holding that failure to control for other explanatory variables makes an expert's table "essentially worthless"Summary of this case from Muñoz v. Orr
Argued September 29, 1988.
Decided November 16, 1988.
Mark J. Rogers, Angermeier Rogers, Milwaukee, Wis., for plaintiff-appellant.
John G. Hill, Jr., Marquette University, Milwaukee, Wis., for defendant-appellee.
Appeal from the United States District Court for the Eastern District of Wisconsin.
Judge Coffey heard oral argument in this case but later recused himself. The case has been decided by the remaining members of the panel. See, e.g., Love v. Young, 781 F.2d 1307, 1308 n. * (7th Cir. 1986) (per curiam); Giacolone v. Abrams, 850 F.2d 79, 80 n. * (2d Cir. 1988); United States v. Allied Stevedoring Corp., 241 F.2d 925, 927 (2d Cir. 1957) (L. Hand, J.) (death or resignation); cf. Tobin v. Ramey, 206 F.2d 505 (5th Cir. 1953). Two judges are a quorum of a three-judge panel, see 28 U.S.C. § 46(d), and "in the absence of a contrary statutory provision, a majority of a quorum constituted of a simple majority of a collective body is empowered to act for the body." FTC v. Flotill Products, Inc., 389 U.S. 179, 183, 88 S.Ct. 401, 404, 19 L.Ed.2d 398 (1967) (footnote omitted).
Dr. Glenn Tagatz, a professor in the Marquette University education school, brought suit against Marquette under Title VII of the Civil Rights Act of 1964, as amended, 42 U.S.C. § 2000 et seq., and the Age Discrimination in Employment Act of 1967, as amended, 29 U.S.C. § 626(c)(1). He charged that he had received smaller pay raises than colleagues who were either Catholic or under 40 years of age. Judge Warren, after a bench trial, entered judgment for Marquette, 681 F. Supp. 1344 (E.D.Wis. 1988), and Tagatz appeals.
The case is remarkable because, for the first time ever so far as we know, the plaintiff testified as his own expert witness. (We have found a case where a party's employees testified as expert witnesses; the practice was criticized by the judge. See Rust Engineering Co. v. Lawrence Pumps, Inc., 401 F. Supp. 328, 334 (D.Mass. 1975).) Dr. Tagatz, a specialist in statistical evidence in employment discrimination cases, prepared the statistical evidence on which his case rides, and was permitted to introduce the evidence as an expert witness. Rule 702 of the Federal Rules of Evidence, which governs the qualification of expert witnesses, is latitudinarian, and nothing in its language suggests that a party cannot qualify as an expert; nor did Marquette object to Dr. Tagatz's testifying as an expert witness. As Dr. Tagatz's counsel pointed out at argument, the fact that a party testifying as his own expert is not disinterested does not distinguish him from any other party who testifies in his own behalf; and hired experts, who generally are highly compensated — and by the party on whose behalf they are testifying — are not notably disinterested. There is a rule against employing expert witnesses on a contingent-fee basis (that is, against paying them more for their testimony if the party that hired them wins), and this rule might be thought to imply that a party — whose "reward" for testifying depends, of course, on the outcome of the suit — is not eligible to be an expert witness. But it is a rule of professional conduct rather than of admissibility of evidence. It is unethical for a lawyer to employ an expert witness on a contingent-fee basis, 3 Weinstein's Evidence ¶ 706 at pp. 706-23 to 706-24 (1987), but it does not follow that evidence obtained in violation of the rule is inadmissible. See United States v. Cervantes-Pacheco, 826 F.2d 310, 316 (5th Cir. 1987) (en banc) (concurring opinion). The trier of fact should be able to discount for so obvious a conflict of interest. In any event, there was no objection to Dr. Tagatz's testifying as an expert witness, so we need not delve deeper into this intriguing subject.
Because the case was tried, the intricate issues concerning prima facie case that Dr. Tagatz presses on us are irrelevant. United States Postal Services v. Aikens, 460 U.S. 711, 715-16, 103 S.Ct. 1478, 1481-82, 75 L.Ed.2d 403 (1983); Morgan v. South Bend Community School Corp., 797 F.2d 471, 480 (7th Cir. 1986); Washington v. Electrical Joint Apprenticeship Training Comm., 845 F.2d 710, 713-14 (7th Cir. 1988). The only question is whether Judge Warren's determination that Dr. Tagatz had not proved that he received smaller raises than he would have if he had been either Catholic or young was clearly erroneous. Although Dr. Tagatz attempts to present a disparate-impact claim, he has not identified any specific employment practice that he contends has a discriminatory effect. See Watson v. Fort Worth Bank Trust, ___ U.S. ___, 108 S.Ct. 2777, 101 L.Ed.2d 827 (1988). The only issue, therefore, is whether he proved deliberate discrimination either against non-Catholic, or against middle-aged, faculty.
