Ambiguity and the First Amendment:
Some Thoughts on All White Advertising
By Michael E. Rosman
Tennessee Law Review, October 29, 1993
Nonetheless, in practice at least, courts appear tougher on conclusions of the Postal Service than those of the FTC as to the effect an advertisement has on the ordinary reader. … Determining what the ordinary reader perceives — not to mention how the court in Hunter divined it — is far less clear. … On appeal, the Second Circuit, in an opinion by Judge Winter, specifically upheld the ordinary reader standard. … The suggestion of the Second Circuit in New York Times that the relevant audience for the discrimination statutes is the ordinary reader of a particular race (as opposed to Judge Sweet’s alternative determination that the standard is the ordinary reader regardless of race) deserves some consideration. … The ordinary reader standard (without reference to subgroups) may leave you a bit cold with these hypotheticals, since the ordinary reader simply does not comprehend the discriminatory content of the message. … First, the relevant audience should not be the ordinary reader but rather the “ordinary potential buyer.” … But then the hypothetical integrated human model advertisements might be false advertising — particularly for our friend the gullible, slow-witted ordinary reader, upon whom the subtle distinctions between what is and what ought to be may be lost. … This may have some implications for the “ordinary reader v. the ordinary potential purchaser” issue. …
This Article examines ambiguous commercial speech as it has been interpreted under various federal statutes that affect the world of advertising. Of particular concern will be recent cases decided under § 3604(c) of the Fair Housing Act. n1 These cases hold that the exclusive use of Caucasian human models in advertisements for the sale or rental of real estate violates § 3604(c) of the Fair Housing Act by sending a message that minorities are not wanted. n2 I then consider recent Supreme Court jurisprudence relating to the protection of commercial speech under the First Amendment and the potential effects it may have on future court decisions requiring the interpretation of advertisements.
Previous academic works concerning the human model cases, written almost exclusively by students, were published in two stages. The first articles were published in 1988, and focused on the question of whether intent was an element of a § 3604(c) claim. n3 That question now seems to [*291] be settled, and the statutory construction espoused by those articles will likely prevail. n4 The second set of student notes, all published in 1992, summarized the subsequent case authorities. n5 However, none of the articles in either stage focused attention on the message being conveyed by all-white advertising. Indeed, the previous works generally assumed without discussion that an advertising campaign that employed human models, all or almost all of whom were white, sent a discriminatory message. n6 This [*292] Article challenges that presumption (or, at least, challenges its standing as a presumption) and asks what kind of proof should be required to show that an advertising campaign using primarily white human models sends a discriminatory message. It does this by exploring analogous statutes and reviewing the type of proof each requires for determining the meaning of advertisements where the alleged representation is not explicit, but is said to be implied in the advertisements.
I also suggest an argument that the First Amendment provides certain protections for defendants exposed to civil liability for commercial speech, and that those protections should apply regardless of the statute in question. Specifically, I hope to advance the argument that the First Amendment may preclude the imposition of damages for ambiguous commercial speech unless: (1) proof exists through some form of empirical analysis that the speech is interpreted by the “relevant reader” — the definition of which is an issue unto itself — to have a meaning that places it beyond the protective carapace of the First Amendment, and (2) the defendant acted with some type of scienter with respect to the effect that the advertisement would have on the relevant reader.
The statutes I will examine fall into two broad categories: (1) those seeking to prevent or punish fraud (The Fraud Statutes), and (2) those seeking to prevent or punish discrimination based upon certain impermissible characteristics (The Discrimination Statutes). In examining each of the statutes that affects advertising, the analysis will focus on the following questions: Who is the relevant reader? How is the relevant reader’s interpretation of the advertisement determined? Finally, does it matter if the relevant reader’s interpretation was inadvertent or unintentional, meaning that the advertiser did not intend the relevant reader to so interpret the advertisement? It is with these questions in mind that I will examine the Supreme Court’s pronouncements on ambiguous commercial speech.
I. The Fraud Statutes
The Federal Trade Commission Act (FTC Act), n7 the Postal Service Act, n8 and the Lanham Act n9 each seek to prevent false or misleading advertising.
[*293] A. The Federal Trade Commission Act
The FTC Act is an omnibus, multifaceted statute regarding trade and commerce. n10 Section 5(a)(1) of the FTC Act states that “[u]nfair methods of competition in or affecting commerce, and unfair or deceptive acts or practices in or affecting commerce, are declared unlawful.” n11 Section 12 of the FTC Act adds that it shall be “unlawful . . . to disseminate, or cause to be disseminated, any false advertisement” n12 and that such dissemination constitutes an “unfair or deceptive act or practice” under section 5(a)(1). n13 Finally, section 15(a)(1) of the FTC Act defines “false advertisement” for purposes of section 12 as:
[A]n advertisement, other than labeling, which is misleading in a material respect; and in determining whether any advertisement is misleading, there shall be taken into account (among other things) not only representations made or suggested by statement, word, design, device, sound, or any combination thereof, but also the extent to which the advertisement fails to reveal facts material in the light of such representations or material with respect to consequences which may result from the use of the commodity to which the advertisement relates under the conditions prescribed in said advertisement, or under such conditions as are customary or usual. n14
1. Defining Deception
Courts have defined a “deceptive advertisement” under the FTC Act as an advertisement with a “tendency or capacity to deceive.” n15 Legal literature [*294] and court opinions have addressed this standard and competing standards, n16 and thus no attempt will be made in this Article to add to those efforts. n17 The following is a brief overview of the decided cases.
a. General Standards for Determining the Meaning of Advertisements
In determining whether an advertisement has a “tendency or capacity to deceive,” the Federal Trade Commission (FTC or the Commission) and the courts upon review must interpret the advertisement in order to determine [*295] its meaning. In doing so, the FTC and the courts have interpreted advertisements broadly such that any potentially misleading impression is subject to the proscriptions of the FTC Act, even if other interpretations that would not be misleading were possible.
