larger, nondisparaging whole likely do not rise to the level of a prohibited government act.113 This included the comment by a visiting minister at one of the town meetings that an “ignorant” religious minority did not respect the history of the country.114 The Capistrano decisions (at the district and appellate levels) both acknowledged this requirement, suggesting that the “religious, superstitious nonsense” statement impermissibly signaled government disapproval of a religion, but that the plaintiff also had to demonstrate “ongoing entanglement.”115 Though courts have offered little guidance on the precise character of remarks required to meet this threshold of religious disapproval, a policy of disparaging remarks can clearly rise to the level of an Establishment Clause violation.

This Comment builds on this line of cases to explore a new type of disapproval-of-religion policy—religious questioning—in order to see how the Establishment Clause can adapt to and inform a new type of government entanglement with religion.

C. How the Establishment Clause Interacts with Other Potential Claims

It is important to note that this Comment’s focus on the Establishment Clause is not intended to suggest that religious questioning does not implicate other constitutional protections. Relevant provisions likely include the Free Exercise Clause116 and the Equal Protection Clause.117 Plaintiffs could also bring statutory claims under the Religious Freedom Restoration Act of 1993118 (RFRA), among other acts.

The Establishment Clause is not merely redundant, though; it provides unique guidance for what types of government entanglement with religion are permissible. The jurisprudence surrounding the Establishment Clause reflects the Court’s attempts to balance the “unbroken history of official acknowledgment” of the role of religion in “American life” with the mandated separation of church and state.119 It provides a framework to navigate the complicated role of religion in American society.120 It also analyzes these questions from a different angle than its sister clause in the Constitution, the Free Exercise Clause.121 Finally, it does so more thoroughly than the Free Exercise Clause—there is significantly more case law and Court guidance on the Establishment Clause.122

Besides offering helpful guidance for religious-questioning cases, the Establishment Clause offers litigants different paths to success in court. Policies analyzed under the Equal Protection Clause,123 the Free Exercise Clause,124 and RFRA125 generally all face strict scrutiny. Some policies that fail this test would nonetheless survive under the various Establishment Clause tests described in Part I, and vice versa. There are also instances when a plaintiff may have standing to bring an Establishment Clause claim but not a Free Exercise or RFRA claim. To bring a Free Exercise claim, a plaintiff must allege a “substantial burden” on his religious practices.126 Courts have held that increased financial costs of practicing a religion127 and a lack of access to halal food in prison128 do not constitute substantial burdens. RFRA has similar practical limitations129 and applies only to federal officials.130 Given these precedents, temporary religious questioning like that in Cherri and Isakhanova that does not alter a plaintiff’s religious practices will likely not reach the level of a substantial burden. The Establishment Clause analysis does not demand an individual demonstrate such a burden.131 As long as standard standing requirements132 are met, a plaintiff can bring an Establishment Clause claim even if the questioning does not alter his religious practices before or after the religious questioning occurs.

* * *

Establishment Clause jurisprudence does not provide clear-cut rules for testing the constitutionality of government policies that invoke religion. Recent court practice suggests that the Lemon test remains the prevailing Establishment Clause test, but that other tests, such as the Larson test, are appropriate in certain factual scenarios. This unsettled legal landscape is even less developed in disapproval-of-religion cases. As the cases discussed in Part I.B demonstrate, courts have generally analyzed disparaging remarks toward religion under traditional Establishment Clause frameworks. Importantly, they have demonstrated a willingness to find that such policies violate the Establishment Clause.

II. Religious Questioning by Government Officials

Though the Establishment Clause provides a helpful lens through which to analyze disapproval-of-religion policies, one subset of these policies that has still received almost no attention from courts is religious questioning. This type of questioning raises thornier Establishment Clause questions than the kind of questioning at stake in cases like Vernon. The questioning of Vernon was prompted by and tailored to his job performance.133 In contrast, the government questioning in the two district-court cases analyzed in this Part, Cherri and Isakhanova, was prompted by nothing other than the plaintiffs’ apparent religion. This Part concludes that religious questioning of this kind raises novel, underexplored Establishment Clause issues that deserve more attention than courts have afforded them.

A. Relevant District-Court Cases

Of the two cases, Cherri presents the more challenging fact pattern. In Cherri, Customs and Border Patrol (CBP) and Federal Bureau of Investigation (FBI) agents (collectively, “border agents”) questioned the plaintiffs—Muslim American citizens—about their religion when crossing the US-Canada border.134 The questioning was prompted “solely” by the apparent religion of the plaintiffs and a perceived connection between this religion and “terrorist activities.”135 Questions included “Which mosque do you go to?”; “How many times a day do you pray?”; “Who is your religious leader?”; and “Do you perform your morning prayer at the mosque?”136 In their complaint, the plaintiffs alleged that the defendants “implemented a policy . . . which include[d] asking Muslim American travelers a substantially similar set of questions about their Islamic beliefs and practices.”137 The plaintiffs alleged similar practices at no fewer than seven other border entry points, with questions including “When did you become a Muslim?”; “Are there any extremists or terrorists at the mosque?”; and “Do you know any terrorists?”138

The Cherri court found the plaintiffs did not have an Establishment Clause claim and granted the defendants’ motion to dismiss the plaintiffs’ First Amendment and RFRA claims.139 The court reached this conclusion because the plaintiffs’ allegations did not state a claim under what it viewed as the three prongs of the Lemon test.140 First, the plaintiffs did not, the court concluded, allege the policy had a “religious objective.” Second, the court held that a reasonable person would not conclude that the religious questioning they experienced constituted an endorsement of religion. Finally, the plaintiffs did not state facts establishing excessive government entanglement.141 The court argued that an Equal Protection Clause claim was “better suited” to the facts.142

A few years later, the Isakhanova court also addressed whether religious questioning by a government official violated the Establishment Clause. In Isakhanova, the Muslim mother of a state prison inmate faced religious questioning and disparaging remarks about Islam after prison guards detained her on suspicion of sneaking tobacco to her son.143 She was asked questions like “What kind of Muslim are you—Sunni or Shia?”; “Do you pray five times a day?”; and “What mosque do you go to?” She was told “All Muslims are terrorists” and “America is no place for Muslims.”144

The court denied the defendants’ motion to dismiss.145 It found that the “derogatory comments” made by the prison guards to the plaintiff violated the Lemon test.146 Because the prison guards were allegedly searching the plaintiff for tobacco, there was no clear “secular purpose” for their religious questioning and “statements such as ‘All Muslims are terrorists’ would be perceived by any reasonable Muslim as ‘disapproval of their individual religious choices.’”147 Finally, “under the third prong of Lemon, statements such as, ‘America is no place for Muslims,’ foster excessive governmental entanglement with religion, because they run afoul of the prohibition against ‘making adherence to a religion relevant in any way to a person’s standing in the political community.’”148

B. Limitations of the District-Court Decisions

Neither Cherri nor Isakhanova provided a satisfying analysis of the Establishment Clause issues raised by religious questioning of the type the plaintiffs in those cases experienced. For one thing, both opinions appear to oversimplify Supreme Court precedent: each applied the Lemon test without discussing whether it was the proper Establishment Clause test for the situation and without acknowledging developments in Lemon jurisprudence, such as the effective elimination of the third prong.149

Furthermore, each case discussed the Lemon test only briefly and the cases came to opposite conclusions through largely conclusory statements. The cases are in some ways distinguishable on their facts. The prison guard in Isakhanova explicitly insulted Islam. It is hard to think of any motivation for his comments other than animus toward the religion. The questions asked by the border agents in Cherri cannot be as easily dismissed as extraneous. Those factual differences should not obscure the fact that the two courts also applied the law inconsistently. While the Isakhanova court relied on Cherri in its decision to apply Lemon, it did not address these inconsistencies.150

Take each court’s discussion of the first Lemon prong (the purpose test). In Cherri, the court stated that the government’s “claimed association between [the plaintiffs’] Islamic beliefs and terrorist activities” did not demonstrate a religious objective.151 The Isakhanova court inferred a religious objective based on the government’s failure to demonstrate a legitimate association between the plaintiff’s Islamic beliefs and criminal activities (sneaking in tobacco).152 At an abstract level, the same thing happened in each instance: a government official asked questions about religious practices in the context of an investigation into whether the plaintiff had committed a crime. The Cherri court saw a secular purpose for this questioning; the Isakhanova court did not. The legal grounding for those different findings is unclear. Because of the different contexts and considerations involved in each case, the opposite findings may have actually been completely justified. As a matter of common sense, questions about Islam are more probative when considering potential terrorist activities than when considering smuggled tobacco.153 The Isakhanova court did not, however, explicitly ground its holding in any type of showing that the interrogation in Cherri was more likely to uncover crimes than that in Isakhanova. The Cherri court did not discuss whether the government’s purpose was pretextual, and thus invalid, as the Isakhanova court did.154 The Cherri court did not suggest that the questions asked were likely to actually root out terrorist activity, implying that an initial association between a religious belief and a crime allows for indiscriminate questioning about that religious belief. It did not distinguish between questions asked or consider whether some expressed unconstitutional animus toward a religion, even if others did not. It did not explore whether a reasonable observer would understand specific questions like “Do you consider yourself a religious person?” and “Are you part of any Islamic tribes?” to have a religious objective.155