Marquette is a Jesuit institution. See Maguire v. Marquette University, 814 F.2d 1213, 1214 (7th Cir. 1987). It gives preference to Catholic applicants for teaching jobs in fields where Catholic doctrine is important, such as theology and philosophy, as it is permitted to do by the exemption in Title VII for religious institutions. See 42 U.S.C. § 2000e-2(e)(2); cf. Pime v. Loyola University, 803 F.2d 351, 357-58 (7th Cir. 1986) (concurring opinion). But its stated policy is not to discriminate against non-Catholics once they are hired; and whether because of this policy or otherwise, Marquette declined to plead the religious exemption as a defense to Dr. Tagatz's claim of religious discrimination.
Dr. Tagatz, who is 54 and an Episcopalian, received his doctorate in educational psychology from the University of Wisconsin and was hired by Marquette as an associate professor in 1968. In 1971 he was made chairman of the department of psychology in the education school but he was removed from that job in 1974. In 1976 he published a book that received a single, brutally unfavorable review, though the dean of the school of education said he rather liked it; since then Dr. Tagatz has published only one article in a professional journal. According to the Social Science Citation Index, Dr. Tagatz's book has never been cited in a scholarly publication; and none of his articles has been cited since 1982. Compare Weinstein v. University of Illinois, 811 F.2d 1091, 1093 n. 2 (7th Cir. 1987).
Beginning in 1976 the dean placed annual raises on an explicitly meritocratic basis. Scholarly output (quality as well as quantity), teaching quality, and administrative and related service to the school or university are the criteria that the dean uses to determine how large a raise to give each member of the department. Between 1976 and 1981, Dr. Tagatz received average annual raises greater than some members of the department, but below the average. He brought this suit in 1982.
The evidence of discrimination is almost entirely statistical. It consists of a series of tables, prepared by Dr. Tagatz, that compare salary raises for Catholic and non-Catholic, over-40 and under-40 faculty members in the school of education. The samples are small, because the school of education has only a small faculty. In the table that Dr. Tagatz's counsel told us at argument contained the strongest evidence for his client, the salary raises between 1975 and 1985 of each of the 28 faculty members were compared with the faculty member's religion. The comparison showed that, on average, the Catholic members of the faculty received significantly larger annual raises than the non-Catholic ones. A similar table showed larger annual raises on a percentage basis for faculty members under 40 compared to those over 40. Marquette did not offer expert testimony, although Dr. Tagatz is incorrect to say that it offered no statistical evidence. It presented a table showing that the most productive scholars in the school of education, as measured by number of publications, happened to be Catholics. (Dr. Tagatz's own conspicuous lack of scholarly productivity — attributed by the school's dean to Dr. Tagatz's having "gone on strike" after he was fired as department chairman — helped depress the average for the non-Catholic members of the department.) Another table presented by Marquette compared Dr. Tagatz's numerical teaching evaluations with the average for the school's faculty, and showed that in the six periods for which comparable data were available the average score was 57.02, but his score was only 53.07. There was testimony that in addition to being a below-average teacher and at the bottom of the faculty in productivity, Dr. Tagatz was a poor citizen of the school — refusing an important committee assignment, failing to go through proper channels, enlisting students in his vendettas against other faculty members (the practice so memorably described in Mary McCarthy's novel The Groves of Academe), and generally behaving in trying ways. Against this there was testimony that Dr. Tagatz is a prominent member of the department and leads it in the number and size of research grants received.
Dr. Tagatz's main argument is that if the plaintiff in a Title VII case presents evidence that passes standard tests of statistical significance, the defendant must present a statistically significant rebuttal or lose. This argument misunderstands the significance of statistical significance, both generally and in this case. One of Dr. Tagatz's tables comparing salary raises for Catholic and for non-Catholic faculty in the school of education reveals a difference (favoring the Catholics) that is significant at the .0048 level. This means (speaking very crudely — see Kruskal, Tests of Significance, in 2 Int'l Encyclopedia of Statistics 944, 957 (1978)) that there is a probability of less than 5 in 1,000 that the difference is due to chance. But to infer as Dr. Tagatz would that there is a less than 5 in 1,000 chance (in the absence of contrary evidence) that the group that received the higher raises was favored because it was composed of Catholics is erroneous. All that the data show is that there is in all likelihood a real, not a spurious, difference between the means of the samples compared. The data do not show that the difference is due to a particular attribute (namely, being Catholic) which the members of the better-off group have and the members of the worse-off group lack. Correlation is not causation. St. Marie v. Eastern R.R. Ass'n, 650 F.2d 395, 400 (2d Cir. 1981) (Friendly, J.). Members of each group may have other things in common, and one or more of those other things, rather than religion, may have caused the difference in mean salary increases. Marquette's evidence suggests that the better-off group is not only Catholic but also more productive, and the group's higher average annual salary raises may be due to its members' greater productivity rather than to their religion. Dr. Tagatz does not question the proposition that "scholarly achievement is clearly recognized as a nondiscriminatory reason for personnel action in academic settings." Anderson v. University of Northern Iowa, 779 F.2d 441, 443 (8th Cir. 1985); see also Lieberman v. Gant, 630 F.2d 60, 66 (2d Cir. 1980) (Friendly, J.).
