Thomas B. Nachbar*
The Supreme Court has asserted the power to review the substance of state and federal law for its reasonableness for almost 200 years. Since the mid-1960s, that review has taken the form of the “familiar ‘rational basis’ test,” under which the Court will strike a statute if it is not rationally related to a legitimate governmental interest. The test is hardly perfect. It lacks, for one thing any textual basis in the Constitution. It has been criticized from both ends, as alternatively a judicial usurpation of legislative power or “tantamount to no review at all.” But the Court has applied it for decades, and while the test is not universally loved, neither is it particularly controversial, at least as rules of constitutional law go.
If rational basis scrutiny itself is largely uncontroversial, the same cannot be said for so-called “rational basis with bite,” “rational basis with teeth,” or—as I shall call it—“rational basis plus” review. Rational basis plus is, as Justice O’Connor describes it, a “more searching form of rational basis review.” The Court has never acknowledged its existence, and Justice Scalia explicitly denied it. But lower courts and scholars have repeatedly identified it, noting a sub-set of cases in which the Court purported to apply rational basis scrutiny but in actuality applied something else—even Justice Scalia eventually relented, conceding the Court was applying a different form of review without explicitly elevating scrutiny above rational basis review.
Identifying instances of the rational basis plus test, what triggers it, and what it consists of has been the subject of much academic sport, increasingly so as the Court has applied the test to a series of cases touching on the hot-button issue of sexual orientation, including Romer v. Evans and United States v. Windsor. Such efforts have borne little fruit in the form of increased understanding. A doctrine that the Court does not acknowledge requires neither a justification nor an underlying theory, rendering inquiry into either the equivalent of a constitutional snipe hunt, and about as productive.
We should be deeply suspicious of a doctrine the Court has not acknowledged applying, none more so than rational basis plus. Rational basis plus lends itself to obfuscation as practically no other doctrine can, in part because it purports to be an application of “rationality,” which is a nearly universally appealing concept. Close examination of the case that gave birth to the doctrine—United States Department of Agriculture v. Moreno—shows how easily rational basis plus can be applied disingenuously. In deciding the case, Justice Brennan applied a standard of rationality far exceeding that demanded in an ordinary case. He was able to do so because, although rationality claims to be objective, a claim of irrationality is not objectively falsifiable. Study of the process by which Moreno and its companion case, United States Department of Agriculture v. Murry, were decided, demonstrates not only that rational basis plus can be used to import fundamental rights conceptions through the language of rationality, but also that Moreno itself was decided on exactly that basis. Far from an exercise in rationality, Brennan’s opinion in Moreno was an attempt to justify a result driven by approaches to fundamental rights that were, for one reason or another, unavailable to him as articulable bases for the decision.
Recognizing both the impetus for rational basis plus and its unparalleled suitability to the to the task of justifying results driven by other approaches demonstrates just how truly exceptional and problematic the standard is. Lacking an articulated basis in principle, rational basis plus is impossible to either apply or constrain in a principled way. That is not to say that we should throw the rationality baby out with the bath water. Understanding how Moreno was decided both supplies a framework for understanding how this unspoken doctrine operates—by using rationality in an exclusive rather than an inclusive sense—and provides a guide for conducting rational basis scrutiny without the problematic aspects of rational basis “plus.”
Acknowledging the dangers of rationality review also offers newfound justification for the Court’s oft-maligned “tiered” approach to scrutiny. Although frequently criticized, the tiered approach to scrutiny is valuable for providing exactly the kind of moral and legal accountability that rationality does not. While rationality purports to be objective, the tiers of scrutiny are themselves acknowledged to be contingent—no one thinks that nature or logic requires a particular form of scrutiny for any particular type of legislation. By forcing the Court to choose among the tiers of scrutiny, we force it to provide a justification for its choice—exactly the kind of justification it avoids by relying on rationality to strike statutes that it believes are problematic for other reasons.
The paper proceeds by describing the issues at play in Moreno and Murry before delving into the process by which they were decided. Reference to the Justices’ internal communications, along with Justice Brennan’s notes, demonstrates a set of related concerns about the two cases. Justice Douglas was originally slated to author Moreno and Brennan took over only when Douglas’s chosen approach proved more than the rest of the Court would accept. But Brennan’s first approach to the case—to strike the statute on “morality” grounds—did not fit the case as argued, and only then did he turn to rationality as the basis of the decision. After discussing that shift in justification for Moreno itself, the paper considers the implications of Moreno, and process of its decision, for the Court’s rational basis “plus” jurisprudence.
In 1964, Congress passed the Food Stamp Act as part of Lyndon Johnson’s Great Society Program. Congress laid out the Act’s purposes in the act itself, connecting social welfare with agricultural policy to “safeguard the health and well-being of the Nation’s population and raise levels of nutrition among low-income households . . . promote the distribution in a beneficial manner of our agricultural abundances and  strengthen our agricultural economy, as well as result in more orderly marketing and distribution of food.”
In 1971, Congress amended the Food Stamp Act to restrict food stamp benefits by redefining an eligible “household” as one in which all the residents were related. Several food stamp recipients who would be denied benefits under the new definition sued, including Jacinta Moreno, a 56-year diabetic requiring special food and medical care who lived with Ermina Sanchez, who was, even without caring for Ms. Moreno, poor enough to qualify for both public assistance and food stamps for her and her three children. Under the change, both Moreno and Sanchez (and Sanchez’s children) would be denied assistance because Moreno was unrelated to Sanchez but living in Sanchez’s home.
The Court, following the three-judge district court, rejected any rational relationship to the stated congressional ends, since familial status is irrelevant to both one’s own nutritional needs and one’s ability to stimulate the agricultural economy in satisfying them—the two statutory purposes. At first blush, this seems unexceptional; people almost certainly eat (and likely buy) the same amount of food whether they’re related to their roommates or not. That approach does raise the question of whether each provision of a statute must individually further the entire program’s stated end. As the government argued, welfare programs necessarily entail choices among priorities, necessitating some exclusion. Any provision limiting the scope of the food stamp program would not directly further either of the stated legislative ends; it would do so only indirectly by making possible the parts of the food stamp program that do.