These same inconsistencies apply to each court’s conclusions about whether a reasonable person would find the questioning to express disapproval of religion under the effects test.156 The effects prong of the Lemon test looks at the message communicated by a policy, not the factors motivating the policy.157 Despite this, the Cherri court did not ask whether a reasonable observer would think that questions like “Are there any extremists or terrorists at [your] mosque?” would make the plaintiffs feel as though they were not complete members of the “political community.”158 It did not explore the salience of a government policy explicitly connecting the practice of a religion with terrorist activities. In fact, it concluded that these questions merely stopped the plaintiffs from “cross[ing] the border in a timely fashion” and did not “endors[e]” a religion—without mentioning that the Establishment Clause also prohibits disapproval of a religion.159 In its decision, the Isakhanova court did not criticize Cherri. It also did not explain why a reasonable observer would think a government official connecting Islam with tobacco smuggling disapproves of a religion, even though connecting it with terrorism (certainly a worse crime) does not. It did not explain why comments like “All Muslims are terrorists” violate the Establishment Clause even though leading questions from Cherri like “Do you know any terrorists?” do not. Neither court acknowledged the likely fact that they were making their assumptions based on common sense. Both courts may indeed have been satisfied that all of these concerns were misplaced, but this lack of discussion leaves unclear which differences supported opposite conclusions in Cherri and Isakhanova, as well as whether the courts were correct to rely on those differences.

Using these cases’ fact patterns as a starting point, this Comment more carefully explores the Establishment Clause concerns triggered by a government policy of asking questions regarding religious practices, and the difficult question of what test should be employed in such cases.

III. Applying the Establishment Clause to Religious-Questioning Policies Using the Lemon-Larson Test

This Part picks up where the Cherri and Isakhanova courts left off, analyzing when religious-questioning policies violate the Establishment Clause. It argues that neither the Lemon nor the Larson test provides an adequate vehicle for analyzing whether religious questioning violates the Establishment Clause. It suggests that courts should instead apply a hybrid version of the two tests, the Lemon-Larson test, which is tailored to the specific concerns that religious-questioning cases raise. Part III.A introduces the mechanics of the hybrid test. Part III.B examines the sources available to courts when applying the test. Part III.C applies the Lemon-Larson test to Cherri, Isakhanova, and various other factual situations to demonstrate how it ought to be applied and how it helps work through the concerns triggered by religious-questioning policies. Part III.D concludes with a summary of why the Lemon-Larson test is the proper test to apply to these novel Establishment Clause cases.

A. A Hybrid Approach: The Lemon-Larson Test

The proposed Lemon-Larson test has three steps. Step One applies Larson strict scrutiny to the policy in question. Step Two applies the Lemon effects test. Step Three applies a balancing test if there are divergent Step One and Step Two outcomes. The Lemon-Larson test therefore replaces the first prong of Lemon, the purpose test, with the Larson strict scrutiny inquiry. It also employs a balancing test at Step Three, rather than necessarily invalidating any government policy that “fails to satisfy” any of the Lemon prongs.160

A court evaluating a religious-questioning policy under Lemon-Larson will start by applying strict scrutiny to the policy. As in Larson, this requires that the court determine whether the policy furthers a “compelling governmental interest” and is “closely fitted to further that interest.”161 The court may find the policy invalid under Step One. In this case, the court does not need to continue the analysis because the policy is unconstitutional.

If the policy passes Step One, meaning it passes strict scrutiny, the court will proceed to Step Two. Substantively, the Step Two analysis remains largely the same as the analysis already undertaken by courts under the Lemon effects test, including the incorporation of the “entanglement” inquiry.162 The court will consider the effects of a government policy and look for policies that “communicat[e] a message” of disapproval toward a religion.163 Failing at Step Two renders a policy presumptively invalid.

Unlike in Lemon, however, a government policy does not necessarily fail under Lemon-Larson if it fails the effects test. Step Three is a balancing test that weighs the effects of a government policy against the government interests underlying the policy. Step Three is triggered in cases in which Step One suggests a policy is valid and Step Two suggests that it is invalid. Although the presumption in favor of invalidity will mean that a failure at Step Two will often doom a policy, Step Three ensures that the effects test does not always prove outcome determinative. Instead, it allows for a persuasive analysis at Step One to influence the outcome even when Step Two would find a policy invalid. This requires a fact-intensive case-by-case analysis: “Every government practice must be judged in its unique circumstances.”164 The practical application of this step is explored in Part III.C. At a general level, a close finding at Step One will never outweigh a presumptive finding of invalidity at Step Two, even if the Step Two finding is also close. Similarly, a Step One analysis that suggests a policy indisputably passes strict scrutiny will not outweigh a Step Two finding that the policy just as indisputably violates the effects test. However, policies that are close calls at Step Two may nevertheless survive if they pass Step One by a wide margin.

To summarize, in order to pass the Lemon-Larson test, a religious-questioning policy must have a compelling governmental interest that is narrowly tailored and that policy must either (1) not have the effect of disapproving of (or endorsing) religion or (2) if it does have the effect of disapproving of a religion, overcome a presumption in favor of finding the policy unconstitutional.

B. Sources to Help Courts Apply the Lemon-Larson Test

Though existing Establishment Clause jurisprudence informs how courts ought to undertake Step Two, analogous areas of law in which courts have applied strict scrutiny to discrimination against protected classes can guide courts when performing the strict scrutiny test in Step One and the corresponding balancing test in Step Three. Courts have provided guidance on what count as compelling government interests. These include protecting national security and preventing crimes.165 Instructive cases also show that not every government activity that involves a protected class necessarily raises alarm.166

Specific cases provide benchmarks that can guide courts’ Steps One and Three analyses. These cases show that even generally compelling interests—such as protecting national security—will not justify a policy that appears to be driven by mere common sense, particularly when that common sense reeks of bias. Hassan v City of New York167 is a recent Third Circuit opinion that denied the city’s motion to dismiss claims concerning the broad surveillance by the New York Police Department (NYPD) of the New York City Muslim community following September 11.168 The city argued that national security and public safety concerns justified the NYPD’s surveillance policy. The court, applying heightened scrutiny to the Equal Protection claim, was not convinced. The “gravity of the threat alone cannot be dispositive of questions concerning what means law enforcement officers may employ to pursue a given purpose.”169 The court held that the city had to ground its “asserted justification” in “objective evidence,” not merely “appeals to ‘common sense’ which might be inflected by stereotypes”—and even then the policy had to “fit” better than any “alternative means.”170

Another instructive case is the Second Circuit’s Tabbaa v Chertoff.171 This case complements Hassan by offering an example of a policy that passed strict scrutiny because it was properly tailored and there were no clear alternative means available to accomplish the government’s goals. The court found a stop-and-search policy that affected certain Muslims at the border to be constitutional under strict scrutiny.172 The narrow tailoring of the policy was crucial to the court’s decision. The policy did not target all Muslims, but only participants of a conference on Islam. The CBP initiated its policy after receiving specific intelligence linking the conference to extremism.173 Finally, the searches were routine and did not include heightened or invasive searches.174

Another related and instructive line of strict scrutiny jurisprudence considers policies that discriminate based on race. While religion and race are governed by different constitutional clauses, they interact closely in the space of discrimination and profiling, making this analogy appropriate.175 One of the few times such a discriminatory policy is valid is when the government is facing a “social emergency”176 and if the chosen policy fits the “compelling goal so closely that there is little or no possibility that the motive for the classification was illegitimate racial prejudice or stereotype.”177 An informative case is United States v Montero-Camargo,178 in which the Ninth Circuit found unconstitutional a policy whereby border agents, based only on the drivers’ Hispanic ethnicity, stopped drivers out of suspicion of their immigration status.179 Echoing the concerns that animate the Lemon test, the court stated that such a policy both had “little probative value” and “send[s] a clear message that those who are not white enjoy a lesser degree of constitutional protection—that they are in effect assumed to be potential criminals first and individuals second.”180

As demonstrated in the following Section, the logic used in these and similar cases helps inform the application of the Lemon-Larson test. Taken together, the cases illustrate the high bar that government policies must clear when they single out a specific protected group. Significant government interests alone, as shown in Hassan, will not make a policy valid, nor will policies that allow room for stereotypes and prejudice to sneak in. Nevertheless, tailored policies that discriminate against a particular class but address specific, compelling government interests when no alternative is available can be constitutional, as demonstrated by the search policy in Tabbaa.