Thus, courts have frequently stated that an advertisement is to be judged by the “overall impression” it makes, not the literal truth or falsity of the representations. n18 Moreover, the overall impression standard considers the interpretation that the general consuming public is likely to ascribe to the advertisement. n19 The public is alternatively referred to by the courts as “the ordinary or average purchaser,” “the ordinary person to whom the advertisement is likely to appeal,” or simply “the general population or public.” n20 One of the points of this Article is to stress the important distinctions between these various formulations of the relevant reader.
b. The Relevant Reader
The courts generally take the position that it is the least intelligent and the most gullible who deserve protection from false advertisements under the FTC Act. n21 As a result, most courts have taken a paternalistic, and not terribly complimentary, view of the general public or at least that portion of [*296] the general public for whom the FTC Act provides protection. In FTC v. Sterling Drug, Inc., n22 the court stated:
“The general public has been defined as ‘that vast multitude which includes the ignorant, and unthinking and the credulous, who, in making purchases, do not stop to analyze but too often are governed by appearances and general impressions.’ The average purchaser has been variously characterized as not ‘straight thinking,’ subject to ‘impressions,’ uneducated, and grossly misinformed; he is influenced by prejudice and superstition; and he wishfully believes in miracles, allegedly the result of progress in science . . . . The language of the ordinary purchaser is casual and unaffected. He is not an ‘expert in grammatical construction’ or an ‘educated analytical reader’ and, therefore, he does not normally subject every word in the advertisement to careful study.” n23
The FTC has stated that it will not condemn an advertisement where only the exceedingly foolish could have been deceived. n24 Accordingly, some formulations state that an advertisement must deceive a “substantial number” of people in order to be considered a false advertisement under the FTC Act. n25 However, the precise number that constitutes a substantial number has never been stated by the FTC or the courts. n26
[*297] It is unclear from the cases whether the relevant readers should be confined to the “target market” (i.e., those readers identified as likely purchasers by the advertisers). The FTC has indicated that, when survey samples are used, they should be taken from the “appropriate population.” n27 However, the authorities have provided little guidance as to how that term should be defined. n28 Commentators have suggested that the standard should be more narrowly defined than “all readers,” and most appear to favor using the target market as the relevant group of readers. n29
The confusion over who the appropriate readers should be is probably best exemplified by the Second Circuit’s decision in ITT Continental Baking Co. v. FTC. n30 In ITT Continental Baking, the FTC charged the makers of Wonder Bread with producing advertisements which gave children the false impression that Wonder Bread was an extraordinary food for assisting children’s growth. n31 The Wonder Bread advertisements under attack contained a fantasy growth sequence in which a small child was seen growing to the size of a twelve-year-old in a few seconds. n32 Since the makers of Wonder Bread conceded that their bread could not produce extraordinary results, “[t]he case turned . . . on the meaning of the Wonder Bread advertisements themselves.” n33 The Commission’s complaint counsel relied on “the Commission’s own expertise” in interpreting the advertisements, but also submitted consumer surveys and psychological testimony on [*298] how children would view the advertisements. n34 Although the administrative law judge held that the advertisements did not make the alleged representation, the full Commission disagreed, and the Second Circuit affirmed. n35
ITT Continental Baking is of interest for several reasons, but the most important for our purposes is the court’s focus on the advertisements’ false representation to children. n36 This decision highlighted the ambiguity found in previous cases that vacillated between describing the relevant population for ascertaining deception as the “general public,” the “buying public,” or the “ordinary purchaser.” As gullible and as easily misled as they may be, seven-year-olds do not do a great deal of food shopping on their own. One could well argue that the purpose of the FTC Act is to prevent the fraudulent or unfair sale of products and services, not to prevent psychological harm to children. Unfortunately, however, the Second Circuit did not identify this as an issue. n37
2. Proving Deception
The FTC begins with an advantage in the cases it brings because it has a standard of interpreting advertisements that focuses on the net impression gleaned by the foolish and gullible. In addition, case law provides the FTC with even more leeway by holding that advertisements mean whatever the FTC says that they mean. In short, almost no evidentiary proof by the FTC is required at all, other than submission of the advertisements. Therefore, it is frequently intoned that “actual deception” is not necessary in a case brought by the Commission. n38 Arguments that the FTC should be required [*299] to poll consumers to ascertain how they interpret advertisements have been repeatedly rejected. n39
The FTC Act affords courts a very narrow power of review. Section 5(c) states that “[t]he findings of the Commission as to the facts, if supported by evidence, shall be conclusive.” n40 For that reason, and because of the purported “expertise” of the Commission in interpreting advertisements, great deference is given to the FTC’s interpretations of advertisements. n41 Moreover, the courts have stated that ambiguities in advertisements can be interpreted against the advertisers. n42
[*300] As a consequence of these broad rules favoring the Commission, the Commission has very little need for proof of what advertisements mean to the general public. The Commission occasionally offers dictionary definitions to define words n43 or places ordinary consumers on the stand to testify as to their understanding of advertisements. n44 However, the FTC’s prosecuting officer can simply place the advertisements in evidence and allow the Commissioners to draw their own inferences.
Beginning in the mid-1980s, the Commission indicated that it wanted extrinsic evidence to be submitted concerning the meaning of certain kinds of advertisements. n45 Specifically, the Commission — purporting to describe its past practice, although the historical accuracy of its description can be debated — stated that it had distinguished between explicit claims (ones that directly state the representation at issue) and implied claims (ones that were [*301] not explicit). n46 Within the range of implied claims, the Commission stated that it had further distinguished between “clear implied claims” (those for which the Commission can “conclude with confidence” that the representation is being made) and those that are less than clear. n47 For the latter type of implied claim, the Commission stated that it would require some sort of extrinsic evidence in order to determine the meaning of the advertisement to the reasonable consumer. n48 The Commission cited consumer perception surveys, conducted consistently with scientific principles, as the best type of extrinsic evidence to be offered. n49 The Commission also described other kinds of extrinsic evidence that it would accept. n50
In moving from their own intuition to empirical evidence of consumer understanding, the FTC Commissioners moved a bit closer to their academic critics, who before, during, and after the mid-1980s, attacked the FTC’s claim of expertise n51 and urged the Commission to utilize extrinsic evidence, n52 [*302] particularly consumer survey evidence. n53 The Commission’s apparent movement in this direction, however, is not irreversible. As Professor Ivan L. Preston has noted, the evidentiary standards set forth by the FTC in In re Thompson Medical have not been adopted by the courts, who continue to defer to the FTC’s untrammelled expertise. n54 Thus, the new evidentiary standards are self-imposed, and future Commissions would likely not be bound by those standards.