Rational basis scrutiny doesn’t require a rational relationship to the legislative end, though—it requires a relationship to any conceivable end, and so the government offered two ends in its brief, both related to the prevention of abuse of the program: First, the government argued that Congress could rationally conclude that the program was more likely to be subject to moral hazard by non-related cohabitants who chose to “remain voluntarily poor” while living off of food stamps, citing the example mentioned in the legislative history of college fraternities or “other collections of essentially unrelated individuals who voluntarily chose to cohabit and live off food stamps.” Similarly, households of non-related individuals, the government argued, were more likely to have financial support from outside the household, rendering them not really “poor at all.” (College students again come to mind, although the government didn’t argue that.) Second, the government argued that Congress could have concluded that households with unrelated persons in them are “[more] fluid living arrangements having little stability over time.” Such households present challenges to the administration of the food stamp program, since the information the Department of Agriculture required to determine eligibility would be harder to obtain and maintain. It was rational, the government argued, that Congress could respond to the increased cost of “eligibility surveillance” by choosing not to make such households beneficiaries of the program.
Justice Brennan’s majority opinion rejected the rationality of the government’s abuse-based justifications by demonstrating that, because fraud was still possible after the changes, “the classification here in issue is not only ‘imprecise,’ it is wholly without any rational basis,” without explaining at what point imprecision crossed into the realm of the irrational. (The Court did not address the distinct moral hazard argument at all.) Instead, the Court identified another purpose in the legislative history: a statement of intent to prevent “hippies” or “hippie communes” from receiving food stamps, which the District Court had cited and the plaintiffs had identified as the “true purpose of the unrelated household provision.” Having disqualified the stated congressional and proffered purposes, the Court identified animosity to hippies as the sole purpose and invalidated the provision as based on “a bare congressional desire to harm a politically unpopular group,” which “cannot constitute a legitimate governmental interest.” Justice Rehnquist dissented. Not reaching the question of hippies (much less their communes), Justice Rehnquist accepted the imperfection of the statutory classification, but saw in it an attempt to limit food stamps to households that had not been formed for the purpose of receiving them, a purpose rationally served by the statute. Clearly some such households were excluded, which to Justice Rehnquist satisfied the standard even though it also excluded some deserving households and failed to exclude some undeserving ones.
On this, Rehnquist seems to have had the better argument. Every legislative distinction necessarily fails at the margin, and so demonstrating that a statute can be over- or under-inclusive does little to demonstrate its constitutional irrationality. Justices Brennan and Rehnquist seem to have been talking past each other, but the Court has engaged in exactly this debate elsewhere. In New York City Transit Authority v. Beazer, the Transit Authority had adopted a policy against employing narcotics users. That might seem sensible for an organization that operates public buses and trains, but the ban included users of methadone, which is frequently prescribed to those recovering from heroin addiction and does not hinder one’s ability to work when taken orally. The Transit Authority nevertheless banned methadone users because of the likelihood of a potential relapse into heroin or other illegal drug use. Some methadone users are likely to relapse into using heroin or other illegal drugs, but many do not, which made the ban as applicable to all methadone users overbroad. Granting that the safe and efficient operation of the transit system was a legitimate end, the Court concluded that, while the coverage of the methadone was an imperfect way to exclude those who might use illegal drugs, there was a causal connection between methadone use and unemployability, rendering the methadone ban rational.
Justice White dissented, pointing out the many ways that the distinction banning all methadone users (as opposed to only those who had been in treatment for a short time) was necessarily arbitrary, especially at the margins. The majority essentially conceded that point but found it inapposite, since every distinction becomes increasingly arbitrary at the margins. In policymaking, there are few “bright lines,” especially so when a behavior is being regulated not for its own sake but because it increases the risk of another bad outcome. If ineffectiveness at the margins were enough to establish constitutional irrationality, no legislation could survive rational basis review.
The standard applied in Moreno departed from nominal rational basis scrutiny not only by requiring a relationship between means and ends that was more exacting than mere rationality but also by rejecting an alternative proffered legitimate end in favor of crediting only the illegitimate end the plaintiffs had advanced. Alternative justifications are a staple of rational basis cases, and nominal rational basis scrutiny requires that they be handled in favor of upholding legislation.
Beazer again is instructive. To his concerns about the rationality of the statute, Justice White added another: that the actual motivation for the ban was not the safety and efficiency of the transit system but rather an invidious one to discriminate against the kinds of people also likely to be drug users:
Heroin addiction is a special problem of the poor, and the addict population is composed largely of racial minorities that the Court has previously recognized as politically powerless and historical subjects of majoritarian neglect. . . . On the other hand, the afflictions to which petitioners are more sympathetic, such as alcoholism and mental illness, are shared by both white and black, rich and poor.
If Justice White truly thought the statute was motivated by an illegitimate end, such as discrimination against methadone users as a proxy for racial discrimination, why did he bother considering whether the statute actually served its avowed purposes of safety and efficiency? After all, purposeful racial discrimination is an equal protection violation in its own right. The answer lies in the structure of rational basis review, which requires the Court to uphold a statute that rationally serves a legitimate government interest, even one the legislature did not consider.
The analysis in Moreno is exceptional even as compared to Justice Brennan’s own rational basis jurisprudence. Eight years later, Justice Brennan would write an opinion in Minnesota v. Clover Leaf Creamery, in which the Court upheld against both an equal protection and dormant Commerce Clause challenge a Minnesota statute prohibiting the sale of milk in plastic cartons despite evidence in the legislative history that the ban was protectionist because there was also evidence to support other, permissible ends. Given a plausible explanation, “it is not the function of the courts to substitute their evaluation of legislative facts for that of the legislature.”
Thus was born rational basis plus, which on the surface looks to be a more exacting form of rational basis scrutiny. On this reading, the Court was simply drawn to an impermissible end by the plaintiffs, and having gotten a whiff of that impermissible end, the Court used the means-ends structure of rational basis review to exclude alternative ends by demonstrating the lack of a (rational) connection between the means chosen and those other ends. After what was essentially a process of elimination, only a single, illegitimate end remained, and the Court’s holding that animosity to hippies is an illegitimate end is firmly entrenched in equal protection and due process review.
Of course, the doctrine would have been much clearer if the Court had actually held that, should it find an illegitimate end in the legislative history, it would credit that end over legitimate ones. That approach appears to be what the plaintiffs were arguing for, would have resembled the Court’s approach to determining legislative intent in cases calling for heightened scrutiny, and would fit the means-ends structure of rational basis review while requiring a closer relationship than rationality—truly Justice O’Connor’s “more searching form of rational basis review.” The Court did not acknowledge it was changing the standard of review in Moreno, though, and so we are left to wonder exactly what form of review the Court thought it was conducting in Moreno.
Moreno was argued the same day as a companion case, United States Department of Agriculture v. Murry. Brought by the same attorneys as Moreno, Murry was a constitutional attack on another part of the 1971 amendments to the Food Stamp Act, one to § 5(b) of the Food Stamp Act making an entire household ineligible if any adult member of the household was claimed as a dependent for federal income tax purposes by a member of an ineligible household. As in Moreno, a three-judge panel had found § 5(b) unconstitutional and the government appealed.