C. Applying the Lemon-Larson Test

To demonstrate the Lemon-Larson test’s applicability to religious-questioning policies, this Section applies it to various factual scenarios. In practice, this test respects both compelling governmental interests and deep wariness about any government entanglement with religion. This Section first applies the test to the policies at issue in Isakhanova and Cherri, both of which would be considered unconstitutional. It then explores similar hypothetical religious-questioning scenarios to demonstrate the nuances of Lemon-Larson that would not be provided by either test alone.181

1. Isakhanova fails the Lemon-Larson test at Step One.

Had the district court applied the Lemon-Larson test to the religious-questioning policy in Isakhanova, the analysis would have been straightforward and the result predictable. There seems to be little question that the policy should be considered unconstitutional under Step One. Recall that the prison guards in Isakhanova paired religious questions (“Do you pray five times a day?”) with facially disparaging comments (“All Muslims are terrorists”). It would be difficult for the prison to argue that this policy was driven by a compelling government interest. As shown in cases like Hassan, broad claims about general governmental interests like “national security” without “objective evidence” do not pass muster as a compelling government interest.182 In Isakhanova, the government did not even present such a broad claim. It “offered no explanation” of how the religious questions related to the prison guard’s investigation of alleged tobacco smuggling.183

Even if the defendants had raised some potentially compelling interest, such as the security of the prison, the religious questioning would still fail under Step One. Not only is this broad justification not grounded in “objective evidence,” but the questions are not narrowly tailored to this end. The derogatory comments certainly added nothing to potential fact gathering by the prison guard, and strongly suggest that religiously motivated humiliation or offense, rather than gathering evidence, was the purpose of the questioning. The Government’s best defense in Isakhanova would be to try to argue that the guard’s comments were one-off and not an official policy; if true, this might protect the government from liability.184

2. Cherri fails the Lemon-Larson test at Step Three.

While the questioning in Cherri would also fail the Lemon-Larson test, this analysis is much closer and the policy likely fails only once the effects test is balanced against the government’s interest in Step Three. This is because the government’s interest—protecting national security—was much more apparent in Cherri than in Isakhanova. As the court put it, the government’s interest in controlling who is allowed into the country is “at its zenith at the international border.”185

The crux of Step One of the Lemon-Larson analysis for Cherri is determining whether the questioning was sufficiently narrowly tailored. The nature of the questioning appears to be much more similar to the invalid surveillance in Hassan (when an entire community was surveilled based on generalized concerns) than the valid stops in Tabbaa (when the Muslim plaintiffs were subjected to standard searches based on specific intelligence about a conference they chose to attend).186 In Cherri, individuals passing through a border checkpoint were stopped for no reason other than their ethnicity and apparent religion and asked questions that are not part of a general security stop, including “How many times a day do you pray?” and “Which mosque do you go to?”187 The plaintiffs also alleged that CBP agents at other ports of entry posed questions including “Are there any extremists or terrorists at the mosque?” and “Do you know Anwar al-Awlaki [a known terrorist]?”188 Asking such an array of questions to individuals who happen to be affiliated with a specific religion does not appear narrowly tailored or grounded in “objective evidence.”189 Rather, this questioning seems to be grounded in “common sense” that is “inflected by stereotypes.”190 As such, it leaves open the “possibility” that the questioning is driven by “illegitimate . . . prejudice or stereotype.”191 As with the impermissible stops in Montero-Camargo, the Cherri questions suggest that, because of their religion, these plaintiffs were “assumed to be potential criminals first and individuals second.”192

Still, it is likely that a court would find the Cherri questioning constitutional under the Step One strict scrutiny test. Courts are “sensitive” to the security needs of border agents.193 Border agents are generally allowed broad discretion to profile individuals as they work to protect national security.194 Given the deference shown to border agents, it is likely that, in practice, a court would not closely scrutinize the rationale of the government’s national security justification at the border.195

But the analysis would not end there. The Cherri case would move to the effects test at Step Two. As discussed in Part I.A.1, the effects test asks what a well-informed reasonable observer would understand the effects of the government policy to be. Benchmarks for unconstitutional remarks come from Capistrano: the description of creationism as “religious, superstitious nonsense” was invalid, but the Mark Twain “con man” quote was valid.196 The Cherri questions ought to be found to be invalid, just as the “superstitious” comment was. The offensiveness of the Cherri questions, however, is more implicit than in the explicitly offensive Capistrano comments. The border agents did not directly slander Islam.

Instead, they used Islam as a proxy to ascertain whether the individual being questioned was a national security risk. A reasonable observer might see this as a justifiable motive. Regardless of whether the motive was proper, the questions still conflate the practice of Islam with terrorism. This communicates a message of disapproval toward Islam that a reasonable observer would find at least as offensive as referring to creationism as “superstitious nonsense,” even if the offense is not as immediately apparent.

This reasonable observer can be pieced together using social science research, rather than conjecture, as appeared to happen in the actual Cherri and Isakhanova cases. Significant research has shown that individuals and the government in post–September 11 America have internalized the perceived connection between Islam and terrorism.197 This would be particularly true for the reasonable observer in 2017. Such an individual would be generally familiar with the fact that the president of the United States had called during his candidacy for a “total and complete shutdown of Muslims entering the United States” based on concerns about “horrendous attacks by people that believe only in Jihad, and have no sense of reason or respect for human life”; had stated that “Islam hates us”; and had suggested that Christian refugees should be given priority over Muslim ones at America’s shores.198 The well-informed reasonable observer in modern America understands the perceived connection between Islam and terrorism. Questions centered on Islam, especially in the context of an interrogation by a government official tasked with protecting American security, express disapproval by sanctioning this negative stereotype of the religion. Furthermore, this was not a one-off comment by a visiting minister, as in Town of Greece.199 This was a formal policy put into place by government officials who determined that the way to gauge whether an individual is a risk is to know the extent to which he follows Islam.200 Given the scope of the policy and the reasonable observer’s awareness of recent American history, the religious questioning would be understood to “denigrate” a religion by equating its followers with security threats.201

Because the Cherri policy fails the Step Two effects test, it would be presumptively invalid. It fails to overcome that presumption at Step Three. Here, the strongly negative effects balanced against the borderline validity under strict scrutiny would render the Cherri questioning unconstitutional. The Cherri policy barely passes strict scrutiny: it does because the government has a compelling interest in protecting national security, with additional discretion at the border. The policy still suffers from a lack of a concrete connection between the questions asked and that interest, and from a failure to narrowly tailor the questioning. Asserting the “gravity of [a] threat alone” does not justify an otherwise-unjustifiable government policy.202 Compared with barely passing under strict scrutiny, the Cherri policy certainly fails Step Two; it expresses animus toward Islam by using Islamic observance as a factor indicating a national security threat. The policy therefore fails: the balancing test at Step Three presumptively finds a policy invalid if it fails Step Two. Because this policy barely passed Step One, it does not overcome that presumption.

The Government would have a few potential defenses against finding the Cherri policy unconstitutional. For one, a court has to weigh the presumption of invalidity under Lemon-Larson against the deference afforded the government at the border, where its power is at its “zenith.” Given this deference, the Government could argue that a court should treat the Cherri policy just as the Second Circuit treated the search policy in Tabbaa.203 That court relied on the CBP’s “extensive expertise in securing the border” and deferred to the agency’s decision that the “routine procedures” were called for.204 These “routine procedures” included targeting attendees at an Islamic conference for searches, fingerprinting, questioning, and photographing.205 The Tabbaa plaintiffs (who did not raise Establishment Clause claims)206 were not questioned about their religious beliefs, but the Government could posit that the connection between their religion and the searches was just as clear as in Cherri. The searches in Tabbaa were potentially more intrusive, at least physically, than the questioning in Cherri. These facts weigh in favor of the same outcome in each case. Such arguments have some merit, but ultimately they fail. The Government in Tabbaa claimed that it had received specific intelligence about potential danger from attendees of the conference at issue,207 and the searches performed, while extensive, were tailored to determining whether the searched individuals posed a threat. The questions asked of the Cherri plaintiffs did not fulfill the same purpose as running a fingerprint or searching a car for dangerous items.

Finding the Cherri policy unconstitutional under Lemon-Larson demonstrates that the test remains consistent with the theory underpinning Lemon. Lemon does not employ a balancing test or engage with strict scrutiny because it prohibits all excessive government entanglement with religion. The Lemon-Larson test gives the government some breathing room for situations in which religion truly does implicate government interests. If the Larson step consistently outweighed the effects test, however, that would reduce the Lemon-Larson test to the Larson test. The Cherri policy signals clear, strong government disapproval of Islam, impermissible under Lemon. Its validity under Step One strict scrutiny is a close call. The fact that the policy may be appropriate under Larson, therefore, is not enough to render it valid at Step Three.

