The New York is an upscale apartment house located on North Lake Shore Drive in Chicago. Believing that the advertisements its owners used to lure potential tenants to The New York were racially discriminatory, in violation of the Fair Housing Act, 42 U.S.C. § 3604, Anthony Tyus, Thomas Walker, and a number of fair housing groups brought this case against the individuals (and their wholly owned companies) who developed, operated, and managed The New York. With one minor exception, defendants prevailed at an unusually contentious jury trial. On appeal, the plaintiffs, supported by the Department of Justice and the National Fair Housing Alliance as amici curiae, raise a host of issues relating principally to the conduct of the trial and the jury instructions. We agree that the case must be reversed and remanded for a new trial, in light of several critical errors in both evidentiary rulings and the instructions.

* In 1985, defendant Development Management Group (DMG), which was owned by defendant Louis Silverman, began construction of The New York. When the building, a 593-unit luxury high-rise, was completed, DMG turned the management responsibilities over to defendant Urban Search Management ("Urban Search"), which was owned by Silverman's wife, defendant Diane Silverman. (Unless the context requires greater specificity, we refer to the defendants collectively as the Silvermans.) Louis and Diane Silverman both served on The New York's marketing committee, which created an aggressive advertising campaign designed to attract "desirable" tenants to the building (professionals between the ages of 25 and 40 with significant disposable income). Plaintiffs Thomas Walker and Anthony Tyus, both African-Americans, noticed in the course of their own private efforts to find housing in the Chicago area that the advertisements for The New York, whether in the form of billboards, newspaper ads, or brochures, featured only White human models. Walker passed along his observations to the Leadership Council for Metropolitan Open Communities, while Tyus called HOPE Fair Housing to report what he had seen.

In short order, the Interfaith Housing Center and the South Suburban Housing Center were also brought into the case, and all four organizations began monitoring the advertisements for The New York, as well as for other apartments. This activity went on from 1989 through 1992. The New York continued its exclusive use of White human models in its billboard and newspaper advertisements. Radio advertisements were aired that made no reference to The New York's commitment to being an equal housing opportunity provider, and many of the print advertisements omitted the Equal Opportunity Logo. This was a significant omission. The federal government provided a $4 million Urban Development Action Grant for use by the Silvermans' companies in development of The New York project. Under the Department of Housing and Urban Development regulations then in effect, the omission of the logo and the lack of any reference to an equal opportunity policy was a factor used in determining whether a landlord was in compliance with the Fair Housing Act's prohibitions against discrimination in advertising. 24 C.F.R. § 109.30 (1995). The Silvermans' companies also enjoyed the benefit of $62.6 million in tax-exempt bonds issued by the City of Chicago in conjunction with The New York project, which obligated them to abide by non-discriminatory practices. Chicago, IL, Municipal Code §§ 22970(c) and 221050 (1990).

Only after the plaintiffs filed the present lawsuit did The New York's advertising consultant prepare any advertisement depicting a racially diverse group of people, and the one ad the consultant prepared did not appear in the publications normally used by The New York, such as the Chicago Tribune and The Chicago Reader. The two individual plaintiffs and the four organizational plaintiffs filed their complaint on April 9, 1992, claiming that the Silvermans and their respective corporations had engaged in the practice of "selective advertising," in violation of the Fair Housing Act, 42 U.S.C. § 3604(c), and the regulations issued under that section. They sought injunctive relief, a declaratory judgment, compensatory and punitive damages, and costs and attorneys' fees. After a seven-day jury trial, however, the jury found in favor of DMG, Urban Search, and Diane Silverman on all points, and it found that although Louis Silverman had caused racially discriminatory advertisements to be printed, the plaintiffs had suffered no compensable damages. The district court awarded the defendants costs, but it declined to grant their request for attorneys' fees.

The Fair Housing Act prohibits racial discrimination of all kinds in housing. Section 3604(c) makes it unlawful in particular

[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 [or] color ... or an intention to make any such preference, limitation, or discrimination.

The implementing regulations in effect at the relevant time provide more detail, stating for example that "an advertising campaign using human models primarily in media that cater to one racial or national origin segment population without a complementary advertising that is directed at other groups" may be a violation. 24 C.F.R. § 109.25(c) (1995). The regulations also state that models used in display advertising "should be clearly definable as reasonably representing the majority and minority groups in the metropolitan area," 24 C.F.R. § 109.30(b) (1995), and that models "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." Id.