At the consolidated conference following oral argument, six Justices voted to affirm in Moreno and five in Murry (with two voting to vacate and remand for further findings). Justice Douglas, the senior Justice in both majorities, assigned both cases to himself and started circulating drafts of the combined decision.
Justice Douglas’s approach in his draft Moreno majority opinions is well-reflected in what eventually became his concurrence in the case. In his draft, Justice Douglas conceded the general rationality between the unrelated-persons provisions and the stated legislative purpose of the act, but “as applied here” (in the case of Moreno herself and others like her) it was “wholly unrelated to the Food Stamp Program’s purposes.” In so doing, Justice Douglas shifted from examining the general rationality of the provision to its accuracy as to every individual—to be no broader than necessary to serve the government’s end. What prompted Justice Douglas to apply so exacting a standard was the connection to a fundamental right: Although Dandridge v. Williams had held three years earlier (over dissents by both Douglas and Justice Brennan) that welfare assistance was subject to nominal rational basis scrutiny, Justice Douglas saw the “unrelated persons” provision as implicating “associational rights that lie in the penumbra of the First Amendment,” requiring that the act be “narrowly drawn” to serve its fraud justification or, as he wrote elsewhere, a “compelling governmental interest.” Justice Douglas described this as “the closest scrutiny.” Douglas translated this form of scrutiny into the language of presumptions, following Stanley v. Illinois, decided earlier that term, and planned to strike the provision in Murry as establishing an irrebuttable presumption that the household was not needy based on a tax filing decision made by someone outside the household in a previous year—a violation of procedural due process.
Justice Brennan was skeptical of Douglas’s approach in both cases and wrote Douglas to express his concern that Douglas’s approach would garner a majority in neither case.
For his own part, Brennan would have applied strict scrutiny, ostensibly to Murry but apparently to Moreno as well, “because the challenged provision involves welfare.” His preferred approach being foreclosed by Dandridge, Brennan suggested that Douglas apply rational basis scrutiny to the provision in Murry, arguing that the connection between tax dependency and indigence was entirely irrational, since the existence of tax dependency did not establish the amount of the support received—an individual could logically be both a tax dependent and indigent, a situation “sufficiently common that the statute cannot be said to have a rational basis.” Brennan didn’t specify what frequency of incidence would qualify as “sufficiently common.” As it happens, Douglas did not follow Brennan’s advice and stuck with the presumptions approach in Murry, albeit in watered down form, highlighting the possibility for erroneous applications (for instance, that the tax dependency determination was made a year prior to the food stamp eligibility determination) and weakly claiming at the end of the opinion that the distinction “rests on an irrebuttable presumption often contrary to fact. It therefore lacks critical ingredients of due process.”
Brennan had a different solution for the problems he saw in Moreno, although it wasn’t rational basis or anything like it. Brennan had noticed that Circuit Judge McGowan, writing for the three-judge court below, had struck the statute because the government had offered a “morality” justification. McGowan, after dispensing with the declared statutory ends as not rationally related to the unrelated persons restriction, found himself with only one possible end advanced by the government: the “fostering of morality.”
Rather than declare the morality justification illegitimate, McGowan actually reasoned backward from the morality justification to the conclusion that it could not be attributed to Congress. Because the statute regulated domestic relationships, a morality justification would raise “serious constitutional questions” as implicating both “the rights to privacy . . . in the home” (citing the Court’s then-budding fundamental rights jurisprudence: Griswold v. Connecticut, Stanley v. Georgia, and Eisenstadt v. Baird) and implicating “First Amendment freedoms.” Avoiding conflict between statutes and the Constitution has a venerable history, suggesting Judge McGowan’s avoidance intuition was well-placed. But in an odd feat of constitutional avoidance gymnastics, Judge McGowan had refused to attribute to Congress an intent that would trigger heightened constitutional scrutiny even though the court’s refusal to do so would require it to strike the statute. In short, the court concluded it would be better to attribute no end to the statute than attribute an end that might require heightened scrutiny, essentially killing the patient to cure the disease or, more accurately, killing the patient to avoid conducting a test that might reveal a disease. To Judge McGowan’s association-in-the-home argument, Justice Brennan proposed applying the same heightened-scrutiny-implies-non-attribution-to-Congress approach to the lack of a close fit in the statute, since it “was not narrowly drawn to serve this purpose” of furthering morality. Brennan admitted the non-attribution approach “rests somewhat on a fiction,” but he felt it was of a piece with the approach he’d taken in Eisenstadt v. Baird (in which Brennan’s opinion for the Court had excluded a number of proffered statutory ends for restricting access to birth control) and, more importantly, could “attract a Court.” With these arguments (and after Douglas attempted another draft), Brennan convinced Douglas to give him the Moreno opinion.
Given how intricate a device he’d constructed to convert the government’s morality justification into an liability, one can imagine Justice Brennan’s dismay when he discovered that the government had dropped the morality justification for the unrelated persons provision in the Supreme Court. The change was of no consequence to Brennan, who quickly shifted to what he considered to be an easy rational basis rationale:
As it turned out, however, [the anti-abuse justification] contention was even less convincing than the “morality” argument. Indeed, in practical operation, the statute was not in any sense rationally designed to serve this goal. The opinion was written along these lines and circulated to the conference, with a good deal of confidence.
Justice Brennan’s conclusion that the unrelated persons provision was so clearly a violation of the rationality requirement is in tension with the scholarly consensus that Moreno actually applied something more strict than rational basis review, suggesting that Brennan’s confidence in his analysis was somewhat misplaced, even if it did manage to “attract a Court.”
Moreno started out (or re-booted) as a case that might have launched a completely new approach to inferring congressional intent (don’t if doing so raises constitutional concerns) based on a novel theory of fundamental rights (that providing welfare benefits based on familial status implicates a fundamental right). The one thing Brennan did not originally plan to do was apply the rational basis test, in either vanilla or “plus” form, although rational basis plus scrutiny proved itself more than able to the task of invalidating a statute that he first intended to dispose of on fundamental-rights grounds.
One could write off Moreno as simply an over-enthusiastic but incorrect application of the rational basis test but for the impact the case has had; indeed, Moreno has had more impact as an assertion of judicial authority than insistence on the rational basis test itself. In the years since Moreno was decided, many cases (including several on the cutting edge of constitutional law) have invalidated provisions as unconstitutional applying Moreno’s standard, while the Court has used nominal rational basis scrutiny to strike a statute only once. Moreno has launched its own line of cases clearly applying something other than mere rational basis review. But it is a mistake to view Moreno as simply inaugurating a heightened form of rational basis review; even in the earliest stages of its decision, Moreno was premised not on rational basis review but on a set of far-reaching propositions of both judicial review and substantive constitutional law.