3. Other factual scenarios demonstrate the Lemon-Larson test’s fit for religious-questioning cases.

The Lemon-Larson test can be applied to a broader set of potential fact patterns than those in Cherri and Isakhanova, both of which similarly concern a law enforcement official questioning individuals who have done nothing to raise suspicions about their Islamic beliefs. Hypothetical fact patterns (loosely based on true events) further demonstrate that Lemon-Larson is well suited to analyzing the unique concerns raised in religious-questioning cases.

There are some fact patterns that can be decided easily under the Lemon-Larson test. These demonstrate that Lemon-Larson does not disrupt existing jurisprudence surrounding longstanding US practices; it creates a way to accommodate new ones. The census, for instance, is still constitutional under both Step One and Step Two of Lemon-Larson (meaning no Step Three balancing is required). The government collects information on citizens’ self-described religious identification through the American Religious Identification Survey (“ARIS”).208 The government interest here is clear and unobjectionable (to gather demographic information)209 and the question is tailored to the specific piece of information the government wants as closely as it could be. There is no reason to believe that a reasonable observer would understand a general question about religious affiliation to disapprove of, or endorse, religion.

The questioning in Vernon210 would likewise still be constitutional. Under Step One, the government’s purpose was clear (making sure an assistant police chief was not breaking the law)211 and the investigation was narrowly tailored to precisely the behavior that had called into question Vernon’s ability to faithfully execute his job. Moving on to Step Two, a reasonable observer would understand that the questioning was driven by Vernon’s own conflation of his religious beliefs with his responsibilities as a police officer, and not by disapproval of his sect of Christianity.

Hypothetical religious-questioning scenarios demonstrate the Lemon-Larson test would not allow law enforcement officers to invoke religion by alleging “compelling interests” that were clearly pretextual. An example of an obviously unconstitutional religious-questioning policy would be police wandering around Borough Park in New York City while investigating a child-abuse scandal involving the Hasidic Jewish community,212 stopping passersby who appear to be Hasidic Jews, and asking them “Why would you believe in such an outdated religion?” and “Why would you be a part of a religion that covers up child abuse?” The government might claim that, as in Vernon, the police are attempting to solve a crime. However, the insulting questions are ill fitted to finding the perpetrator of the crime. The individuals being questioned have demonstrated no connection between their potentially criminal actions and their religion, and there is presumably no “objective evidence” to prove a connection between the interest and the questions. This policy would fail at Step One.

The Lemon-Larson test allows for a nuanced, informative analysis in situations in which religious questions may actually uncover helpful information and the policy has a facially legitimate, reasonable government purpose. One difficult and close, but ultimately constitutional, fact pattern would be if the police gain intelligence that members of the Kingston Group, a radical religious sect holding fundamentalist Mormon beliefs in Salt Lake City, Utah, are engaging in polygamy and making forced marriage arrangements via letters.213 The government wants to figure out who is engaged in this scheme, but members of the Kingston Group wear “conventional clothing” with no obvious markers.214 With no other clues to go on other than the group’s stated religious beliefs, postal workers at area post offices are instructed to discreetly pull aside everyone who they believe to be Mormon and ask them questions including “Do you believe in polygamy?”; “Do you know anyone in the Kingston family?”; and “Why are you Mormon?” Starting at Step One, the government has a compelling interest in stopping crime. The questions are also more narrowly tailored than Hassan or Cherri. As in Tabbaa, the government is attempting to target members of a specific community engaged in a specific flagged activity (mailing letters). The activity is legal (as is attending an international conference), but the government has reason to believe it is being used to facilitate a crime. The similarities between this situation and Tabbaa suggest the policy passes strict scrutiny.

Though it passes Step One, this policy fails Step Two. This failure turns on how much a hypothetical reasonable observer knows. The Kingston Group has no affiliation with the mainstream Church of Latter Day Saints (LDS), which disavowed polygamy long ago.215 It is a small splinter group, and the questions are trying to target individuals affiliated with that group. For an individual well versed in this background, these questions would pass Step Two. They are not disapproving of the practices of Mormonism, but instead trying to weed out members of a group that is breaking the law. There is no research, however, on what proportion of the population has heard of the Kingston Group and its radical beliefs. For this reason, assume the reasonable observer has not heard of this small group. A reasonable observer who has general knowledge about the history of polygamy, the Mormon Church, and American disapproval of polygamy—but not about the splinter Kingston Group—might miss the nuance of the questioning. No matter the government’s actual motives, the focus on Mormons in attempting to investigate the forced-marriage scheme could very well be seen as claiming that it is Mormons, rather than members of the Kingston Group, who engage in illegal, forced polygamy. This would treat Mormons as “second-class citizens” and disapprove of the religion.216

Because it fails at Step Two, the policy moves on to—and passes—Step Three. While the policy clearly passes Step One strict scrutiny, it only narrowly fails Step Two. Its failure at Step Two is also mitigated by the fact that a reasonable observer in Salt Lake City, where the Kingston Group is based and where it presumably has increased notoriety, would more likely be familiar with the group and understand the true thrust of the questions than a random American. Overall, it is a relatively narrow policy, which attempts to implicate religion only insofar as it relates to solving a specific crime. Some questions, such as “Why are you Mormon?”, may not be tailored as narrowly as possible. It is unlikely this question will be answered with “To engage in forced polygamy!” It appears plausible, though, for the government to argue that its questions are truly trying to root out followers of the Kingston Group. Assuming the police have evidence of the criminal activities of the Kingston Group but no way to uncover its members other than through this type of individual interrogation, then these questions are plausibly among those most able to narrow the population down to members of the Kingston Group who may be engaged in the criminal enterprise. The policy is not perfect, but courts are deferential (within limits, as shown by Hassan) to stated government interests when they are supported with more than superficial evidence.

A hypothetical, revised Cherri policy presents another challenging application of the Lemon-Larson test. Consider the following: The Cherri policy is found unconstitutional, and the government hopes to conform with the law by making the questions religiously neutral with a “Cherri-plus” policy. Border agents are now charged with asking questions including “Do you consider yourself religious?”; “How often do you attend a house of worship?”; and “Are there any religious zealots or extremists of which you are aware at your house of worship?” If the answers suggest that the individual is devout and the border agent feels something is amiss, the agents have more particular questions to ask that are tailored to specific religious practices. Agents have discretion over which people to question, but have been directed that Arab Muslims and people with pro-life bumper stickers (who the officials see as representing a threat to people at clinics that offer abortions)217 should be of particular interest. With the Cherri-plus policy, the government has diluted its focus on Islam and attempted to use facially neutral questions that invoke religion but only appear to call the legitimacy of specific religious practices into question once the agent believes red flags are raised by the screening questions.

Under Lemon-Larson, the Cherri-plus policy would still be unconstitutional. The core problem with this type of questioning, hard to overcome in any permutation of the policy, is that it equates an individual’s religion with a threat—without specific reason to do so. It invokes religion when the individual being questioned has not given the government a reason to think his religious beliefs pose a threat. Even though the new policy overtly takes the focus off of Islam, it impermissibly treats devotion to religion of any kind as an incriminating characteristic; this line of questioning is worse than what occurred in Cherri because it continues to treat Islam as a threat but sweeps more religions into the umbrella of beliefs that trigger the government’s scrutiny.

Therefore, rather than becoming more acceptable by incorporating more religions, the Cherri-plus policy is actually more concerning under Lemon-Larson than Cherri was. There are two potential reasons for the expansion of the policy: the government thinks that devoted individuals of other creeds are also a threat or it is asking those questions to diffuse the focus on Islam. If the former, the policy is invalid at Step One. Under the logic of Larson and Hassan, the government cannot, for example, cite a “security” interest in questioning pro-life Christians without presenting an “actual concrete problem.”218 This policy sweeps an entire community of innocent individuals into questioning that imprecisely probes a poorly defined threat. The questions are also poorly tailored—most pro-life Christians pose no threat to abortion clinics, and the questions are too vague to efficiently identify those few who might.

On the other hand, if the government is merely asking these questions to make it appear that the focus is not on Islam, then the Cherri-plus policy also fails the Lemon-Larson test. Under Step One, rather than being narrowly tailored to an “actual concrete problem,” this policy would just obfuscate the true purpose of questioning Muslims. This purpose can be understood in classic Lemon terms—the stated purpose of a policy cannot be a pretextual “sham,” covering up the actual intent to disapprove of or endorse a religion.219

If the Cherri-plus policy moved on to Step Two despite the lack of narrow tailoring, a court would consider the policy’s effects. The government in this hypothetical considers religion a key factor in determining whether an individual poses a threat. This certainly impermissibly “communicat[es] a message” of disapproval to religion.220 An altered Cherri-plus policy is unconstitutional under Lemon-Larson, whether the government genuinely crafted it to target several religions viewed by the government as risks or whether the government meant to obscure the true target of Islam.