No showing of intent by the advertiser is needed for the FTC to enjoin an advertisement. n55 The FTC has stated that an “interpretation will be presumed reasonable if it is the one the [advertiser] intended to convey.” n56 [*303] In regards to an advertising agency or a distributing retailer, courts will consider their involvement in creating the advertisement when deciding whether to issue an injunction against the agency or retailer. n57
B. The Postal Service Act
Section 3005(a) of Title 39 of the U.S. Code, the Postal Service Act, provides that “[u]pon evidence satisfactory to the Postal Service that any person is engaged in conducting a scheme or device for obtaining money or property through the mail by means of false representations . . . the Postal Service may issue an order which . . . requires the person or his representative to cease and desist from engaging in any such scheme . . . .” n58 Cases brought under § 3005 frequently deal with advertisements sent through the mail, either alone or in newspapers and magazines. n59
The standard for determining what an advertisement means under § 3005 was set forth in Donaldson v. Read Magazine, Inc., n60 where the Supreme Court adopted an “ordinary mind” standard for determining what an advertisement means under the Postal Service Act. n61 At first glance, this standard does not appear to differ much from the standard used under the FTC Act, and, indeed, courts have occasionally invoked the language of cases decided under the latter Act. n62 Nonetheless, in practice at least, [*304] courts appear tougher on conclusions of the Postal Service than those of the FTC as to the effect an advertisement has on the ordinary reader. n63
1. The Relevant Audience and Proving the Meaning of Advertisements
Judge Clark’s concurrence in Peak Laboratories, Inc. v. United States Postal Service n64 illustrates the contrast between the perception of the ordinary buyer in the context of the Postal Service Act to those seen in the cases involving the FTC Act:
The typical shopper in 1977, who often marches into the market place armed with a current consumer report, would not be led astray by this assertion [of favorable test results of a product]. One is hard pressed to find advertisements today that fail to extol how well various products have fared in recent independent tests. . . . I say the mind of the ordinary reader says maybe so and maybe no when it notes an advertiser report of favorable tests or studies. n65
Other courts, while perhaps not as sanguine about our modern consumer society as Judge Clark, have nonetheless closely scrutinized claims that the ordinary mind would be deceived by an advertisement. n66 Despite this, no [*305] case has set forth any evidentiary requirement of extrinsic evidence that the Post Office must meet in order to prove that its interpretation of how an ordinary mind would perceive an ambiguous advertisement is correct.
Prior to 1968, the Postal Service Act required a finding of fraudulent intent before an order could issue. However, Congress passed an amendment in 1968 that eliminated any scienter requirement. n67
C. The Lanham Act
Section 43(a) of the Lanham Act prohibits any person from using any “false or misleading description of fact, or false or misleading representation of fact, which . . . in commercial advertising or promotion, misrepresents the nature, characteristics, qualities, or geographic origin of his or her or another person’s goods, services, or commercial activities.” n68 The Lanham Act further creates a private right of action for the benefit of “any person who believes that he or she is or is likely to be damaged by such act.” n69 Although originally used primarily for common law “palming off” or “passing off” cases, courts have also applied section 43(a) to false advertising cases. n70
[*306] 1. The Relevant Audience and Elements of Proof
At first blush, the general standards for determining whether an advertisement is false or misleading under the Lanham Act seem quite similar to those under the FTC Act and the Postal Service Act. n71 First, as with those other acts, courts will examine the advertisement as a whole in order to determine its message. n72 Second, an advertisement need not make a literally false statement under these acts in order to create liability if the overall message is misleading. n73
However, significant differences do exist between the acts. For one, the characterization of the relevant audience under the Lanham Act is usually of a relatively normal, if not astute, group of purchasers, eschewing the pejorative depictions favored by courts in opinions discussing the FTC Act. n74 Also, the audience is generally deemed to be “purchasers or prospective purchasers,” and a court will generally take into account whether the audience has special knowledge. n75
[*307] More importantly, while covering both “literally false” and “true but misleading” representations, the Lanham Act makes a key distinction between the two in the quantum of proof necessary to prove a Lanham Act cause of action. n76 This distinction was first raised in American Brands, Inc. v. R.J. Reynolds Tobacco Co. n77 The court was faced with a claim that an advertiser’s representation that its cigarette was “lowest” in tar falsely implied that no other cigarette had the same tar content (a not altogether unreasonable interpretation of the word “lowest”). n78 The court rejected that claim in the following oft-quoted language:
The subject matter here is different [from outright false statements]. We are dealing not with statements which are literally or grammatically untrue . . . . Rather, we are asked to determine whether a statement acknowledged to be literally true and grammatically correct nevertheless has a tendency to mislead, confuse or deceive. As to such a proposition “the public’s reaction to [the] advertisement will be the starting point in any discussion of the likelihood of deception. . . . If an advertisement is designed to impress . . . customers, . . . the reaction of [that] group will be determinative.” A court may, of course, construe and parse the language of the advertisement. It may have personal reactions as to the defensibility or indefensibility of the deliberately manipulated words . . . but the court’s reaction is at best not determinative and at worst irrelevant. The question in such cases is — what does the person to whom the advertisement is addressed find to be the message? . . . The fatal defect of both the [plaintiff’s] and [defendant’s] cases is that in neither instance has any evidence of substance been submitted which establishes the “determinative” reaction of the consumer to the respective advertisements. n79
[*308] Subsequently, the courts have repeatedly rejected claims of misleading or false advertising where no consumer survey evidence has been submitted. n80
Why are the FTC and the Postal Service permitted to simply declare an advertisement misleading, but judges in Lanham Act cases are not? Probably no good reason exists, n81 but several courts have attributed the difference in standards between the Lanham Act and the FTC Act to the FTC’s expertise in construing advertisements. n82 Courts have ruled that such expertise calls for deference to the FTC’s judgment, deference to which a Lanham Act plaintiff is not entitled.