As Justice Brennan’s notes and his memo to Justice Douglas show, Justice Brennan’s opinion in Moreno was rooted in something quite different from a somewhat more rigorous form of rational basis review. Rather, Justice Brennan’s approach to Moreno was driven by two distinct theories, both of them soundly rejected by the Court: The first was an inclination to apply strict scrutiny to welfare legislation, which was rejected outright in Dandridge. The second was similarly rejected in Dandridge, albeit in a different way.
Given Dandridge, Brennan had to accept that food stamp benefits themselves were not subject to heightened scrutiny, but Judge McGowan’s interpretive strategy was itself predicated on the impact of the food-stamp decision on a fundamental right: that of privacy in the home. This was the fundamental right that prompted Judge McGowan to refuse to follow the normal approach in rational basis cases—to rely on any conceivable legislative end that might uphold the provision in question. By following Judge McGowan, Justice Brennan would similarly have imported the Court’s fundamental-right-of-privacy doctrine into the welfare arena, substantially limiting the reach of Dandridge by shifting the focus away from the government’s welfare decision (which was subject to rational basis) and toward the individual’s conduct that triggered the different treatment (which implicated fundamental rights because it took place in the home).
Indeed, when one looks at the reasoning that Justice Brennan had originally planned to base Moreno upon, the differences between Justice Brennan’s thinking and Douglas’s seem to be more in degree than kind. Both Justices thought the statute to be problematic because of its impact on a fundamental right. Justice Douglas thought the rights were “associational rights that lie in the penumbra of the First Amendment,” and although Brennan had objected to Douglas’s reliance on the First Amendment, Judge McGowan had, in addition to the right of privacy, relied on First Amendment rights as the basis for insisting on a closer relationship in the statute before attributing that end to Congress. Thus, while the tool was one of statutory interpretation rather than judicial review, the fundamental-rights structure itself remained, Dandridge notwithstanding. The major advantage of Judge McGowan’s approach, though, was not that it relied on a more well-established set of rights but that it avoided debate over fundamental rights at all by refusing to attribute the (constitutionally problematic) end to Congress and thereby avoiding a test of either the legitimacy of the “morality” end itself or the relationship necessary to uphold a statute intended to serve an end that raised “serious constitutional questions.”
Viewing welfare conditions as regulation of private relations, as Judge McGowan had done and Justice Brennan proposed to do, would have subjected a variety of welfare regulations to heightened scrutiny. Although lacking the color of the occasional reference to hippies, the debate over the 1970 Food Stamp Act amendments was dominated not by the unrelated-persons or tax-dependency provisions (neither of which appear to have been remotely controversial) but rather by an amendment requiring all able-bodied adult members of a household to be willing to accept work lest the entire household lose food stamp benefits. When viewed as a regulation of the private right of familial association, the work requirement could have been seen as forcing a choice for parents between work or abandoning their children (a point that Senator McGovern made in the debates over the food stamp amendments act), potentially implicating the very same associational and family rights that both Justices Brennan and Douglas had viewed as being at issue in Moreno. The “fundamental rights equal protection” doctrine exemplified by Shapiro v. Thompson that the Court rejected in Dandridge had relied on exactly this connection between welfare and fundamental rights: to describe the limitation of nominal right to welfare payments against a fundamental right (in Shapiro, the right to travel among the States) and rely on the impact on the fundamental right to trigger heightened scrutiny. Justice Brennan (and Justice Douglas) viewed the unrelated persons provision as a limitation not on the nominal right to food stamps but on the fundamental right of association. Justice Douglas had taken the same approach in his Dandridge dissent. The parallels between Brennan’s original approach in Moreno and Douglas’s Dandridge dissent are even closer when one considers that Justice Douglas’s Dandridge dissent was grounded not on constitutional, equal protection grounds, but on statutory interpretation. The restriction at issue in Dandridge was Maryland’s imposition of an absolute cap on benefits regardless of family size in implementing a federal welfare program. Justice Douglas had (as had Justice Marshall in a dissent joined by Justice Brennan) argued in Dandridge that Maryland’s restriction was invalid under the federal welfare statute because the incentive it provided to “break up large families” failed to further the congressional purpose of the act, much as Justice Brennan would have questioned whether Congress had intended to regulate morality with the food stamp program. Although couched in terms of statutory interpretation, the implications of Justice Brennan’s original approach were both constitutional and far-reaching, even more so than Justice Douglas’s dissent in Dandridge because of the constitutional justification for the narrow construction Judge McGowan and Justice Brennan would give the statute.
In this way, Justice Brennan’s original plan for Moreno could have resulted in a ground-breaking shift in constitutional law, essentially resurrecting the heightened scrutiny of the fundamental rights equal protection line of cases through a new version of the well-accepted avoidance canon of statutory interpretation. On the other hand, it’s possible that Justice Brennan would have been no more successful than he anticipated Justice Douglas was going to be. The fiction, and its connection to fundamental rights, would have been obvious to all. Eisenstadt v. Baird—the model for Brennan’s approach to Moreno—was decided by a seven-Justice court, with Rehnquist (who dissented in Moreno) and Justice Powell not participating. Justice Brennan’s majority in Eisenstadt attracted only four votes (including his own), with a strong dissent from Chief Justice Burger, and a concurrence by Justice White joined by Justice Blackmun to distinguish the statute from economic legislation. It is doubtful Brennan would have received any of these votes to extend the Eisenstadt approach to what was clearly an economic regulation in Moreno. (Indeed, Blackmun had originally voted to overturn the district court in Moreno.) We will never know what would have happened had Brennan pursued his fundamental-rights approach to the avoidance canon simply because the government failed to make the morality argument in the Supreme Court.
Although the realized Moreno did not live up to Brennan’s idealized version, there are several lessons to be taken from the way in which it was decided.