D. Why to Apply the Lemon-Larson Test

This Comment proposes applying a novel test to religious-questioning cases under the Establishment Clause because existing tests fail to adequately address the concerns raised in these cases. The application of Lemon-Larson to Cherri, Isakhanova, and the other scenarios described in Part III.C demonstrates how the Lemon-Larson test accommodates both the unique role of religion in American society (captured by the Lemon test) and the unique concerns raised when the government truly has a compelling reason to become entangled in religion (captured by the Larson test). The test does so without requiring a wholesale invention of a new test that would disrupt Establishment Clause jurisprudence.

The Court’s Establishment Clause tests are incapable of adequately analyzing religious questioning. One could argue the opposite, pointing to the use of these tests in analogous cases like Capistrano that consider disparaging remarks toward religion.221 It is true that disparaging remarks are similar in form to religious questioning. However, they do not implicate similarly forceful government interests, such as the national security interests raised by policies like the one in Cherri. For that reason, Lemon is inadequate to handle the asserted compelling interests in religious-questioning cases. Lemon invalidates policies that violate any of its prongs. This provides no breathing room for the government when religion truly implicates compelling interests. Cases like Hassan and Tabbaa demonstrate that courts are generally willing to provide governments with this breathing room in other spaces. The facts in Cherri are a good example of this shortcoming. While Cherri is unconstitutional under Lemon-Larson, the test compels a court to grapple with the national security interests claimed by the government even though the policy communicates a message of disapproval toward a religion. Lemon cannot accommodate a compelling government interest that entangles religion in the same way. On the other hand, Larson alone cannot address these concerns. The Larson test does not incorporate concerns about religious liberty in the same way as the other Establishment Clause tests—a possible reason for its rare invocation by the courts. Larson suffers from the same shortcomings as the strict scrutiny tests applied in the Free Exercise Clause and RFRA contexts.222 In addition, the Court has applied strict scrutiny only to those Establishment Clause cases that concern policies that “patently” discriminate among religions.223 Although religious questioning may implicate discrimination among religions, it does not necessarily do so. The Larson test, as a stand-alone test, should not be applied to these policies: the Court created it to apply to a specific subset of cases and did not tailor it to assess the validity of religious policies more broadly.

The Lemon-Larson test succeeds in analyzing religious-questioning policies when either test on its own would come up short. The applications of the Lemon-Larson test to the fact patterns from Part III.C demonstrate how the Lemon-Larson test shores up Establishment Clause jurisprudence. The Utah post office example demonstrates the way the Lemon-Larson test can offer greater flexibility for government policies that become entangled with religion if the entanglement is for demonstrably compelling reasons. Yet the Cherri and the Cherri-plus examples show that flexibility does not come at the expense of longstanding Establishment Clause values embedded in Lemon. The Larson test would render Cherri and parts of the Cherri-plus policy constitutional, but the Lemon-Larson test invalidates them. In this way, the Lemon-Larson test sets a higher bar for government policies that implicate religion than strict scrutiny, remaining faithful to historic Establishment Clause jurisprudence.

In essence, the two tests complement each other. Under Lemon, courts invalidate a policy if it does not have a secular purpose or if it has the effect of endorsing or disapproving of a religion.224 The compelling-interest balancing that is at the core of Larson strict scrutiny225 addresses Lemon’s primary shortcoming by allowing the government to adopt religious-questioning policies when truly called for. It provides courts with more flexibility without sacrificing the rigorous analysis required whenever a government policy implicates religion. At the same time, the Lemon-Larson test does not abandon Lemon’s values; it just uses Larson to tailor the inquiry. Step Two ensures that religious questioning is not merely subjected to a strict scrutiny analysis. Courts will still presume it is invalid if it fails the effects test in Step Two. This enhanced analysis successfully addresses the shortcomings from which either test applied alone suffers, making it the proper test to apply to religious questioning.

Adopting the Lemon-Larson test is also preferable to creating an entirely new test for religious-questioning policies. There are benefits to this latter option. An entirely new test could be completely tailored to religious questioning, free of the baggage of previous analyses and tangled jurisprudence. There are several reasons not to take this step. The primary one is that religious questioning, while unique, is still analogous to other Establishment Clause cases. The Lemon-Larson test can benefit from related, existing case law rather than require an entirely new set of decisions to flesh out its proper application.

Furthermore, Lemon-Larson will likely be less controversial to adopt than a wholly new test. Courts already frequently use multiple Establishment Clause tests in their opinions.226 The Court in Santa Fe Independent School District v Doe227 may not have explicitly adopted a new Establishment Clause test when it considered student-led prayers before football games,228 but it was willing to tweak its existing tests to work with the facts at hand. The opinion mentions Lemon only in its final paragraphs.229 The core of the discussion centers on related concerns, including endorsement, the coercion inherent in school–student relationships, and the difference between private and government speech, but does not confront these issues explicitly using the Lemon steps.230 The Court appeared to think that this type of discussion was more helpful in determining the validity of the policy considered in Santa Fe Independent School District than a Lemon approach, which silos each step. Van Orden is another example of the Court adapting its existing analyses when considering new manifestations of Establishment Clause cases.231 It took a more dramatic approach than Santa Fe Independent School District; the Court thought Lemon itself did not need to be applied because of the different concerns triggered by the historical dimensions of the statue.232 The Lemon principles still motivated the decision, though. The Court’s ultimate decision focused on the “dual significance”—both religious and governmental—of the statue (reflecting the search for a secular purpose) and that the plaintiff “walked by the monument for a number of years” before finding it upsetting (suggesting the monument did not have an invalid effect on him).233 Another case in which the Court stepped outside of the traditional Establishment Clause framework is Zelman v Simmons-Harris.234 The Court did not think the Lemon test necessary in its discussion of whether the school voucher program at issue violated the Establishment Clause.235 It found more helpful a close analysis of the difference between government programs that directly provide aid to religious schools and those that offer “true private choice.”236 Its analysis did remain motivated by the Lemon principles.237 Both Lemon and Larson are already Supreme Court–promulgated Establishment Clause tests. This Comment merely proposes adapting them to a new manifestation of Establishment Clause question. As with statues of historical significance and programs promoting school choice, religious-questioning policies raise unique concerns that merit unique attention within Establishment Clause jurisprudence.

Finally, Larson fits naturally into the Lemon framework. Lemon’s first prong already looks at the purpose behind a government policy. Larson is also interested in the purposes of a policy. As compared to Lemon, which automatically invalidates a policy once it becomes overly entangled with religion, strict scrutiny allows for a more nuanced look at the narrowness and legitimacy of that purpose. The strict scrutiny framework provides the additional tools to assess the constitutionality of government purposes that necessarily implicate religion. It does so without requiring courts to significantly change the way that they approach Establishment Clause questions. The Lemon-Larson test will not clash with existing Establishment Clause jurisprudence; it seeks to take advantage of the way that Larson complements the Lemon test’s ability to navigate the role of religion in American society by offering a preexisting framework through which to measure the importance of a governmental interest.

Conclusion

Religious questioning by government officials implicates, but has yet to be fully analyzed under, the Establishment Clause. President Trump’s recent emphasis on refugees’ religions shows the need to understand how this type of policy—which invokes religion but in meaningfully different ways than traditional Establishment Clause cases—should be treated under the Establishment Clause. This Comment addresses that gap in the literature. Traditional Establishment Clause jurisprudence, created to accommodate the unique relationship between the American government and religion, is best applied to this type of case through a modified analysis, the Lemon-Larson test. This test pairs the traditional Establishment Clause inquiry with a strict scrutiny analysis that allows for some government entanglement with religion when that entanglement is truly driven by compelling government interests. This hybrid test would conclude that most types of religious questioning, including the policies in Cherri and Isakhanova, violate the Establishment Clause. At the same time, it affords the government breathing room to implicate religion when truly called for. By tailoring existing Establishment Clause analyses to the concerns raised by religious-questioning policies, the Lemon-Larson test allows for a thorough Establishment Clause analysis of a new form of government entanglement with religion.