The Lanham Act does not require any showing of intent. n83 Some cases, however, have held that a showing of intent can be used as proxy for [*309] consumer perceptions. n84 The rationale for this rule is that if the advertisers wanted consumers to receive a misleading or false message, then the court will presume that consumers did receive such a message. n85
II. The Discrimination Statutes
Section 704(b) of the Civil Rights Act of 1964, n86 section 4(e) of the Age Discrimination in Employment Act (ADEA), n87 and § 3604(c) of the Fair Housing Act, n88 all provide for liability for advertisements that indicate a preference based upon certain impermissible characteristics. This Part first examines the employment discrimination statutes and then the Fair Housing Act and analyzes the leading authorities interpreting each. Two problems that are unique to the discrimination statutes will also be considered: (1) whether the relevant readers for these statutes can be limited to a narrow subset of the entire population, and (2) what balance is required between truth and nondiscrimination?
A. The Employment Discrimination Statutes
Section 704(b) of the Civil Rights Act of 1964 states:
It shall be an unlawful employment practice . . . to print or publish or cause to be printed or published any notice or advertisement relating to employment . . . indicating any preference, limitation, specification, or discrimination, based on race, color, religion, sex, or national origin, except that such a notice or advertisement may indicate a preference, limitation, specification, or discrimination based on religion, sex, or national origin when religion, sex, or national origin is a bona fide occupational qualification for employment. n89.
Section 4(e) of the Age Discrimination in Employment Act states, “It shall be unlawful for an employer, labor organization, or employment agency to print or publish, or cause to be printed or published, any notice or advertisement [*310] relating to employment . . . indicating any preference, limitation, specification, or discrimination, based on age.” n90
A variety of different regulations interpret these statutes. The Equal Employment Opportunity Commission (EEOC) has promulgated a regulation which states, “The placement of an advertisement in columns classified by publishers on the basis of sex, such as columns headed ‘Male’ or ‘Female,’ will be considered an expression of a preference . . . based on sex.” n91 An EEOC regulation promulgated under the ADEA states that advertisements which contain, inter alia, the terms “college student,” “recent college graduate,” “boy,” “girl,” “retired person,” or “supplement your pension” indicate preferences based upon age. n92
At one time, the EEOC approach was generally considered to be a “trigger words” approach meaning that certain words automatically would be deemed to indicate a preference (as would the use of sex-segregated “help wanted” columns). n93 Some of the terms proscribed by the EEOC unarguably would indicate a preference, limitation, or discrimination — for example, the EEOC adduces the word “young” and phrases “age 25 to 35” and “age 40 to 50” as violating the ADEA. n94 However, other words and phrases proscribed by the EEOC indicate that the EEOC will characterize a phrase as discriminatory upon the basis of age even where there is no unambiguous reference to age but only a reference to a group that tends to be of similar age (e.g., retired persons or college students). n95 Thus, the EEOC does not require an unambiguous reference to a particular age group to find a violation of section 4(e). In specific applications, the EEOC seems to have followed this approach. n96
[*311] Courts have taken a similar, if somewhat more eclectic, approach to this issue. Hodgson v. Approved Personnel Service, Inc. n97 is perhaps the most well-known and comprehensive court decision in this area, dealing with a wide variety of help wanted advertisements published by an employment agency. The Hodgson court explicitly rejected the trigger words approach in favor of one that read the advertisements in “context.” n98 The context approach used by the court, however, seemed to do little more than distinguish between the employment agency’s advertisements for particular jobs and advertisements that generally offered the employment agency’s services. n99
In any event, the circuit court found that advertisements with phrases like “recent grads” and “those unable to continue in college” violated the ADEA, as did those with the words “girl” or “career girl.” n100 On the other hand, the court held that the phrases “junior secretary” and “junior accountant” had no age connotations, but merely described the seniority level of the position. n101 The court held that the phrase “athletically inclined” was proper because it “merely state[s] qualifications relating to . . . physical characteristics which can exist in persons of any age.” n102
[*312] Perhaps in response to Hodgson, the EEOC has tempered its view of trigger words. In a policy statement issued on job advertising, the EEOC stated that “judicial and Commission interpretations have added insight as to [the law’s] application” and that “generally there remains the need for a careful, case-by-case assessment as to whether a particular job advertisement runs afoul of sec. 4(e) [of the ADEA].” n103 The analysis requires an examination “not only of the language used in the advertisement but also the context in which it is used to determine whether persons in the protected age group would be discouraged from applying.” n104
Both the 1989 Policy Guidance and the EEOC Compliance Manual (EEOC Manual), guides for those enforcing the employment discrimination statutes, suggest that ambiguous advertisements require an investigation of actual hiring practices. In the “Compliance Procedures” section of the EEOC Manual, the EEOC states that two types of advertisements may deter those in the protected classes even though they make no explicit reference to those categories. n105 One type, according to the EEOC, is advertisements that use “college student” and “recent college graduate”; the other is the use of phrases like “young office group.” n106 The EEOC Manual states:
These types of ads can point to discriminatory hiring practices by respondents with a higher level of sophistication and awareness of Title VII and the ADEA. In such cases, consider investigating the respondent’s hiring practices, focusing initially on preserving the evidence (by on-site investigation, if possible) which identifies protected class applicants . . . . n107
The EEOC regulations notwithstanding, the terms “college student,” “recent grad,” and the like are simply not discriminatory on their face. Those phrases purport to state qualifications for the job, and it is altogether [*313] unclear that older college students or older recent graduates (as opposed to older people who did not meet the qualification) would be deterred by such advertisements, or that most readers would interpret those phrases as age preferences. Rather, the problem with these phrases is that they are job requirements that are usually not bona fide occupational qualifications. (It would be an odd job that could only be handled by a recent graduate and not one ten years out of school.) Advertising nonexistent requirements or requirements that have no true business purpose does indicate a preference, but for reasons that are different from explicit references to young people and males only. n108
As the EEOC Manual, but not the regulations, seems to recognize, references to “recent graduates” are discriminatory in the same way that the “some college required” advertisement in EEOC Decision No. 71-2237 n109 was discriminatory. Such phrases have a disparate impact. They discriminate because they discourage qualified people who do not meet the false requirement and such people fall disproportionately into a protected category.