Extensive evidence relating to the advertisement campaign and its discriminatory qualities was introduced at the trial. Indeed, the jury must have been persuaded that the campaign was discriminatory, given its verdict against Louis Silverman. Nevertheless, the plaintiffs argue that the jury's finding that Louis Silverman's actions caused them no damages, as well as its exoneration of the other three defendants, is the product of errors that riddled the trial from start to finish. Moreover, they argue that they are entitled to a new trial because both the presiding judge and defense counsel prejudiced their case so severely in front of the jury that they did not get a fair hearing. For example, they assert that the judge intervened inappropriately on a number of occasions and allowed defense counsel to treat the plaintiffs' lawyer in a sexist and demeaning way. The defendants, in addition to defending the trial and its outcome, argue essentially that all's fair in love and war, and that test cases are somehow not entitled to the respect that other litigation enjoys. We have nothing good to say about the tone of this trial, and we emphasize how important it is for judges, attorneys, and litigants alike to treat everyone in the court with respect, in an atmosphere free of racial, sex-related, or other forms of bias. Defense counsel behaved inexcusably, referring to plaintiffs' lawyer disparagingly as "an amateur" on numerous occasions. He interfered with counsel's legitimate attempts to raise objections, barking at her to "sit down" and "stop interfering," and claiming in the presence of the jury that she "didn't know the rules." He literally grabbed papers out of her hands (all of this, we note, without any reprimand from the judge). Personal attacks on counsel and remarks that disparage others on the basis of sex or race have no place in the courtrooms of this Circuit. Here, we do not need to consider whether this conduct standing alone would have required a new trial, because we conclude that the specific errors discussed below require this step in any event. We turn then to the most important specific errors, which fall into three categories: error in the jury selection process, error in the treatment of plaintiffs' proffered expert witnesses, and error in the jury instructions.

During the initial voir dire of the jury panel, the following exchange took place between the court and one of the prospective jurors, Mr. Jackson:

THE COURT: Does anybody on the panel reside in public housing? I am sure everybody knows what public housing is.

MR. JACKSON: Yes.

THE COURT: Would you stand up and tell us about it, please?

MR. JACKSON: Public housing?

THE COURT: I can't hear what you're saying.

MR. JACKSON: I didn't hear you.

THE COURT: Do you reside in public housing?

MR. JACKSON: Yes.

THE COURT: What is the name of it?

MR. JACKSON: Chicago Housing Development.

THE COURT: Chicago Housing & Development. Would the fact that you reside there--is that an Afro-American housing or mixed?

MR. JACKSON: Mostly Afro-American.

THE COURT: Has there been any racial problems in that project?

MR. JACKSON: Yes.

THE COURT: Would that interfere in any way with you coming up with a fair and impartial verdict if you're selected on a jury?

MR. JACKSON: I can't really say.

Based on this exchange, the court excused Mr. Jackson for cause.

The problem with this decision was not the fact that the court excused someone for cause once he said that he was unable to say whether or not the subject under discussion would influence his ability to reach a fair and impartial verdict. It was instead the anterior problem of the court's decision to inject this line of inquiry, in the manner it did, into the questioning. While the trial court is granted broad discretion in directing inquiry at voir dire, that discretion is "subject to the essential demands of fairness." Aldridge v. United States, 283 U.S. 308, 310, 51 S. Ct. 470, 471, 75 L. Ed. 1054 (1931); United States v. Guy, 924 F.2d 702, 707 (7th Cir.1991). It is the duty of the trial court to ensure that collateral or unrelated issues are not brought into the case through the voir dire process. United States v. Daily, 139 F.2d 7, 9 (7th Cir.1944). Here, the question whether any members of the venire lived in public housing was not pertinent to any issue in the case. A trial judge may not determine juror eligibility based on external associations that are irrelevant to the case before him, regardless of what assumptions the judge himself may harbor about those associations. See United States v. Salamone, 800 F.2d 1216, 1225 (3d Cir.1986) (holding in the context of a criminal jury that to allow a determination of juror eligibility based solely on perceptions of the external associations of a juror would threaten the right of an accused to a fair trial by an impartial jury as well as the integrity of the judicial process as a whole). Furthermore, it was inappropriate for the court to ask about the racial composition of Mr. Jackson's residence, implying perhaps that public housing is likely to be plagued by racial problems, or that racial problems might arise if African-Americans were present. This injected a note of prejudice into the trial that plaintiffs had no way of overcoming, because the entire group of prospective jurors was exposed to it. The questioning was stigmatizing not only to Mr. Jackson, who was put on the spot in an embarrassing way in front of his fellow prospective jurors, but also by implication to the plaintiffs and their case. Inquiry of this nature would be unacceptable had it come from counsel; that it came from the court itself makes it all the more egregious, for "a critical function of a trial judge is to ensure that th[is] very type of highly inappropriate and irrelevant racially charged commentary ... gets no play before the jury." Rush v. Smith, 56 F.3d 918, 924 (8th Cir.1995) (Hansen, J., concurring).