• 1See Cherri v Mueller, 951 F Supp 2d 918, 924–26 (ED Mich 2013) (detailing similar questioning faced by Ali and his coplaintiffs).
• 2See International Refugee Assistance Project v Trump, 857 F3d 554, 594–95 (4th Cir 2017), vacd and remd, 2017 WL 4518553 (US) (listing times when Trump “expressed anti-Muslim sentiment”); Daniel Burke, Trump Says US Will Prioritize Christian Refugees (CNN, Jan 28, 2017), archived at http://perma.cc/BM7H-MNLP.
• 3951 F Supp 2d 918 (ED Mich 2013) (granting in part and denying in part defendants’ motion to dismiss).
• 4Id at 923–27.
• 52016 WL 1640649 (ND Cal) (denying defendants’ motion to dismiss).
• 6Id at *5–6.
• 7US Const Amend I, cl 1. See also Cherri, 951 F Supp 2d at 935; Isakhanova, 2016 WL 1640649 at *5–6.
• 8See Everson v Board of Education of Ewing Township, 330 US 1, 15–16 (1947).
• 9See Amanda Holpuch and Ashifa Kassam, Canadian Muslim Grilled about Her Faith and View on Trump at US Border Stop (The Guardian, Feb 10, 2017), archived at http://perma.cc/TFR3-35YU (reporting that questions included “Which mosque do you go to? What is the name of the imam? How often do you go to the mosque? What kind of discussions do you hear in the mosque? Does the imam talk to you directly?”). See also International Refugee Assistance Project, 857 F3d at 572 (affirming a preliminary injunction against the executive order “that in text speaks with vague words of national security, but in context drips with religious intolerance, animus, and discrimination”).
• 10Durbin asked Barrett, “Do you consider yourself an ‘orthodox Catholic’?” while Feinstein noted, “[T]he dogma lives loudly within you. And that’s of concern.” Alexandra Desanctis, Did Durbin and Feinstein Impose a Religious Test for Office? (National Review, Sept 8, 2017), archived at http://perma.cc/LQL3-W7RY.
• 11See Lemon v Kurtzman, 403 US 602, 612–13 (1971); Larson v Valente, 456 US 228, 246 (1982).
• 12US Const Amend I, cl 1. Despite the word “Congress,” the Supreme Court has interpreted the First Amendment as applying to all federal government officials, as well as all state officials, through the Fourteenth Amendment. See Engel v Vitale, 370 US 421, 429–30 (1962).
• 13330 US 1 (1947).
• 14Id at 15–16.
• 15See, for example, Lynch v Donnelly, 465 US 668, 692 (1984) (O’Connor concurring).
• 16See Engel, 370 US at 430.
• 17Van Orden v Perry, 545 US 677, 686 (2005), quoting Hunt v McNair, 413 US 734, 741 (1973).
• 18See, for example, Santa Fe Independent School District v Doe, 530 US 290, 314–17 (2000) (applying the Lemon test but also considering the endorsement test and historical practice).
• 19Van Orden, 545 US at 697 (Thomas concurring) (“The unintelligibility of this Court’s precedent raises the further concern that, either in appearance or in fact, adjudication of Establishment Clause challenges turns on judicial predilections.”).
• 20See, for example, American Civil Liberties Union of Ohio Foundation, Inc v DeWeese, 633 F3d 424, 431 (6th Cir 2011). Use of the Lemon test continues despite considerable criticism. See, for example, Santa Fe Independent School District, 530 US at 319 (Rehnquist dissenting) (listing cases that fault Lemon and bemoaning “the sisyphean task of trying to patch together the blurred, indistinct, and variable barrier described in Lemon”) (quotation marks omitted). See also Emily Fitch, Comment, An Inconsistent Truth: The Various Establishment Clause Tests as Applied in the Context of Public Displays of (Allegedly) “Religious” Symbols and Their Applicability Today, 34 NIU L Rev 431, 445 (2014).
• 21See Committee for Public Education and Religious Liberty v Nyquist, 413 US 756, 773–80 (1973) (holding that tax benefits to parents whose children enrolled in nonpublic schools were unconstitutional); Roemer v Board of Public Works of Maryland, 426 US 736, 748–54 (1976) (holding that grants to private religious colleges were constitutional); Santa Fe Independent School District, 530 US at 314–16 (holding that school policy permitting student-led prayers before school football games was unconstitutional); McCreary County, Kentucky v American Civil Liberties Union of Kentucky, 545 US 844, 859–65 (2005) (holding that displays of the Ten Commandments in courthouses were not necessarily unconstitutional).
• 22403 US 602 (1971).
• 23See id at 612–13.
• 24See text accompanying notes 52–55.
• 25Though originally articulated as a secular legislative purpose, the Court has since applied the purpose prong broadly to “legislation or governmental action.” Lynch, 465 US at 680.
• 26Lemon, 403 US at 612–13.
• 27Edwards v Aguillard, 482 US 578, 583 (1987).
• 28Lynch, 465 US at 690 (O’Connor concurring).
• 29See Wallace v Jaffree, 472 US 38, 56 (1985).
• 30McCreary County, 545 US at 860, quoting Corporation of Presiding Bishop of Church of Jesus Christ of Latter-Day Saints v Amos, 483 US 327, 335 (1987).
• 31Wallace, 472 US at 56.
• 32Stone v Graham, 449 US 39, 41 (1980).
• 33Edwards, 482 US at 586–87.
• 34See Zelman v Simmons-Harris, 536 US 639, 649, 653 (2002).
• 35DeWeese, 633 F3d at 434 (quotation marks omitted).
• 36School District of Abington Township v Schempp, 374 US 203, 222 (1963).
• 37Lynch, 465 US at 692 (O’Connor concurring) (emphasis added).
• 38See Santa Fe Independent School District, 530 US at 316 (“Therefore, the simple enactment of this policy, with the purpose and perception of school endorsement of student prayer, was a constitutional violation. . . . Government efforts to endorse religion cannot evade constitutional reproach based solely on the remote possibility that those attempts may fail.”).
• 39Lynch, 465 US at 680.
• 40Id at 681.
• 41Santa Fe Independent School District, 530 US at 312 (finding that the effect of pregame prayers was to impermissibly pressure students to participate in worship).
• 42DeWeese, 633 F3d at 434–35 (finding that this display of the Ten Commandments had the effect of endorsing religion).
• 43See McCreary County, 545 US at 862 (noting that “the eyes that look to purpose belong to an ‘objective observer’” and do not call for “psychoanalysis of a drafter’s heart of hearts”).
• 44Lynch, 465 US at 688 (O’Connor concurring).
• 45See id at 690 (O’Connor concurring); Santa Fe Independent School District, 530 US at 308.
• 46Catholic League for Religious and Civil Rights v City and County of San Francisco, 624 F3d 1043, 1049 (9th Cir 2010).
• 47Lund v Rowan County, North Carolina, 863 F3d 268, 290 (4th Cir 2017) (en banc).
• 48Felix v City of Bloomfield, 841 F3d 848, 863–64 (10th Cir 2016).
• 49See, for example, Utah Highway Patrol Association v American Atheists, Inc, 565 US 994, 1004 (2011) (Thomas dissenting from denial of certiorari) (“One might be forgiven for failing to discern a workable principle that explains these wildly divergent outcomes.”).
• 50Cressman v Thompson, 798 F3d 938, 958 (10th Cir 2015). See also American Civil Liberties Union of Kentucky v Mercer County, 432 F3d 624, 636 (6th Cir 2005).
• 51See, for example, Freethought Society, of Greater Philadelphia v Chester County, 334 F3d 247, 260 (3d Cir 2003) (“[W]e will assume that the reasonable observer is informed about the approximate age of the plaque and the fact that the County has done nothing with the plaque since it was erected; we also conclude that the reasonable observer is aware of the general history of Chester County.”).
• 52521 US 203 (1997).
• 53Id at 232–35.
• 54Zelman, 536 US at 668–69 (citation omitted). See also Agostini, 521 US at 218, 232–33.
• 55See Agostini, 521 US at 218, 232–33. As discussed in Part II.B, the two district-court decisions that inspired this Comment both incorrectly analyzed this prong under Agostini, treating it as an independent part of the analysis.
• 56Lemon, 403 US at 614.
• 57Id.
• 58Id.
• 59Id at 606–07.
• 60Lemon, 403 US at 619.
• 61Id at 615.
• 62456 US 228 (1982).
• 63Id at 230.
• 64Id at 245.
• 65See id at 251–52 (“The [Lemon test is] intended to apply to laws affording a uniform benefit to all religions, and not to provisions, like [the law considered in Larson], that discriminate among religions.”) (citation omitted).
• 66Larson, 456 US at 246–47.
• 67Awad v Ziriax, 670 F3d 1111, 1129 (10th Cir 2012), citing Brown v Entertainment Merchants Association, 564 US 786, 799 (2011).
• 68Larson, 456 US at 247.
• 69670 F3d 1111 (10th Cir 2012).
• 70Id at 1131.
• 71See Jeremy Patrick-Justice, Strict Scrutiny for Denominational Preferences: Larson in Retrospect, 8 NY City L Rev 53, 76–87 (2005) (discussing the infrequent use of Larson by the Supreme Court).
• 72See Awad, 670 F3d at 1127–28.
• 73See Patrick-Justice, 8 NY City L Rev at 76–87, 120 (cited in note 71) (discussing how Larson is on the “peripher[y]” of the Court’s Establishment Clause jurisprudence, but remains an active doctrine).
• 74See id at 76.
• 75See note 92.
• 76See, for example, C.F. v Capistrano Unified School District, 615 F Supp 2d 1137, 1145 (CD Cal 2009), vacd on other grounds, 654 F3d 975 (9th Cir 2011).
• 77465 US 668, 681 (1984). For more discussion of the complexity of Establishment Clause jurisprudence, see Part I.A.3 (introducing cases that rely on neither the Larson nor the Lemon test).
• 78Lynch, 465 US at 687 n 13.
• 79See id. See also Patrick-Justice, 8 NY City L Rev at 81 (cited in note 71).
• 80See Awad, 670 F3d at 1128.
• 81For a detailed discussion of these tests, see Fitch, Comment, 34 NIU L Rev at 435–44 (cited in note 20).
• 82See Lynch, 465 US at 687–90 (O’Connor concurring). O’Connor’s concurrence from Lynch is widely cited, including by the Supreme Court, and treated as very influential, if not binding, law. See Fitch, Comment, 34 NIU L Rev at 436–37 (cited in note 20).
• 83See Good News Club v Milford Central School, 533 US 98, 114 (2001); Rosenberger v Rector and Visitors of University of Virginia, 515 US 819, 839 (1995) (“A central lesson of our decisions is that a significant factor in upholding governmental programs in the face of Establishment Clause attack is their neutrality towards religion.”).
• 84See Van Orden, 545 US at 694 (Thomas concurring).
• 85See, for example, County of Allegheny v American Civil Liberties Union Greater Pittsburgh Chapter, 492 US 573, 659–60 (1989) (Kennedy concurring in part and dissenting in part).
• 86See, for example, Van Orden, 545 US at 694–98 (Thomas concurring).
• 87545 US 677 (2005).
• 88See id at 686.
• 89Id.
• 90Town of Greece v Galloway, 134 S Ct 1811, 1819 (2014).
• 91See id at 1818–19 (citations omitted):