First, the process of deciding Moreno demonstrates the shaky intellectual ground on which rational basis plus scrutiny stands. Both Brennan’s notes and his memo to Justice Douglas suggest that Brennan had decided for reasons unrelated to rational basis scrutiny that the statute should be struck and landed on rational basis as the means to do so only when his preferred reasoning was no longer supported by the facts (or rather, by the government’s justification). One might object to that claim by arguing that it is equally likely that Brennan independently thought that the statute failed the deferential form of the rational basis test, but virtually no one has read Moreno as embodying the deferential form of the rational basis test in the decades since its decision. The credibility of a claim that Brennan simply shifted from one basis of his decision to another equally applicable one depends on his fidelity to the rational basis test. Moreno’s status as the standard for rational basis plus scrutiny combined with Brennan’s original attempt to apply a much stricter form of review undermines any claim that Brennan was also convinced that the provision failed the deferential form of rational basis review. Brennan’s shift to the rational-basis justification in Moreno was at best an obfuscation of his real basis for striking the statute and at worst a disingenuous ploy to re-purpose rational basis scrutiny to avoid a fight over fundamental rights while following a fundamental-rights approach to resolving the case: either his view that welfare regulation should be subject to strict scrutiny or his view, shared in slightly different form by Justice Douglas, that residential associational rights are so fundamental as to affect the standard of review applicable to economic or social legislation touching on the home (or both).
Second, identifying the connection between Justice Brennan’s eventual Moreno opinion and its methodological origins in Judge McGowan’s opinion below demonstrates the ways in which the Moreno approach is not really rational basis “plus” at all but an altogether different form of scrutiny. In a typical rational basis case, the Court will use rationality inclusively—to identify potential ends in hope of finding one that supports the statute. In Moreno, Justice Brennan used rationality exclusively—to reject the government’s proffered ends, eventually landing on one that the Court considered illegitimate.
The distinction between inclusive and exclusive rationality is presented by the different approaches to ends taken by the majority and dissent in Beazer. Justice White had no direct evidence of an illegitimate motive, and so in order to land upon the one he eventually found, rational basis scrutiny required him to exclude all the other (legitimate) ends potentially served by the policy, which he did by showing that the policy did not rationally serve those alternative ends. Thus, Justice White employed rationality in a different but related way to that of the majority. The majority used rationality to demonstrate a causal connection between the provision and a legitimate governmental interest—to include legitimate ends as within the ambit of the provision. Justice White used rationality to exclude potential ends, eventually finding only an illegitimate one remaining.
It is tempting to allow evidence of illegitimate ends to alter the rationality inquiry, but the two are distinct. Beazer presents a more pristine example of exclusive rationality because there was no evidence in the record to support the illegitimate end Justice White eventually landed upon, while in Moreno, the legislative history provided it. But the possibility that the regulation can rationally serve some illegitimate ends does not change whether it can rationally serve other, legitimate ones. In Clover Leaf Creamery, for example, the Court had before it both legitimate (environmental) and arguably illegitimate (protectionist) ends, and Justice Brennan used rationality inclusively—to identify a permissible end to which the statute could be rationally related. It may be that the presence of illegitimate ends should warrant heightened scrutiny, but it does not change the nature (or existence) of rationality itself.
The two inquiries—using rationality to include ends or using rationality to exclude them—might seem like logical complements, but they are not because of the low standard that mere rationality requires. The sort of instrumental rationality employed by the Court is defined by a causal relationship between means and ends, but neither causation nor rationality includes a self-defining description of how close that relationship has to be to qualify as “rational,” what statisticians would refer to as an “effect size.” Rationality describes a relationship between inferences and a conclusion, and the point of the rational basis test is to require the state to articulate that the necessary relationship can plausibly exist, not to demonstrate that it exists to any particular degree.
The lack of reciprocity between the inclusive and exclusive uses of rationality becomes clear when one considers a third lesson one can draw from how Moreno was decided: that while rationality can effectively constrain legislative discretion, it cannot effectively constrain judicial discretion. The lack of an accepted effect size in order for a particular causal relationship to qualify as rational is not a problem when including potential ends because the Court is only looking for the existence of the relationship, not its strength. When using rationality to exclude ends, though, the Court requires a threshold below which it might find a causal relationship but not rationality, and it is the Court itself that determines how large an effect size is required in order to find a relationship rational. Thus, Justice Brennan did not make the strong claim that the unrelated persons provision would stop no abuse (for surely it would stop some), just that it would not stop enough in order to qualify as rational.
Like Justice White did in his Beazer dissent, Justice Brennan used rationality to exclude potential legitimate governmental ends from consideration. Justice Brennan had an even stronger case for the illegitimacy of the statute because, unlike in Beazer, there actually was direct evidence of an illegitimate legislative motive in the legislative history: statements that the provision was intended to discriminate against “hippies” and “hippie communes.” But because rationality review will uphold a statute if it is supported by any legitimate governmental interest (unless the illegitimate one is one that leads to heightened scrutiny, which discrimination against hippies did not), it was necessary for Justice Brennan to not only identify the illegitimate governmental end he did but also to eliminate the other, legitimate ends advanced by the government. He was able to do so only because of the lack of an agreed upon description of the minimum rationality required to uphold a statute; once Justice Brennan insisted on more than a plausible description of the causal relationship to establish the rationality of the provision, his claim that the statute was not rational enough to qualify as rational became nonfalsifiable. By decrying an inadequate quantity of something that no one was prepared to quantify, the Court’s discretion became unbounded by the constraints of rationality.
It is possible to restate the rational causal relationships mathematically by describing the relationship between accurate and inaccurate classifications, but even then disagreement as to application swamps mathematical comparison. In Craig v. Boren, for example, the majority found the statute (a differential drinking age for men and women) to fail the heightened standard of substantial relationship to an important governmental objective that is applied to sex-based classifications. Oklahoma offered a variety of statistical arguments, which the Court rejected:
Viewed in terms of the correlation between sex and the actual activity that Oklahoma seeks to regulate—driving while under the influence of alcohol—the statistics broadly establish that .18% of females and 2% of males in that age group were arrested for that offense. While such a disparity is not trivial in a statistical sense, it hardly can form the basis for employment of a gender line as a classifying device. Certainly if maleness is to serve as a proxy for drinking and driving, a correlation of 2% must be considered an unduly tenuous “fit.”
The 2% number represents a comparison between all males age 18-20 (the class actually affected by the restriction) and males who drove drunk (the class whose behavior the statute attempted to modify). The dissent, on the other hand, did not see the relevant comparison as between the number of men affected and the men who drove drunk but between the number of men who drove drunk (2%) and the number of women who drove drunk (only .18%), because the statute discriminated on the basis of sex. If evaluated by that distinction, the relevant number is not the .02 likelihood of a man 18-20 driving drunk but that men are eleven times more likely to drive drunk than similarly aged women, an effect size likely large enough to convince even the most skeptical.
At issue in such cases is not the mathematics of probability and rationality but a normative question about how to define the affected and targeted classes. Given that underlying dispute, it is perhaps fortunate that the Court in Craig eschewed the false determinacy of statistics in applying its review.