This questioning went beyond the bounds of permissible inquiry during voir dire. Had this been an isolated incident, it may have qualified as harmless error. Unfortunately, it was not; it was instead a harbinger of the judge's approach toward the entire trial. For example, the judge permitted the defendants to argue that the Chicago Tribune's notice that the newspaper did not engage in discriminatory practices in its housing advertisements meant that The New York's discriminatory ads were harmless. The judge's remarks implied that the publisher's notice "certifies" that all advertisements were by advertisers who observe the fair housing laws--an inaccurate statement. Because we find below that other errors in the trial even more clearly call for reversal and a new trial, it is unnecessary to decide whether these incidents would also have justified such a ruling. Nevertheless, they demonstrate that the other matters were, in effect, just the most egregious in a trial that suffered from many problems.

Plaintiffs argue that the district court erred in excluding the testimony of their two expert witnesses: Dr. Douglas Massey, a professor of sociology at The University of Chicago, who would have testified about the history and patterns of housing discrimination in Chicago, and Dr. John Tarini, a psychologist, statistician, and chair of the Department of Marketing Communication at Columbia College (Chicago), who would have testified about how advertising sends a message to its target market and how an all-White advertising campaign affects African-Americans. The court excluded Dr. Massey's testimony on the grounds that it would range too far afield into the history of housing discrimination in Chicago and would thus be prejudicial and not sufficiently linked to the issue of The New York's advertising. After a lengthy proffer relating to Dr. Tarini's background and proposed testimony, the court concluded that he was not qualified to testify to the particular question of how the ordinary reader would react to The New York ads. In neither of its rulings on the experts did the court make use of the framework established by Daubert v. Merrell Dow Pharmaceuticals, 509 U.S. 579, 113 S. Ct. 2786, 125 L. Ed. 2d 469 (1993).

We have noted a number of times since the Supreme Court decided Daubert that its framework for assessing expert testimony is applicable to social science experts, just as it applies to experts in the hard sciences. See, e.g., United States v. Hall, 93 F.3d 1337 (7th Cir.1996) (requiring application of Daubert standard to social psychologist tendered as expert witness); Rosen v. Ciba-Geigy Corp., 78 F.3d 316, 318-19 (7th Cir.1996); Frymire-Brinati v. KPMG Peat Marwick, 2 F.3d 183, 186 (7th Cir.1993) (requiring application of Daubert standard to accountant tendered as expert witness). Social science testimony, like other expert testimony proffered under Fed.R.Evid. 104(a) for admission under Rule 702, must be tested to be sure that the person possesses genuine expertise in a field and that her court testimony "adheres to the same standards of intellectual rigor that are demanded in [her] professional work." Braun v. Lorillard Inc., 84 F.3d 230, 234 (7th Cir.1996); cf. Rosen v. Ciba-Geigy Corp., 78 F.3d at 318-19. At that level of generality, the Daubert framework is appropriate for all kinds of expert testimony; we do not agree with the Tenth Circuit's decision in Compton v. Subaru of America, Inc., 82 F.3d 1513, 1518-19 (10th Cir.1996), that it is limited to cases of novel scientific theories or methodologies. It is true, of course, that the measure of intellectual rigor will vary by the field of expertise and the way of demonstrating expertise will also vary. Furthermore, we agree with the implication in Compton that genuine expertise may be based on experience or training. In all cases, however, the district court must ensure that it is dealing with an expert, not just a hired gun.