Neither the cases nor the EEOC regulations have specifically adopted an ordinary reader test for discriminatory advertising provisions. Nonetheless, where ambiguous phrases like “junior accountant” are at issue, ascertaining the meaning of such phrases to those whom the advertisement is addressed (i.e., those seeking employment as accountants) is the only means of ascertaining meaning consistent with Title VII and the ADEA. The court’s own reaction to such phrases should be irrelevant, particularly since the judiciary probably has little experience in the accounting profession and cannot, without extrinsic evidence, pass judgment on the meaning of particular phrases within the profession.
Finally, no issues of scienter have been raised in the discriminatory employment advertisement cases. One assumes, I suppose, that those advertising for recent college graduates intend to attract and hire recent college graduates. In general, however, scienter is not required under the employment discrimination statutes.
B. The Fair Housing Act
Title VIII of the Civil Rights Act of 1968 constitutes the Fair Housing Act. n110 Section 3604(c) provides that:
[*314] [I]t shall be unlawful [t]o make, print, or publish, or cause to be made, printed, or published any notice, statement, or advertisement, with respect to the sale or rental of a dwelling that indicates any preference, limitation, or discrimination based on race, color, religion, sex, handicap, familial status, or national origin, or an intention to make any such preference, limitation, or discrimination. n111
Until recently, few § 3604(c) decisions were published. In one of the few, United States v. Hunter, n112 the Fourth Circuit found that “the ordinary reader” would find an advertisement for a “white home” to indicate a preference based upon race. n113 The court also held that the property owner intended to express such a preference. n114 The reference to the ordinary reader in Hunter survived and became the standard for § 3604(c) cases. n115 Determining what the ordinary reader perceives — not to mention how the court in Hunter divined it — is far less clear.
The Department of Housing and Urban Development has had little difficulty determining what it believes the ordinary reader would consider exclusionary. It has promulgated a series of regulations that are intended to provide guidance to publishers of advertisements. n116 It advises that certain “catch words” like “private,” “integrated,” “traditional,” “board approval” or “membership approval” should be avoided. n117 It also states that directions “made in terms of racial or national origin significant landmarks, such as an existing black development” or one “known for its exclusion of minorities” could send improper signals. n118 The regulations also advise that “[n]ames of facilities which cater to a particular racial, national origin or religious group, such as country club or private school designations” or those “which are used exclusively by one sex” should not be used to describe an area. n119 In addition, the use of “English language media alone” when non-English media is available, or selectively using media “in particular geographic coverage editions” of newspapers designed “for reaching a particular segment of the community” may indicate a violation. n120
[*315] Most importantly for this Article, the regulations provide that “selective advertising” may indicate an illegal preference. n121 Selective advertising under the regulations includes an advertising campaign using human models that caters primarily to one race or sex, without a “balancing” campaign in media directed at other groups, or racially mixed models used in advertisements for one building owned by a developer but not in advertisements for another. n122 Section 109.30(b) of the regulations adds:
Human models in photographs, drawings, or other graphic techniques may not be used to indicate exclusiveness because of race, color, religion, sex . . . or national origin. If models are used in display advertising campaigns, the models should be clearly definable as reasonably representing majority and minority groups in the metropolitan area . . . . Models, if used, should portray persons in an equal social setting and indicate to the general public that the housing is open to all without regard to race, color, religion, sex . . . or national origin, and is not for the exclusive use of one such group. n123
1. Consumer Psychology in Black and White
Since recent cases have focused on the use of human models in advertisements, a brief review of consumer psychology literature in this area will be useful. At least one popular theory, the “in-group bias” theory, posits that people are attracted to advertisements which use human models that are similar to them. n124 Race is one — but certainly not the only — characteristic that people appear to use to determine “similarity.” n125 Other similar characteristics include age and socioeconomic status. n126 Thus, according to the in-group bias theory, all other things equal (a big if), black readers will be more attracted to advertisements that include black human models, Oriental readers will be more attracted to advertisements that include Oriental human models, and so forth. n127
[*316] Although the theory is intuitively appealing, the evidence supporting the in-group bias theory is not overwhelming. With respect to white viewers, the empirical evidence supporting the theory is weak. n128 Although empirical evidence exists that an advertisement using models that are all black will receive the least favorable reaction from white viewers, almost all studies have concluded that whites evaluate integrated advertisements (those with both white and black human models) approximately the same as those using only white human models. n129 The “white backlash” from integrated human model advertisements, much feared by advertisers in the 1960s and 1970s, n130 has largely been rejected by available empirical evidence.