  [Marsh v Chambers] is sometimes described as “carving out an exception” to the Court’s Establishment Clause jurisprudence, because it sustained legislative prayer without subjecting the practice to “any of the formal ‘tests’ that have traditionally structured” this inquiry. The Court in Marsh found those tests unnecessary because history supported the conclusion that legislative invocations are compatible with the Establishment Clause. . . . Yet Marsh must not be understood as permitting a practice that would amount to a constitutional violation if not for its historical foundation.

  See also Marsh v Chambers, 463 US 783, 800–02 (1983) (Brennan dissenting).

• 92See American Family Association, Inc v City and County of San Francisco, 277 F3d 1114, 1122 (9th Cir 2002) (“[B]ecause it is far more typical for an Establishment Clause case to challenge instances in which the government has done something that favors religion or a particular religious group, we have little guidance concerning what constitutes a primary effect of inhibiting religion.”). See also Jay Wexler, Government Disapproval of Religion, 2013 BYU L Rev 119, 120–24 (discussing the “exceedingly rare” use of the disapproval side of the Establishment Clause).
• 93See Vasquez v Los Angeles County, 487 F3d 1246, 1255 (9th Cir 2007) (“Although Lemon is most frequently invoked in cases involving alleged governmental preferences to religion, the test also ‘accommodates the analysis of a claim brought under a hostility to religion theory.’”), citing American Family Association, 277 F3d at 1121.
• 94Everson, 330 US at 15–16.
• 95487 F3d 1246 (9th Cir 2007).
• 96Id at 1255.
• 9727 F3d 1385 (9th Cir 1994).
• 98See id at 1396–1401.
• 99Id at 1388, 1396–1401.
• 100Id at 1388–89.
• 101See Vernon, 27 F3d at 1388–89.
• 102See id at 1388–89, 1398–99. See also American Family Association, 277 F3d at 1121–23 (holding that a city prohibition on anti-gay advertisements paid for by religious groups had a secular purpose and a primary effect of “encouraging equal rights for gays and discouraging hate crimes,” not of “inhibiting” religion).
• 103Awad, 670 F3d at 1116, 1126–29.
• 104Id at 1130 (quotation marks omitted).
• 105463 US 783 (1983).
• 106Id at 794–95 (“The content of the prayer is not of concern to judges where, as here, there is no indication that the prayer opportunity has been exploited to proselytize or advance any one, or to disparage any other, faith or belief.”) (emphasis added).
• 107615 F Supp 2d 1137, 1146–49 (CD Cal 2009), vacd on other grounds, 654 F3d at 978, 988 (granting the teacher qualified immunity, but acknowledging that “[a]t some point a teacher’s comments on religion might cross the line and rise to the level of unconstitutional hostility”). For more discussion, see generally Jennifer L. Bryant, Note, Talking “Religious, Superstitious Nonsense” in the Classroom: When Do Teachers’ Disparaging Comments about Religion Run Afoul of the Establishment Clause?, 86 S Cal L Rev 1343 (2013).
• 108Capistrano, 615 F Supp 2d at 1146.
• 109Id.
• 110Id at 1149.
• 111Id.
• 112134 S Ct 1811 (2014).
• 113See id at 1824.
• 114Id (“Absent a pattern of prayers that over time denigrate, proselytize, or betray an impermissible government purpose, a challenge based solely on the content of a prayer will not likely establish a constitutional violation.”).
• 115Capistrano, 615 F Supp 2d at 1153. See also Capistrano, 654 F3d at 986.
• 116US Const Amend I, cl 1 (preventing the government from “prohibiting the free exercise” of religion).
• 117US Const Amend XIV, cl 1 (“No State shall . . . deny to any person within its jurisdiction the equal protection of the laws.”).
• 118Pub L No 103-141, 107 Stat 1488, codified at 42 USC § 2000bb et seq.
• 119See Van Orden, 545 US at 683–86 (quotation marks omitted), quoting Lynch, 465 US at 674.
• 120See, for example, Van Orden, 545 US at 683–84 (discussing how every Establishment Clause case must recognize that “[o]ur institutions presuppose a Supreme Being” and that the Court must flexibly respond to this history by “neither abdicat[ing] our responsibility to maintain a division between church and state nor evinc[ing] a hostility to religion by disabling the government from in some ways recognizing our religious heritage”).
• 121See Stephen Pepper, Taking the Free Exercise Clause Seriously, 1986 BYU L Rev 299, 306 (“[T]he establishment clause focuses on the protection of government from the encroachment of the church . . . while the free exercise clause reflects [ ] the . . . view of protecting religion from the state.”).
• 122Id at 306 n 34.
• 123See, for example, City of Richmond v J.A. Croson Co, 488 US 469, 493–94 (1989).
• 124Although the phrase “strict scrutiny” is not often used in the Free Exercise Clause context (whose case law at times mirrors the confusion seen in Establishment Clause jurisprudence), courts apply the “most exacting scrutiny” and look for “compelling” government interests. Trinity Lutheran Church of Columbia, Inc v Comer, 137 S Ct 2012, 2021, 2024 (2017).
• 125See, for example, Gonzales v O Centro Espirita Beneficente Uniao do Vegetal, 546 US 418, 430 (2006).
• 126See Patel v United States Bureau of Prisons, 515 F3d 807, 813 (8th Cir 2008) (describing how this burden must “significantly inhibit or constrain conduct or expression that manifests some central tenet of a person’s individual religious beliefs,” “meaningfully curtail a person’s ability to express adherence to his or her faith,” or “deny a person reasonable opportunities to engage in those activities that are fundamental to a person’s religion”), citing Murphy v Missouri Department of Corrections, 372 F3d 979, 988 (8th Cir 2004).
• 127See Braunfeld v Brown, 366 US 599, 605 (1961).
• 128See Patel, 515 F3d at 814.
• 129The Court has found that RFRA applies when an individual “alleges a substantial burden on his or her free exercise of religion.” City of Boerne v Flores, 521 US 507, 532 (1997).
• 130Id at 536. Only twenty-nine states have adopted state-level RFRAs. See Ira C. Lupu, Moving Targets: Obergefell, Hobby Lobby, and the Future of LGBT Rights, 7 Ala CR & CL L Rev 1, 48–49 (2015). RFRA expands the protections granted by the Free Exercise Clause by extending scrutiny to “neutral, generally applicable” laws that “incidentally” burden religious exercise, which are not the type of policy considered in this Comment. Holt v Hobbs, 135 S Ct 853, 859 (2015).
• 131See Galloway v Town of Greece, 681 F3d 20, 30 n 4 (2d Cir 2012), revd on other grounds, 134 S Ct 1811 (discussing the broad standing available for Establishment Clause claims).
• 132For a brief summary of modern standing requirements, see Lujan v Defenders of Wildlife, 504 US 555, 560–61 (1992).
• 133See Vernon, 27 F3d at 1388–89.
• 134See Cherri, 951 F Supp 2d at 923–27.
• 135Id at 927, 935.
• 136Id at 924.
• 137Id.
• 138Cherri, 951 F Supp 2d at 926. The defendants had previously explored the legality of these policies internally, finding little guidance from the courts. See id at 924–25.
• 139See id at 937–38. The court did not dismiss the plaintiffs’ Fifth Amendment claims. Id.
• 140Id at 933–36.
• 141Id at 936.
• 142Cherri, 951 F Supp 2d at 936–37. For a discussion of what an Establishment Clause analysis adds, see Part I.C.
• 143See Isakhanova, 2016 WL 1640649 at *1, 5–6.
• 144Id at *5–6.
• 145See id at *1.
• 146Id at *5–6.
• 147Isakhanova, 2016 WL 1640649 at *6.
• 148Id, citing Lynch, 465 US at 687 (O’Connor concurring).