In Hall, we described in some detail the methodology that is appropriate in these cases. The court must first consider whether the proffer demonstrated that the work that the scientific expert has done meets the standards of intellectual rigor described in Braun, using criteria like those described in Daubert for guidance. Next, it must determine whether the proffered testimony is based upon the expert's special skills. Finally, Rule 702 requires the court to decide whether the proffered testimony will assist the trier of fact to understand or determine a fact in issue. Social scientists in particular may be able to show that commonly accepted explanations for behavior are, when studied more closely, inaccurate. These results sometimes fly in the face of conventional wisdom. The court below erred insofar as it assumed that only evidence completely inaccessible to the jury could come in under Rule 702. As we noted in Hall, a trial court is not compelled to exclude expert testimony "just because the testimony may, to a greater or lesser degree, cover matters that are within the average juror's comprehension." 93 F.3d at 1342.

In this case, Dr. Tarini was prepared to testify about the way an advertising campaign sends a message to its target market and how an all-White campaign affects African-Americans. This kind of social research, which would demonstrate the way one of the most important industries in our country actually operates, would have given the jury a view of the evidence well beyond their everyday experience. (Indeed, anyone who watches television knows that the major consumer product companies in the country spend billions crafting their advertising campaigns, to reach exactly the right audience, with exactly the right pitch. It is doubtful they are making these decisions by asking six or twelve random people on the street what might appeal to them.) The proffer indicates that both the materials on which Dr. Tarini relied, which were peer-reviewed articles accepted in his profession, and the particular methodology he used, were well within the range contemplated by Daubert for expert scientific testimony. Using his admittedly extensive background, Dr. Tarini looked at The New York's ads in particular, using the well known "focus group" method, to see how they came across to African-American viewers. We are satisfied on the record as it was developed in the trial court that it was error to exclude this testimony.

With respect to Dr. Massey, the court had the power to consider under Rule 403 whether the evidence should have been excluded because its relevance was too slim, in light of potential prejudice, to warrant its admission. Testimony about the history and patterns of housing discrimination in Chicago may well have drifted too far from the question whether the advertising campaign for The New York was discriminatory, and it might even have opened the door to rebuttal testimony seeking to show that ads do not affect the racial makeup of a building. Nevertheless, the court failed to use anything like the Daubert framework in assessing Dr. Massey's testimony, and it cut off plaintiffs' efforts at establishing a record on the proposed scope of his testimony. This puts us at a significant disadvantage in trying to decide at this remove whether Massey's testimony, properly limited, was pertinent and, if so, how the court should weigh the likely prejudice against the usefulness of the testimony. We believe the best course is to leave it open on remand for the court to consider a new proffer of Dr. Massey's testimony, in light of its relevance to the issues to be tried and the evidence offered by the defendants.

Our review of jury instructions takes a practical approach, under which a new trial is appropriate only if it appears that the jury was misled and its understanding of the issues was seriously compromised, to the prejudice of the losing party. See EEOC v. AIC Sec. Investigations, Ltd., 55 F.3d 1276, 1283 (7th Cir.1995). Regrettably, that standard is satisfied for four separate issues: mitigation of damages, "actual injury," the statute of limitations, and punitive damages. We cannot agree with the defendants that any error was harmless, because the jury's decision not to award any relief to the plaintiffs (in spite of its verdict of liability against Louis Silverman), as well as its decision in favor of the other three defendants, may have been influenced by any one of these errors.

The most serious problem lay in the lengthy instruction the district court gave, at the defendants' request, on mitigation of damages. The court told the jury that a person "has a duty under the law to use reasonable diligence" to avoid or minimize damages and that the plaintiffs could not recover for any item of damage that they "could have avoided through a reasonable effort." At the trial, the defendants repeatedly stressed that individual plaintiffs Tyus and Walker had placed themselves in harm's way by purposely exposing themselves to The New York ads. Defendants argued that the plaintiffs could have avoided the harm if they had only notified The New York's management that the ads were offensive and discriminatory. Finally, the defendants criticized the plaintiffs for monitoring the ad campaign for a three-year period, asserting that they should have brought suit or notified the defendants as soon as they saw an offending ad. This, defendants claimed, would have avoided the alleged emotional and organizational injuries plaintiffs suffered.