With respect to black consumers, some studies have provided strong support for the in-group bias theory. n131 Others, however, suggest that the evaluation of advertisements by blacks will not be significantly affected by the race of the models in the advertisements. n132 Still others have provided [*317] mixed support for the theory. n133 A number of researchers in this area have conceded that the results have not been entirely consistent. n134
Aside from their inconsistency, the studies analyzing the consequences of using human models of a particular race are not without flaws. Many of [*318] the studies used college students as subjects, thus casting some doubt on the relevance of the results to the population at large. n135 Some of the studies used black middle-class subjects, whose reactions might be more like those of middle-class whites than other blacks. n136 Furthermore, most of the studies were done a rather long time ago, and focused on “white backlash,” an important issue of the 1960s and 1970s (the concern that advertisers would not be able to attract white customers if they used advertisements with black human models). n137
In any event, the limitations of the in-group bias theory itself, disregarding any lack of consensus in the scientific community, are of far greater concern. In-group bias theory only tells us what certain viewers will find more attractive. It does not state that an alternative that is less attractive to certain groups will either be unattractive to those groups or will send a message of exclusion to them. In fact, even where the studies support the in-group bias theory, they frequently find that blacks have positive reactions to advertisements with all white human models, just not as positive as their reaction to advertisements with some black human models. n138 Nor does [*319] in-group bias theory state that those who prefer those same advertisements will interpret those advertisements as a sign that the product being sold is only for them. And the fact that the in-group bias theory has less support in tests of white subjects, who tend to evaluate integrated model advertisements similarly to all white model advertisements, raises further doubts about whether any message of exclusion is being received by them. n139
The literature relating to the role of race in consumer reactions has not addressed two additional problems that are raised in the context of a lawsuit concerning the use of all white human models. First, these lawsuits generally do not rely on one advertisement, but rather a series of advertisements, all of which depict white human models. n140 On the one hand, the repeated use of white human models may intensify the reaction of some readers. n141 On the other hand, this consideration also raises the problem of retention. When advertisements that appeared over several months are seen all at once (as the courts see them), the absence of one group is more conspicuous. But if one surveyed ordinary readers in their home after reviewing one particular week’s advertisement, they may not remember the advertisements from previous weeks. n142 People who do not recall the previous weeks’ all-white advertisements are far less likely to identify the current advertisement as discriminatory. n143 Accounting for this fact is a [*320] substantial problem for both the science of measuring consumer reaction and the law. n144
A second problem raised in the context of these lawsuits is whether housing is a product that is unique, and for which previous studies are of minimal use. Most people do not care who else is buying such products as Coca-Cola or Wonder Bread. In an apartment complex, or even a neighborhood, this may not be the case — a racist may not want to live in an integrated neighborhood and a strong advocate of pluralism and diversity may not want to live in an area that is all one race. n145
Finally, even if it is true (and no doubt it is) that some racial minorities will react unfavorably to all-white advertisements, these minorities cannot [*321] necessarily be assumed to have attributed a racist motivation to the property owner. They might blame the advertising agency, society at large, a market in which whites predominate, or just plain carelessness. However, it is the attribution by the reader of racist motivation to the person offering the property for sale or rental that is the sine qua non of a violation of § 3604(c); the advertisement must indicate to the ordinary reader a racial preference in the sale or rental of housing (as opposed to, for example, a racial preference in the employment of models).
In short, the determination of whether advertisements that use all white human models indicate a racial preference in the sale or rental of housing — and to whom such preference is indicated — is a complex factual issue, one in which the evidence provided by consumer psychology is less than clear. For better or worse, however, the judiciary has rushed in where social science feared to tread.
2. The Judicial Response
In the progenitor of the all white human model cases, Saunders v. General Services Corp., n146 the plaintiffs submitted expert testimony that a brochure for an apartment complex in which the models were almost all white indicated a racial preference. n147 Although the court found the research methodology employed by the experts questionable and weak, it nonetheless held that their findings comported with “the average layman’s knowledge” of advertising:
It requires no expert to recognize that human models in advertising attempt to create an identification between the model, the consumer, and the product. In other words, advertisers choose models with whom the targeted consumers will positively identify, hoping to convey the message that people like the depicted models consume and enjoy the advertised product. Therefore, if the consumer wants to emulate the model, he or she will use the product, too.
Thus, it is natural that readers of the . . . brochure would look at the human models depicted as representing the kinds of individuals that live in and enjoy GSC apartment complexes. If a prospective tenant positively identified with these models, the message conveyed would be that “I belong in these apartments. ‘My kind of people’ live there.” Conversely, if the prospective tenant reading the brochure saw no models with whom he or she could identify, the reader would obtain a message that “these apartments are not for me or ‘my kind.'” n148
Most of what Judge Merhige stated is a restatement of the in-group bias theory, the validity of which already has been discussed. It is the glib and [*322] unsubstantiated last sentence that is simply not supported by the work of most consumer psychologists. The absence of models attractive to particular viewers may mean that the viewers will not like the advertisement as much as others or even that they are not attracted to them at all. However, that is not the same as receiving a message that “these apartments are not for me.” n149 Affirmative marketing to one group is not necessarily tantamount to negative marketing to another.
In the recent case of Ragin v. Harry Macklowe Real Estate Co., n150 Judge Sweet refused to consider any consumer survey evidence at all, rejecting the one survey that was offered. n151 The paucity of evidence on the reaction of the “ordinary reader,” however, did not “impugn the evidentiary significance of the ads themselves.” n152 Based upon the ads alone, Judge Sweet found that ordinary readers, both black and white, would find that the advertisements indicated a preference based upon race. n153 On appeal, the Second Circuit explicitly rejected the argument that “survey evidence or expert testimony” was required, holding that the meaning of the advertisements to the ordinary reader could be discerned by “applying common sense.” n154
Most other published decisions concerning human models involve motions to dismiss and, perhaps not unreasonably, find that the exclusionary effect of advertisements is a question of fact. n155 In Ragin v. New York Times Co., n156 Judge Haight denied the Times’s motion to dismiss because he could not conclude, at the pleading stage, that a reasonable trier of fact might not find the advertisements published in the Times discriminatory in nature. n157 He stated that he expected “corroborative testimony from expert witnesses” at trial. n158
On appeal, the Second Circuit, in an opinion by Judge Winter, specifically upheld the ordinary reader standard. n159 The court stated that [*323] “[t]he ordinary reader is neither the most suspicious nor the most insensitive of our citizenry” and that the ordinary reader “does not apply a mechanical test to every use of a model of a particular race.” n160 In one (and only one) of its references to the ordinary reader, the court made an extraordinarily important ruling: an ad that would “discourage an ordinary reader of a particular race from answering it” violates § 3604(c). n161 This issue is discussed in depth in the next section. n162
In rejecting the Times’s argument that only plainly discriminatory advertisements — those depicting swastikas or burning crosses — violate the statute, the Second Circuit held that “Congress used broad language” and that “[o]rdinary readers may reasonably infer a racial message from advertisements that are more subtle.” n163 The court went on to hold, like Judge Haight, that a trier of fact might find that the racial configuration of an advertisement conveyed a racial preference. n164
The holding of the Second Circuit is relatively unobjectionable in that the court simply recognized the normal standards to be applied at the pleading stage. However, the court went further and offered several gratuitous observations that deserve closer analysis.