Of course, the requirement to produce statistical evidence to support a statute’s application goes far beyond both the demands of rational basis scrutiny and the likely capacity of even the federal government were it put to such a test in the potentially limitless number of rational basis cases it could face. Imagine the Department of Agriculture being required to produce statistical evidence about the relative number of deserving and abusing food stamp recipients as a percentage of those who live in households with unrelated persons in order to demonstrate the statute’s rationality. The administrative cost of detecting abusers was itself the government’s justification for the provision; it would approach irony to require it to do so in order to defend the constitutionality of the statute at the insistence of even a single objector willing to litigate. As Justice Stevens wrote in Beazer, irrationality is going to exist at the margins of every regulation, since every regulation is over- and under-inclusive at the margin. The rational basis test is designed with this in mind: If the government can establish a reasoned causal relationship between the means and a legitimate end, that should be the end of the inquiry.
The lack of a framework for applying rationality in its exclusive rather than inclusive sense opens the door to sophistry in both its malicious and innocent senses. A requirement of not just a reasoned causal relationship but some minimum effect size combined with both the conceptual and practical difficulty of quantifying effects in rational basis cases allows Justices who wish to strike statutes for other reasons to make non-falsifiable claims that a particular provision fails to demonstrate “enough” rationality to be rational. Comparing Justice Brennan’s majority with Justice Rehnquist’s dissent is an exercise in frustration largely because the two opinions are talking past each other. Rehnquist argued the relationship between the unrelated persons provision and the prevention of abuse, and Brennan, even while assuming the existence of a causal relationship between unrelated persons and abuse, did not think the relationship rose to the level of rationality. There is not only no way to know who was right, the lack of an accepted principle for deciding what constitutes the relevant effect size in rational basis cases means that there is no way to know how we’d decide who was right.
The fourth lesson of Moreno is that the lack of any inherent principle for applying rationality in its exclusive sense makes rational basis plus review effectively insurmountable. Indeed, the way scholars identify cases as being rational basis “plus” cases is by noting that the Court claims to apply rational basis review but strikes the statute. Justice O’Connor’s description of rational basis plus scrutiny in Lawrence v. Texas tellingly continues “to strike down such laws” not “to evaluate such laws.” Not even strict scrutiny is so consistently fatal. Rational basis plus allows the Court to strike statutes without actually acknowledging that it is applying a higher standard, avoiding criticism of its choice to do so. In Adarand Constructors v. Pena, concerning the application of strict scrutiny to racial classifications, Justice O’Connor only knew to defend strict scrutiny from charges of being “strict in theory but fatal in fact,” because of her willingness to acknowledge that the Court was in fact applying strict scrutiny. No such defense is necessary for the heightened scrutiny of rational basis plus exactly because the Court does not feel compelled to acknowledge its existence, leaving litigants with little opportunity to argue how the statute in question satisfies the standard. It is difficult to imagine a government brief citing Moreno as part of an argument that that provision satisfies the standard applied in that case. Rational basis plus is a one-way street toward constitutional invalidation.
I am not suggesting that other forms of review are necessarily more deterministic than rationality review, after all there is no machine for measuring whether a governmental interest rises to the level of being “compelling.” In cases requiring an “important” or “compelling” governmental interest, the Court is not identifying such interests as though they exist in nature, it is defining them. When the Court explains that promoting diversity in higher education is a compelling governmental interest, it is not calling upon a concept of “compelling” as recognized in broader thought; it is making a claim that the Court (as opposed to some outside authority) believes this interest is important enough to support race-based classification and the Court necessarily takes responsibility for making that claim. The same is not true of rationality review, in which the Court is invoking the concept of rationality—a concept whose primary meaning is exogenous to law—to justify its role in conducting constitutional review.
Fifth, the pliability of rational basis plus demonstrates that that the real work the Court does in equal protection cases is in choosing a level of scrutiny for a particular form of discrimination, not in applying the standard of review that level of scrutiny demands. The question of whether we should view race-based classifications with greater suspicion than we view other classifications is, unlike the question of whether denying food stamps to households with unrelated persons will actually curb food stamp abuse, a question of constitutional dimensions. The Court instructs that the Constitution is more concerned about race than it is about practically any other form of discrimination, and in telling us that, the Court has said something important about the Constitution. In order to make such claims, the Court needs to speak with clarity, which it cannot do when it makes largely unsupported (and generally unsupportable) claims that are tied to the efficacy of a particular statute, as it does in applying rational basis review.
Rational basis plus scrutiny should stand as Exhibit 1 in the case for retaining and building upon the Court’s tiered approach to equal protection scrutiny. Each equal protection case the Court confronts currently requires it to resolve two issues: what level of scrutiny to apply, which tells us about how constitutionally sensitive the classification is, and whether the provision in question actually satisfies the chosen level of scrutiny, which tells us very little. Suggestions that the Court abandon the tiers of scrutiny, as Justice Marshall famously did and some academics have, would result in even less clarity and even more intractable arguments, as every inquiry would devolve into a combined inquiry that produces only one determination: whether the specific statute survives. Although the answer to that question is important to the litigants in a particular case, it tells us very little about the Constitution.
The question we should be asking ourselves after Moreno is not whether the unrelated person restriction was rationally related to reducing abuse of the program (it clearly was) but rather is how Justice Brennan could so easily strike a statute that he was convinced implicated fundamental rights—in two separate ways—without actually applying a different standard of review. The answer lies in the rational basis test itself, which claims legitimacy by virtue of its connection to the normatively neutral concept of rationality but provides practically no restraint on judicial discretion.
Even at its best, the use of rationality review to exclude ends lends itself to intractable disagreements over whether a means is adequately proximate to the end it serves. At its worst, rationality used as part of an exclusive rather than inclusive inquiry into legislative ends can too readily serve as cover for outcomes driven by other justifications, as appears to have happened in Moreno itself. My complaint is not that the Court is being disingenuous when it claims to apply rationality review but actually applies something stricter, it is that rationality review particularly lends itself to such misuse, intentional or otherwise, and that it is possible to counter this potential misuse by insisting that rationality be used only in its inclusive rather than exclusive sense.
Although one might at first blush see my proposal—to prevent the use of rationality to exclude proffered ends as not rationally related to means—as an extreme modification of rational basis review that guts it of any force, it is actually fairly modest and unlikely to have much effect on current practice. My proposal affects only a particular use of one part of the test. The test remains available to strike statutes in singular pursuit of an illegitimate end, as occurred in Zobel v. Williams, striking Alaska’s retrospective distribution of oil dividends, or in cases where the means is not rationally related to a stipulated end, as was the case in Allegheny Pittsburgh Coal Co. v. County Commission of Webster County, in which the Court struck a taxation scheme as irrational because the state had limited itself to a particular means of assessing land value. Moreover, my modification only limits the use of rationality as an interpretive method. The Court would still have available to it other interpretive tools to connect means to ends, much as it does in cases calling for heightened scrutiny. That is actually what the Court has been doing for decades. Cases striking statutes while applying the most deferential form of rational basis review are practically unheard of, and yet the legal system endures.