The instruction given to the jury, in the context of this case, did not accurately state the law. While mitigation may be appropriate for cases in which a person is denied housing and must find alternative housing, see, e.g., Miller v. Apartments & Homes of N.J., Inc., 646 F.2d 101, 111-12 (3d Cir.1981); Young v. Parkland Village, Inc., 460 F. Supp. 67, 71 (D.Md.1978), it is not required for intangible losses or organizational claims, both of which are recoverable in Fair Housing Act cases. See, e.g., United States v. Balistrieri, 981 F.2d 916, 933 (7th Cir.1992), cert. denied, 510 U.S. 812, 114 S. Ct. 58, 126 L. Ed. 2d 28 (1993); Ragin v. Harry Macklowe Real Estate Co., 6 F.3d 898, 904-05 (2d Cir.1993). Thus, the very essence of the court's instruction--that these plaintiffs had a duty to avoid their damages--was wrong. The Fair Housing Act does not require notification prior to filing a suit, nor does it require a suit to be filed upon viewing or hearing the first discriminatory ad. Indeed, such a claim would certainly fail, because one ad standing alone would normally be insufficient evidence to show a pattern of discriminatory advertising. See, e.g., Housing Opportunities Made Equal, Inc. v. Cincinnati Enquirer, Inc., 943 F.2d 644, 648 (6th Cir.1991); Ragin v. New York Times Co., 923 F.2d 995, 1002 (2d Cir.), cert. denied, 502 U.S. 821, 112 S. Ct. 81, 116 L. Ed. 2d 54 (1991). The discriminatory character of an advertising campaign is often not self-evident, and the law does not automatically make every ad showing persons of only one race actionable. Only when enough ads have run to allow a pattern to emerge is a violation likely to become apparent. Thus, contrary to the defendants' arguments, repeated exposure to discriminatory ads is not masochistic behavior, or something designed only to run up the damages tab, but instead is normally the only way to tell if an advertising campaign as a whole violates the Fair Housing Act. See Balistrieri, 981 F.2d at 933; Ragin, 923 F.2d at 1002. The jury's mistaken view, based on this instruction, that the plaintiffs could not recover unless they had sued after the first ad, or notified The New York about their concerns, or simply averted their eyes, could easily have made a difference in its verdict.

Second, the court instructed the jury that one element of a Fair Housing Act violation is proof that the defendants' advertisement caused "actual injury." Rejecting plaintiffs' proposed instructions, which would have clearly stated that Tyus and Walker were entitled to compensation for intangible items like distress, indignity, humiliation, or embarrassment, the court instructed that "[a]n individual suffers injury when he proves that the advertisements would indicate a racial preference to the ordinary reader and that he has read the challenged advertisements. Injury to an organizational plaintiff can be established by proof of an actual injury to its efforts to assist equal access to housing." Only much later, when the court was instructing the jury on damages, did it inform them that they could award compensatory damages "if you find that the plaintiff suffered any emotional distress because of the defendants' conduct occurring after April 9, 1990 ... [and] any amount adequate to compensate [the organizational plaintiffs] for any impairment of [their] objectives and any diversion of [their] resources to programs independent of this suit." We address the problem with the date mentioned by the court below; here, we focus only on the confusing nature of the "actual injury" instruction, bearing in mind that the jury heard it in conjunction with the erroneous mitigation instruction.

These instructions could have misled the jury, because the individual plaintiffs did not need to prove they suffered from a tangible injury. Following the instructions step by step, the jury logically would have considered first whether any of the plaintiffs suffered an "actual injury," before it turned to the question whether any component of that injury might have included emotional distress. Again, to the extent the instruction allowed the jury to think that it could not consider the kinds of intangible harms mentioned in the plaintiffs' proposed instruction at that first step, it failed to follow this court's Balistrieri decision. As before, this went to the heart of the individual plaintiffs' claim to recover; the error was not harmless.

The statute of limitations error arose because the instructions confused the date by which a suit had to be filed with the period of time for which recovery is possible for a continuing violation. The Fair Housing Act requires that a suit be filed within two years "after the occurrence or termination of an alleged discriminatory housing practice." 42 U.S.C. § 3613(a)(1)(A). No one argues that the plaintiffs' suit was late here. Instead, the problem is that the district court limited damages to those experienced by plaintiffs between April 1990 and April 1992, apparently believing that the two-year period also created a cut-off point for damages. In a suit claiming that the defendant engaged in a continuous course of conduct that causes damages, however, a plaintiff can recover for damages that preceded the limitations period if they stem from a persistent process of illegal discrimination. Stewart v. CPC International, Inc., 679 F.2d 117, 121 (7th Cir.1982); McKenzie v. Sawyer, 684 F.2d 62, 72 (D.C.Cir.1982); see generally, Moskowitz v. Trustees of Purdue Univ., 5 F.3d 279, 282 (7th Cir.1993) (discussing continuing violation doctrine). The jury may have thought that all the damages suffered between the two dates specified by the court were self-inflicted, given the mitigation instruction they had heard, and they may therefore have failed to look at the defendants' course of conduct as a whole.