First, the Second Circuit addressed the Times’s concern that, as a practical matter, a rule that required the inclusion of minority human models would, in essence, be a “quota” rule. n165 The court pointed out that [*324] “advertising is a make-up-your-own world in which one builds an image from scratch . . . . The deliberate inclusion of a black model where necessary to avoid [a preferential] message seems to us a far cry from the alleged practices that are at the core of the debate over quotas.” n166 The court’s statement is true enough, and would be of significant force if there were only two races in the world. Unfortunately, the court missed the point. The quota problem arises if any racial, national, religious, physically handicapped, or other minority plaintiffs can sue — and get past the pleading stage — by noting the absence of models like them in an advertisement or series of advertisements. Logically, by conceding that advertisements that exclude blacks conceivably could send an exclusionary message to blacks, one must also concede that the absence of Chinese, Puerto Ricans, Hasidic Jews, paraplegics, or families with children could also send exclusionary messages to these groups. If it is the ordinary reader of a particular protected group that counts, as the Second Circuit suggested, then the liability issues become rather severe. Even in a “make-up-your-own world in which one builds an image from scratch,” the quota problem becomes one that is insurmountable for conscientious advertisers who wish to avoid a lawsuit. n167 The only intelligent legal advice for them is that they simply never use human models. n168
Second, like Judge Merhige, the court quite inappropriately decided to voice its own rather limited knowledge of advertising practices and consumer psychology. n169 The Second Circuit stated the view that “[a]n ad depicting a single model or couple of one race that is run only two or three times would seem . . . outside section 3604(c)’s prohibitions as a matter of law” but that “[a] housing complex that runs ads several times a week for a year depicting numerous white models as consumers and black models as doormen or custodial employees would have difficulty persuading a trier of fact that its ads did not facially indicate a preference.” n170 While this might be true, the court ignored significant learning in the field of consumer psychology, specifically the distinction between positive appeals and exclusionary messages, and the serious problem of retention in drawing this [*325] conclusion. But, more importantly, while the court’s advice as to what may be dismissed as a matter of law is always welcome, its prejudging of factual issues outside that category might be deemed inappropriate. n171 Indeed, in Ragin v. Harry Macklowe Real Estate Co., Judge Sweet, when faced with the complete absence of evidence concerning the ordinary reader, relied on the size and the number of advertisements — and then quoted the Second Circuit’s New York Times opinion. n172
Indeed, the most intriguing aspect of the human model decisions is the courts’ inevitable, unguided forays into the world of consumer psychology. The Second Circuit held that advertisements using only a few models for a few weeks is not a violation as a matter of law. n173 The Sixth Circuit has ruled that a single advertisement with only a few white models does not violate § 3604(c) as a matter of law, but that one with more models might. n174 The D.C. Circuit has stated that it is “highly unlikely” that a single ad containing only white models could violate § 3604(c). n175 But where is the line between cases that are immediately dismissable and cases that must go to a jury? Three white models? Five? Four models for three weeks? It seems safe to assume that whatever line is drawn will be arbitrary and will be based not upon consumer psychology but rather — as with Judge Merhige’s discussion of the layman’s understanding of advertising — upon the idiosyncratic perspectives of the black-robed among us. n176
[*326] C. The Relevant Audience
The suggestion of the Second Circuit in New York Times that the relevant audience for the discrimination statutes is the ordinary reader of a particular race (as opposed to Judge Sweet’s alternative determination that the standard is the ordinary reader regardless of race) deserves some consideration. n177 Again we need to ask who the relevant audience is in determining the meaning of an advertisement. Should subgroups count, as the Second Circuit suggests? n178 Should readers who are not potential purchasers of the housing at issue count?
Consider the following hypotheticals concerning various advertisements published in a newspaper in metropolitan area A, a diverse population in which 3% of the population is Chinese.
Hypothetical No. 1: An advertisement for housing states “No Chinese Allowed” in Chinese letters at the bottom of the page. However, no discrimination takes place at the housing site. The Chinese minority understands this admonition; the rest of the population, none of whom read Chinese, do not.
Hypothetical No. 2: Because of assimilation, only 10% of the Chinese population (.3% of the total population) understands the warning; the rest do not.
Hypothetical No. 3: The advertisement is for luxury condominiums in the heart of metropolitan area A. Only 1% of the Chinese population (.03% of the total population) is wealthy enough to even consider the apartments in question, and that 1% constitutes only .5% of the total population that are wealthy enough to purchase a luxury condominium. That is, the Chinese population has disproportionately low representation among the rich — six times lower than their representation in the population as a whole. Because of assimilation, none of the rich Chinese in question reads Chinese or understands the advertisement’s discriminatory statement.
The ordinary reader standard (without reference to subgroups) may leave you a bit cold with these hypotheticals, since the ordinary reader simply does not comprehend the discriminatory content of the message. Our intuition is that any statement which is, at least to some parts of the population, discriminatory on its face should fall within the aegis of the Fair Housing Act. These hypotheticals suggest that subgroups protected by the law present the appropriate relevant audience.
But even this standard may not be good enough. Consider that in the second hypothetical, the ordinary Chinese person does not understand the [*327] discriminatory content of the message, and, in the third hypothetical, not one person is deterred from visiting the housing because of the discriminatory content of the advertisement. If the purpose of the Fair Housing Act is to provide equal opportunities in housing, it should have no concern with those few people who understand the discriminatory content of the message, but could not in any event afford the condominiums. (Just as if the FTC Act is concerned with businesses profiting from fraudulent and dishonest practices, it should have no concern with the psychological effects of Wonder Bread commercials on seven-year-olds.)
However, permitting the relevant audience to include only particular protected groups, and relying only upon their reaction to the advertisements, may present other problems. Consider the following hypotheticals.
Hypothetical No. 4: A series of advertisements published in newspapers in metropolitan area A picture a large number of human models enjoying the facilities of an apartment complex. Black, white, and Oriental people are pictured, but no Eskimos. There are fifty Eskimos in metropolitan area A (.00003% of the population), all of whom interpret these advertisements as a message that Eskimos are not wanted at the apartment complex, although no one else interprets the advertisement in this way.
Hypothetical No. 5: The same advertising program is published for a luxury condominium complex. None of the offended Eskimos can afford the condominiums.