In practice but not word, the Court has done exactly what I propose it do: elevate the scrutiny when it believes doing so is justified for some reason exogenous to the rational basis test itself. The only implication under my proposal is that, by limiting the use of rationality to strike a statute, the Court must acknowledge it is elevating scrutiny in the cases in which it currently does sub silentio. Escalating scrutiny, in turn, will prompt the Court to supply a justification for doing so. In such justification lies the best hope for the Court to develop the constitutional law of equal protection.
* Professor of Law, University of Virginia School of Law. I would like to thank Charles Barzun, Jill Hasday, Debbie Hellman, Greg Mitchell, David Strauss, and Geoff Stone for helpful comments and suggestions. I am also indebted to Jennifer Talbert for excellent research assistance.
. See, e.g., Beach Comm’ns, 508 U.S. at 323 n.3 (Stevens, J., concurring). Kenji Yoshino, The New Equal Protection, 124 Harv. L. Rev. 747, 760 (2011) (describing rational basis review as a “free pass”).
. Among scholars, the preferred term appears to be “rational basis with bite,” garnering 501 hits in the Westlaw JLR database, well ahead of either “rational basis with teeth” with 98 hits and “rational basis plus” with only 76 (with some overlap among them). None of the terms are popular with the Court, with zero hits for any of the three terms in the slightly more influential Westlaw SCT database. I prefer “rational basis plus,” both out of a general aversion to dental metaphors and because it avoids any potential confusion in rational basis cases actually involving teeth. See, e.g., Douglas v. Noble, 261 U.S. 165 (1923) (upholding state licensing restrictions on dentists against due process challenge).
. Attention to rational basis plus started shortly after Moreno was decided, see Gary J. Simson, A Method for Analyzing Discriminatory Effects Under the Equal Protection Clause, 29 Stan. L. Rev. 663, 675-77 (1977), and continues to this day, see Raphael Holoszyc-Pimentel, Reconciling Rational Basis Review: When Does Rational Basis “Bite”?, 90 N.Y.U.L. Rev. 2070 (2015).
. 413 U.S. 528 (1973). See, e.g., Yoshino, supra note 6, at 760 (citing Moreno for “rational basis with bite”); Russel K. Robinson, Unequal Protection, 68 Stan. L. Rev. 151, 165 (2016) (citing Moreno as the first of one of “three key cases” developing “heightened scrutiny” rational basis analysis); Susannah W. Pollvogt, Unconstitutional Animus, 81 Fordham L. Rev. 887, 903 (“Moreno set the pattern for the one-two punch of animus analysis.”). Most importantly, this is the point that the Court itself has identified, see supra notes 14–15, even without admitting that rational basis plus exists. In situating rational basis plus on Moreno, I am excluding some earlier cases in categories that were ostensibly reviewed under rational basis but later employed standards of heightened scrutiny, such as in the case of sex. Cf. Reed v. Reed, 404 U.S. 71 (1971). The Court having later announced a heightened scrutiny standard in such cases, it is reasonable to conclude that it was applying something more than the most permissive rational basis scrutiny in those earlier cases even if relying on the language of rationality. In his comprehensive survey of rational basis plus cases, Raphael Holoszyc-Pimentel identifies six cases applying rational basis plus before Moreno. Holoszyc-Pimentel, supra note 12, at 2106–10. One of those cases is Reed, which concerns sex discrimination, one was Eisenstadt v. Baird, which addressed access to contraceptives, one addressed illegitimacy (which like sex the Court later ruled was subject to intermediate scrutiny, see Clark v. Jeter, 486 U.S. 456, 461 (1988)), one upheld the statute under rational basis review but struck it as providing unequal access to courts, and two related to criminal procedure, an area not generally subject to rational basis review. See also Robert C. Farrell, Successful Rational Basis Cases in the Supreme Court from the 1971 Term Through Romer v. Evans, 32 Ind. L. Rev. 357, 361–70 (1999) (collecting seven cases from the 1971 term evaluated by Gerald Gunther and explaining their relationship to later heighten scrutiny review).
. See Amendments to the Food Stamp Act of 1961, 116 Cong. Rec. 42021 (statement of Rep. Belcher) (“If you really want to kill the food stamp program, just jack it up so high that the taxpayers will completely revolt. Keep all of these gadgets in the bill, keep the students, the hippies, the strikers, and everybody else, enabling all of them to get on the food stamp plan, and it will not take very long for those people who want to kill the food stamp plan to get the job done.”).
. Id. at 534 (“The legislative history that does exist, however, indicates that that amendment was intended to prevent so called ‘hippies’ and ‘hippie communes’ from participating in the food stamp program.”) (quoting H.R. Conf. Rep. No. 91-1793, p. 8; 116 Cong. Rec. 44439 (1970) (statement of Sen. Holland)). But see id. at 543 (Douglas, J., concurring) (citing “hippy communes” rather than “hippie communes”).
. See id. (“[T]he uncertainties associated with the rehabilitation of heroin addicts preclude [the District Court] from identifying any bright line marking the point at which the risk of regression ends. By contrast, the ‘no drugs’ policy now enforced by TA is supported by the legitimate inference that as long as a treatment program (or other drug use) continues, a degree of uncertainty persists.”).
. See Washington v. Davis, 426 U.S. 229, 239 (1976) (“The central purpose of the Equal Protection Clause of the Fourteenth Amendment is the prevention of official conduct discriminating on the basis of race.”). See also Kim Forde-Mazrui, Traditional Justification: The Case of Opposite-Sex Marriage, 78 U. Chi. L. Rev. 281, 308 (2011) (Regulation “must not be tainted by illegitimate purposes, beliefs, or assumptions. An interest is tainted when the reasoning or motivation leading a state to pursue an ostensibly legitimate interest includes an illegitimate assumption or belief, such as an irrational fear or impermissible stereotype.”).
. FCC v. Beach Comm’ns, Inc., 508 U.S. 307, 313–14 (1993); Yoshino, supra note 6, at 760 (“In other words, even if the legislature had provided no rationale or an inadequate rationale, the state action would be upheld so long as the Court could supply one. Because judges could imagine many things, ordinary rational basis review was tantamount to a free pass for legislation.”).