Finally, the court clearly erred in granting judgment as a matter of law to the defendants on plaintiffs' claim for punitive damages. Like all questions concerning judgments as a matter of law, we review this one de novo. Sokol Crystal Products v. DSC Communications, 15 F.3d 1427, 1432 (7th Cir.1994); Harrison v. Dean Witter Reynolds, Inc., 974 F.2d 873, 884 (7th Cir.1992), cert. denied, 509 U.S. 904, 113 S. Ct. 2994, 125 L. Ed. 2d 688 (1993). Punitive damages are available in Fair Housing cases when a defendant has shown "reckless or callous disregard for the plaintiff's rights, [or] intentional violations of federal law." See Balistrieri, 981 F.2d at 936 (quoting Smith v. Wade, 461 U.S. 30, 51, 103 S. Ct. 1625, 1637, 75 L. Ed. 2d 632 (1983)). The plaintiffs presented evidence of knowledge of the law and of intentional disregard of the law in implementing this particular advertising campaign. For example, plaintiffs showed that the Silvermans ran ads that omitted the fair housing logo, even though the Silvermans admitted that they knew that the law required its inclusion. This was enough to raise a jury question on the issue of the appropriateness of punitive damages.

Because we are remanding this case for a new trial, we see no need to provide an exhaustive discussion of other errors in this trial that may not recur. We comment here briefly only on some points raised on the appeal that have a reasonable chance of coming up again, in the hopes of moving the case along. Both plaintiffs and one of the amici argue that, if the district court was going to mention the HUD regulations applicable to advertising cases at all, it should have told the jury that the regulations pointed to liability here, and that the jury should have deferred to those regulations. This, we believe, misapprehends the nature of the problem with the court's instructions. The judge told the jury that it could consider the regulations as examples of practices that might violate the Act, but that the regulations carried no legal weight and required no particular action. The mistake here was twofold. First, the judge should not have referred the jury directly to the regulations; it was the court's job to describe the law clearly to the jury so that it could decide whether the practices before it violated the Act. Second, the court was plainly wrong to state that the regulations had no legal effect. They do, and the substance of the court's instructions to the jury had to reflect the law accurately, including, where appropriate, rules set forth in HUD regulations. Thus, plaintiffs are right to object generally to the court's approach to the instructions, but the remedy was not to tell the jury about deference to administrative agencies under Chevron, U.S.A., Inc. v. Natural Resources Defense Council, Inc., 467 U.S. 837, 104 S. Ct. 2778, 81 L. Ed. 2d 694 (1984). It was instead to give the jury an accurate description of the legal rules governing discriminatory advertising.

Plaintiffs also claim that the district court erred in rejecting their proposed instruction directing the jury to evaluate the advertisements from the standpoint of "the ordinary African-American reader," rather than from the standpoint of a generic ordinary reader. In Jancik v. Dept. of Housing and Urban Development, 44 F.3d 553, 556 (7th Cir.1995), this court held that an objective "ordinary person" standard should be used in determining what is "indicated" by an ad. The Fair Housing Act is violated "if an ad suggests to an ordinary reader that a particular [protected group] is preferred or dispreferred from the housing in question ... [or the ad] would discourage an ordinary reader of a particular [protected group] from answering it." Id. The district court's instruction said that "[a]n ad expresses a preference, limitation, or discrimination if it would discourage an ordinary reader of a particular race from answering it. An ad may be simply a neutral expression. The plaintiffs need not show that the defendants had a discriminatory intent." Bearing in mind the standard of review for jury instructions, we see nothing here that would have misled the jury or affected its understanding of the issues in a prejudicial way. See AIC Sec. Investigations, Ltd., 55 F.3d at 1283.

The remainder of the issues raised by both parties are better left to reconsideration by the new district court judge on remand. Both the relevance of the issues and the factual context in which they arise may be somewhat different. We hereby VACATE the district court's order awarding costs to the defendant, and we REVERSE and REMAND for further proceedings.