Our intuition here (or at least mine) is that the Eskimos should not have a claim under the Fair Housing Act unless there is some showing that the exclusion of Eskimos as human models in the advertisements was purposefully effected for the purpose of conveying an exclusionary message. n179
Between the exclusionary message written in Chinese and the exclusion of Eskimo human models lie any number of hypotheticals that are far more difficult to intuit — and that come far closer to factual scenarios in the real world. Suppose, for example, that a series of advertisements depicts only [*328] white human models. Blacks constitute 25% of the population in metropolitan area A, and 35% of them (just under 8% of the total population) find that the series of advertisements convey an exclusionary message. The entire white population (which, for the purposes of this hypothetical, constitute the other 75% of the population) is insensitive to the exclusion of blacks and does not read any discriminatory message into the advertisements. Do the messages indicate a preference, limitation, or discrimination based upon race? The answer, of course, depends upon who the relevant audience is deemed to be.
Suppose further that the advertisements are for luxury condominiums. Because of historic societal discrimination, blacks make up only 4% of the individuals and families that can afford apartments at the complex. Thus, only 1.4% of the population who can afford the apartments read an exclusionary message into it.
How does one escape this morass? In my opinion, the Lanham Act provides the appropriate guide. First, the relevant audience should not be the ordinary reader but rather the “ordinary potential buyer.” n180 The purpose of the Fair Housing Act is to prevent discrimination in the sale and rental of housing; n181 to include the perceptions of those who are not within the population of those who can afford the housing at issue is to ignore that purpose. Of course, determining the parameters by which one defines the ordinary potential buyer may be difficult, but no more difficult than defining the potential buyer of other products. n182
[*329] Moreover, I believe a focus upon subpopulations is unworkable. Although I have noted that problems may exist for either standard, the difficulties of allowing a single minority group to be the standard for determining liability seem the more intractable. If we allow the discriminatory nature of all-white advertisements to be determined solely by the perceptions of blacks, no logical reason exists why the perceptions of the Eskimos (and only the Eskimos) should not prevail in the hypothetical advertisements that excluded them. This would create the intractable quota problem about which the New York Times complained (and which the Second Circuit ignored). n183
Moreover, some of the problems of focusing on the population as a whole can be remedied by incorporating the Lanham Act’s treatment of intent. Consider the sly advertiser who specifically uses a message that he knows only blacks or some other protected group will understand — the pictorial equivalent of a message in Chinese. Under Lanham Act decisions, those who intend their advertisements to have a false or misleading meaning are assumed to have been successful. Similarly, housing advertisers who include a “No Chinese Allowed” message in their advertisements, even if written in Chinese and even if no one understands the message, should be found liable under the Fair Housing Act.
Finally, in proving how the ordinary potential buyer perceives advertisements, the Lanham Act’s distinction between false and misleading advertisements and its emphasis on consumer perception surveys is also worth adopting. When an advertisement is not discriminatory on its face, the law should require a plaintiff to produce a consumer perception survey demonstrating that the advertisements have a discriminatory message. If the opinions of the triers of fact in Lanham Act cases are “at best not determinative [*330] and at worst irrelevant,” it is difficult to understand why their perceptions should be given greater weight under the Fair Housing Act. n184 The somewhat bumptious musings of Judge Merhige in Saunders about the layman’s knowledge of advertising notwithstanding, the only way to know how an advertisement is perceived by ordinary consumers is to ask them. No self-respecting social scientist would say otherwise; the law should do no less.
Application of Lanham Act standards was presented and rejected directly by the court in Ragin v. Harry Macklowe Real Estate Co.: n185
Defendants’ reliance on our requirement that plaintiffs submit [survey or expert] evidence to prove confusion in trademark infringement cases, see, e.g., Johnson & Johnson * Merck Consumer Pharmaceuticals Co. v. Smithkline Beecham Corp., 960 F.2d 294, 298 (2d Cir. 1992), is misplaced. In trademark infringement cases, the inquiry focuses on whether there is a likelihood that a defendant’s mark will confuse a group of customers. In contrast, the inquiry directed by Ragin [v. New York Times] is whether a hypothetical ordinary reader would find that a defendant’s ads expressed an impermissible racial preference. Like the inquiry in negligence cases concerning whether a defendant’s conduct conformed with that of the reasonable person, this question is one that the factfinder can answer by viewing the ads and the defendants’ conduct and then applying common sense. No expert testimony or survey evidence is necessary, although such evidence no doubt is admissible. n186
Apparently, the Second Circuit found the the difference between “readers” and “potential purchasers” compelling. n187 I have argued here [*331] that the “potential purchasers” standard is more appropriate, n188 but, in any event, is there really a significant difference between readers of advertisements for headache remedies n189 or antacids, n190 and potential purchasers of those products? One would think that for such widely consumed products, the difference between readers and potential purchasers would be slim. n191 If a consumer perceptions survey is required when dealing with a group as broad as potential purchasers of headache remedies, it is hard to understand why the addition of a few more readers would remove that requirement.
Perhaps consumer psychology as a whole is a pseudo-social science and an utter waste of time, and the determination of what advertisements mean should always be determined by “applying common sense.” n192 If so, then the requirement of extrinsic evidence in Lanham Act cases, and the increasing emphasis on such evidence by the FTC, should be reversed. Personally, I find reliance on the application of common sense to be less than comforting, since, at least in the area of advertising using models of one or another race, the common sense of many members of the judiciary seems to differ so dramatically from what some scientifically conducted surveys conclude. n193 But, under any circumstances, the [*332] need for extrinsic evidence ought not to change depending on which statute is at issue.
With the Lanham Act standards in mind, then, let me briefly analyze the decision of Judge Sweet and the Second Circuit in Macklowe. n194 First, Judge Sweet explicitly rejected the proposition that the relevant audience should be limited to “those having the financial capacity sufficient to become renters” in the apartments in question, and thus refused to admit survey evidence from that group. n195 This, as I have noted, is inconsistent with the purpose of the Fair Housing Act. n196
Judge Sweet correctly, in my view, rejected the “ordinary black reader” standard in favor of the “ordinary reader” standard, n197 but his decision fails to suggest what effect this holding had on his final determination. As noted previously, consumer psychology suggests that a substantial difference may exist in the way that the ordinary black and the ordinary white person would view advertisements containing only white human models, with blacks being more sensitive to the absence of black human models than whites. n198 Nonetheless, Judge Sweet’s opinion suggests that the ordinary white reader would have noticed the absence of black models as readily as the ordinary black reader — and would have ascribed the same discriminatory i