• 149See Cherri, 951 F Supp 2d at 933–36; Isakhanova, 2016 WL 1640649 at *5.  See also text accompanying notes 53–55.
• 150See Isakhanova, 2016 WL 1640649 at *5.
• 151Cherri, 951 F Supp 2d at 935.
• 152Isakhanova, 2016 WL 1640649 at *6.
• 153For a discussion on why courts cannot merely rely on this type of “common sense,” see text accompanying notes 167–81.
• 154See text accompanying notes 30–34.
• 155Cherri, 951 F Supp 2d at 926.
• 156See id at 935–36; Isakhanova, 2016 WL 1640649 at *6.
• 157See text accompanying notes 35–38.
• 158Lynch, 465 US at 687.
• 159Cherri, 951 F Supp 2d at 936.
• 160Edwards v Aguillard, 482 US 578, 583 (1987).
• 161Larson, 456 US at 246–47.
• 162See Part I.A.1.
• 163Lynch, 465 US at 692 (O’Connor concurring).
• 164Id at 694 (O’Connor concurring).
• 165See Hassan v City of New York, 804 F3d 277, 306–07 (3d Cir 2015); Schall v Martin, 467 US 253, 264 (1984).
• 166See, for example, Lewis v Ascension Parish School Board, 806 F3d 344, 357 (5th Cir 2015) (noting that “the [Supreme] Court has unequivocally stated that a legislative body’s mere awareness or consideration of racial demographics in drawing district boundaries will not alone trigger strict scrutiny” unless race is the “predominant” motivating factor).
• 167804 F3d 277 (3d Cir 2015).
• 168Id at 306.
• 169Id.
• 170Id (quotation marks omitted).
• 171509 F3d 89 (2d Cir 2007).
• 172See id at 107.
• 173See id at 106.
• 174See id at 99.
• 175Jesse H. Choper, Religion and Race under the Constitution: Similarities and Differences, 79 Cornell L Rev 491, 491–92 (1994) (“There is a powerful resemblance between the government singling out persons for imposition of adverse consequences because of their skin color and because of [ ] their ideological beliefs. This likeness calls for analogous handling under the Constitution.”) (citations omitted). See also generally Mary Anne Case, Lessons for the Future of Affirmative Action from the Past of the Religion Clauses?, 2000 S Ct Rev 325; Ming Hsu Chen, Note, Two Wrongs Make a Right: Hybrid Claims of Discrimination, 79 NYU L Rev 685 (2004). See also Employment Division, Department of Human Resources of Oregon v Smith, 494 US 872, 901 (1990) (O’Connor concurring) (“[T]he First Amendment unequivocally makes freedom of religion, like freedom from race discrimination and freedom of speech, a ‘constitutional nor[m],’ not an ‘anomaly.’”).
• 176City of Richmond v J.A. Croson Co, 488 US 467, 521 (Scalia concurring) (noting that racial classification can also be appropriate when used to remedy past discrimination).
• 177United States v Montero-Camargo, 208 F3d 1122, 1134 (9th Cir 2000), quoting J.A. Croson Co, 488 US at 493.
• 178208 F3d 1122 (9th Cir 2000).
• 179Id at 1135.
• 180Id.
• 181These hypotheticals all concern American citizens, thereby avoiding the different legal framework triggered when noncitizens are involved.
• 182Hassan, 804 F3d at 306–07.
• 183Isakhanova, 2016 WL 1640649 at *6.
• 184See Town of Greece, 134 S Ct at 1824.
• 185Cherri, 951 F Supp 2d at 931–32, citing United States v Flores-Montano, 541 US 149, 152 (2004).
• 186Compare Hassan, 804 F3d at 306, with Tabbaa, 509 F3d at 95.
• 187Cherri, 951 F Supp 2d at 924.
• 188Id at 926.
• 189See ISPU American Muslim Poll Key Findings *1–2 (Institute for Social Policy and Understanding, 2016), archived at http://perma.cc/24X3-H7PN; National Survey of American Muslims Finds Mosques Help Muslims Integrate into American Political Life *1–2 (Muslim American Survey, Mar 8, 2011), archived at http://perma.cc/J4CL-3BFY.
• 190Hassan, 804 F3d at 306.
• 191Montero-Camargo, 208 F3d at 1134.
• 192Id at 1135.
• 193Cherri, 951 F Supp 2d at 926.
• 194See Carrie L. Arnold, Note, Racial Profiling in Immigration Enforcement: State and Local Agreements to Enforce Federal Immigration Law, 49 Ariz L Rev 113, 133–34 (2007). See also Deborah A. Ramirez, Jennifer Hoopes, and Tara Lai Quinlan, Defining Racial Profiling in a Post–September 11 World, 40 Am Crim L Rev 1195, 1209 (2003).
• 195See Grutter v Bollinger, 539 US 306, 351–53 (2003) (Thomas concurring in part and dissenting in part) (clarifying that national security is one of very few bases that constitutes a compelling government interest that can justify racial discrimination).
• 196Capistrano, 615 F Supp 2d at 1146.
• 197See Amjad Mahmood Khan, Religious Freedom as a National Security Imperative: A New Paradigm (Harvard National Security J, Mar 22, 2016), archived at http://perma.cc/E36W-PFKP (discussing how current US government strategies “overemphasiz[e] a terrorism-focused analysis of Islam”). See also Sahar F. Aziz, Sticks and Stones, the Words That Hurt: Entrenched Stereotypes Eight Years after 9/11, 13 NY City L Rev 33, 37–38 (2009).
• 198See International Refugee Assistance Project v Trump, 857 F3d 554, 594 (listing times when President Trump “expressed anti-Muslim sentiment”); Burke, Trump Says US Will Prioritize Christian Refugees (cited in note 2).
• 199See Town of Greece, 134 S Ct at 1824.
• 200See Cherri, 951 F Supp 2d at 924–25.
• 201See Town of Greece, 134 S Ct at 1824.
• 202Hassan, 804 F3d at 306.
• 203See text accompanying notes 171–74.
• 204Tabbaa, 509 F3d at 106–07.
• 205Id at 94.
• 206Id at 92.
• 207Id.
• 208See Section 75. Self-Described Religious Identification of Adult Population (US Census Bureau, 2012), archived at http://perma.cc/XC6V-BMAV. The Census Bureau itself does not inquire into the religion of citizens, but it incorporates the findings of the privately operated ARIS into the Statistical Abstract of the United States. About ARIS (ARIS), archived at http://perma.cc/235L-DEW5.
• 209See Lewis, 806 F3d at 357.
• 210See text accompanying notes 99–102.
• 211See Schall, 467 US at 264 (emphasizing crime prevention as an undoubtedly compelling state interest).
• 212See Lucy Westcott, Newsweek Article on Alleged Hasidic Child Abuse Sparks Brooklyn Yeshiva Protest (Newsweek, Mar 18, 2016), archived at http://perma.cc/2S59-XUL8.
• 213See Nate Carlisle and Brian Maffly, Federal Agents Raid Utah Offices of Polygamous Kingston Group (Salt Lake Trib, Feb 10, 2016), archived at http://perma.cc/E8JT-33Y4.
• 214Id.
• 215For a (colorful) description of the alleged criminal activities of the Kingston Group and its differences from the LDS Church, see Jesse Hyde, Inside “The Order,” One Mormon Cult’s Secret Empire (Rolling Stone, June 15, 2011), archived at http://perma.cc/KE2M-M82D.
• 216Catholic League for Religious and Civil Rights v City & County of San Francisco, 624 F3d 1043, 1049 (9th Cir 2010).
• 217See Jerry Newcombe, Can Your Pro-life Bumper Sticker Actually Get You in Trouble? (Christian Post, Sept 27, 2012), archived at http://perma.cc/2LZQ-KLNH.
• 218Awad, 670 F3d at 1129.
• 219Edwards, 482 US at 586–87.
• 220Lynch, 465 US at 692 (O’Connor concurring). As expounded in Everson, the Establishment Clause prohibits discrimination against religion generally or certain religions specifically. See Everson, 330 US at 15.
• 221See text accompanying notes 107–22.
• 222See text accompanying notes 119–38.
• 223See Lynch, 465 US at 687 n 13; Larson, 456 US at 230, 246.
• 224See Part I.A.1.
• 225See Part I.A.2.
• 226See Part I.A.
• 227530 US 290 (2000).
• 228See id at 314 (“[W]e assess the constitutionality of an enactment by reference to the three factors first articulated in Lemon.”).
• 229See id.
• 230See id at 302–16.
• 231See Van Orden, 545 US at 686.
• 232Id at 685–86.
• 233Id at 691–92.
• 234536 US 639 (2002).
• 235See generally id.
• 236Id at 649.
• 237See id 668–69 (O’Connor concurring).