. Minnesota v. Clover Leaf Creamery Co., 449 U.S. 456, 471, n.7 (1980) (dismissing the protectionist statements in the legislative history as an “economic defense of an Act genuinely proposed for environmental reasons”).
. Id. at 470. As Justice Brennan himself wrote several years before Moreno, if anything, dormant Commerce Clause analysis suggested a more demanding standard than mere rational basis. See Fl. Lime and Avocado Growers, Inc. v. Paul, 373 U.S. 132, 154 (1963) (“Other state regulations raising similar problems have been found to be discriminatory or burdensome notwithstanding a legitimate state interest.”).
. Draft Opinion dated May 3, 1973, at 4, U.S. Dep’t of Agric. v. Moreno, No. 72-534 (on file with the Library of Congress) (William J. Brennan Papers, box I:302, folder 10) [hereinafter Douglas First Draft].
. Judge McGowan explained that the court’s refusal to attribute the “morality” end to Congress would not change the outcome, because if the court did so, the statute could not survive the resulting scrutiny. In a portion of the case more closely reflecting the mores of the time than even Judge McGowan understood, he explained that in order to save the statute under those circumstances, it would be necessary to read into the statute a classification limiting the provision to households “of both sexes as distinct from all other households,” apparently in the belief that only households containing members of both sexes could embody the types of living arrangements that would need to be discouraged in order to foster morality. See Moreno, 345 F. Supp. at 315.
. Draft Opinion dated May 3, 1973, at 4, U.S. Dep’t of Agric. v. Moreno, No. 72-534 (on file with the Library of Congress) (William J. Brennan Papers, box I:302, folder 10); 73; Memo from Justice Brennan to Justice Douglas (May 17, 1973) (William J. Brennan Papers, box I:302, folder 10).
. Nor, for that matter, did Brennan plan to base the case on what has become the most durable part of the case actually acknowledged by the Court as a rule: that a bare desire to harm group cannot be a legitimate governmental interest.
. The case is Allegheny Pittsburgh Coal Co. v. County Comm’n of Webster County, 488 U.S. 336, 338 (1989). In Allegheny Pittsburgh Coal, the state had limited itself (in the state constitution) to assessing land based on its present value. The Court found that the local taxing authority had impermissibly assessed some land based on its current value while assessing some on historical value. One could arguably characterize Murry itself as a second rational-basis case, although the Court has largely treated it as addressing the use of irrebuttable presumptions. See, e.g., Levine v. Milne, 424 U.S. 577, 584 n.9 (1976) (“Since nothing is conclusively presumed against the applicant, who is clearly required to prove his eligibility if he is to receive relief, this Court’s prior cases dealing with so-called irrebuttable presumptions [including Murry] are not in point.”). See generally Nachbar, supra note 1.
. Farrell, supra note 17, at 358 (Rational basis plus creates “two sets of rationality cases, one deferential and one heightened, operating as if in parallel universes with no connection between them.”).
. Id. at 634 (“The waiting-period provision denies welfare benefits to otherwise eligible applicants solely because they have recently moved into the jurisdiction. But in moving from State to State or to the District of Columbia appellees were exercising a constitutional right, and any classification which serves to penalize the exercise of that right, unless shown to be necessary to promote a compelling governmental interest, is unconstitutional.”).
. Id. at 502 (Douglas, J., dissenting) (“The District Court correctly states that this incentive to break up family units created by the maximum grant regulation is in conflict with a fundamental purpose of the Act.”); id. at 513–14 (Marshall, J., dissenting).
. Cf. Shelby Cty. v. Holder, 133 S. Ct. 2612, 2631 (2013), in which the Court avoided confronting the question of whether legislation passed pursuant to Section 2 of the Fifteenth Amendment is subject to the same, heightened, congruence and proportionality test applied to legislation passed pursuant to Section 5 of the Fourteenth Amendment by showing that the statute did not satisfy what it purported to be rational basis scrutiny.
. Such a move was lost on no one, except perhaps the Court itself. Gerald Gunther pointed out a similar use of means-ends scrutiny in Reed, Eisenstadt, and Griswold the Term before Moreno was decided. See Gerald Gunther, Foreword: In Search of Evolving Doctrine on a Changing Court: A Model for a Newer Equal Protection, 86 Harv. L. Rev. 1, 30 (1972) (“The resort to means-oriented scrutiny in all these cases is at least partly attributable to its attractiveness as an avoidance device.”).
. See Barbara G. Tabachnick and Linda S. Fidell, Using Multivariate Statistics 199 (5th ed. 2006). In statistics, what is measured is not causation but correlation, which has to be interpreted in order to make causative claims, but the relationship to effect size is the same. Id. at 596.
. See John Ladd, The Place of Practical Reason in Judicial Decision, in NOMOS VII: Rational Decision 127–28 (Carl J. Friedrick ed. 1964) (“By a ‘rational decision’ I mean a decision for which the agent can give good reasons.”).
. Id. at 204 (“[P]roving broad sociological propositions by statistics is a dubious business, and one that inevitably is in tension with the normative philosophy that underlies the Equal Protection Clause.”).
. 2 The Works of Samuel Johnson in Nine Volumes (No. 31) (Oxford 1825) (“[M]en who cannot deceive others, are very often successful in deceiving themselves; they weave their sophistry till their own reason is entangled, and repeat their positions till they are credited by themselves; by often contending, they grow sincere in the cause; and by long wishing for demonstrative arguments, they at last bring themselves to fancy that they have found them.”).
. Note that by definition I am excluding cases in which the Court acknowledges that it is applying heightened scrutiny, including cases that were previously decided applying some form of a “rational basis” standard but which the Court later acknowledged would be decided under heightened scrutiny, as in the case of sex. See supra note 17.
. Lawrence v. Texas, 539 U.S. 558, 580 (2003) (O’Connor, J., concurring in the judgment) (“When a law exhibits such a desire to harm a politically unpopular group, we have applied a more searching form of rational basis review to strike down such laws under the Equal Protection Clause.”).
. San Antonio Indep. School Dist. v. Rodriguez, 411 U.S. 1, 98–99 (1973) (Marshall, J., dissenting) (“A principled reading of what this Court has done reveals that it has applied a spectrum of standards in reviewing discrimination allegedly violative of the Equal Protection Clause.”). Cf. id. at 59 (Stewart, J., concurring) (“I join the opinion and judgment of the Court because I am convinced that any other course would mark an extraordinary departure from principled adjudication under the Equal Protection Clause of the Fourteenth Amendment. The unchartered directions of such a departure are suggested, I think, by the imaginative dissenting opinion my Brother MARSHALL has filed today.”).