All this is not to say that Dr. Tagatz's tables are no evidence. Cf. MacDissi v. Valmont Industries, Inc., 856 F.2d 1054, 1058 (8th Cir. 1988). Correlations can be suggestive of causality; scientific induction means inferring causality from correlation. Having shown a correlation between religion and salary raises, Dr. Tagatz has shown that it is possible that notwithstanding Marquette's stated policy of treating its Catholic and non-Catholic faculty members alike, Marquette is discriminating against non-Catholics. But it is only a possibility. There are many other possibilities. One is that a Catholic university such as Marquette is more attractive to Catholic academics than to non-Catholic ones; this might explain why the Catholics in the school of education are more productive on average than the non-Catholics. Maybe it is a pure coincidence that the most productive members of the school of education are Catholic; for the sample on which both parties' tables are based is very small. Certainly in a case such as this where the plaintiff's statistical evidence is weak and equivocal, the defendant can rebut it by casual statistical evidence of its own, by nonstatistical evidence, or by a combination of the two types. Then it is up to the district judge to decide whether the plaintiff has carried his burden of persuasion.
Dr. Tagatz objects very strongly to Judge Warren's attempt to do a little statistical analysis of his own. One of Dr. Tagatz's tables compares the salary raises faculty members who had been continuously employed since 1969, and shows that over this period the average salary raise for the Catholics was higher than for the non-Catholics (including Dr. Tagatz). All five Catholics were full professors at the end of the period, while only two of the non-Catholics were; and Judge Warren noted that if one compared the five Catholic full professors with the two non-Catholic full professors the latter group had done better. Dr. Tagatz points out correctly that a larger sample is more reliable than a smaller one, and complains that Judge Warren made an already very small sample significantly smaller by casting out three of its ten observations. "It is an unacceptable statistical procedure to turn a large sample into a small one by arbitrarily excluding observations." Washington v. Electrical Joint Apprenticeship Training Comm., supra, 845 F.2d at 713. And perhaps the exclusion of the three non-Catholic professors of lower rank was arbitrary. But Dr. Tagatz's failure to control for differences in rank, like his failure to control for differences in scholarly productivity and teaching evaluations, made his table essentially worthless — and such variables can be controlled for. See, e.g., Tuckman, Gapinski Hagemann, Faculty Skills, and the Salary Structure in Academe: A Market Perspective, 67 Am.Econ. Rev. 692 (1977). Judge Warren's reworking of the table was a means of demonstrating just how worthless the table was rather than an impermissible exercise in generating factual data on which the parties were given no opportunity to comment.
Judges, by the way, are not wallflowers or potted plants; and a district judge's effort to test the strength of a party's statistical evidence by determining how sensitive it is to the design of the sample — a standard method of evaluating statistical evidence — is rather to be commended than condemned. Compare Mister v. Illinois Central Gulf R. R., 832 F.2d 1427, 1432-33 (7th Cir. 1987).
We have focused on the evidence of religious discrimination. The evidence of age discrimination was if anything less impressive. The discovery that young faculty members receive larger annual raises on a percentage basis is no evidence at all of age discrimination. As is well known, the professoriat, like the Roman Catholic hierarchy, has only a few ranks. And unlike the Catholic hierarchy, the top rank — full professor — is normally reached relatively early in the academic's career (Dr. Tagatz became a full professor in 1970, when he was 36). This rank structure implies, and one observes, that academics' salaries tend to rise rapidly in the early stages of their career and to reach a plateau when the academic becomes a full professor; and this regardless of age. The phenomenon of diminishing returns to years of experience is well documented in studies of academic salaries. See, e.g., Tuckman, et al., supra, at 697. Of course there are exceptions, but Dr. Tagatz's evidence was only of averages. It is unlikely — and there is no evidence — that the characteristic age profile of academic salaries reflects age discrimination. Among alternative explanations, tenure — and the job security that results from it — tends to take some of the edge off academic ambition. So we would expect tenured faculty to be somewhat less productive on average than nontenured faculty, and of course most nontenured faculty are under 40, and most faculty over 40 have tenure. A countervailing factor is that tenured faculty have passed a more careful screening process than untenured. A comparison not between a faculty member before and after he got tenure but between tenured and (different) nontenured faculty members might therefore reveal that tenured faculty were more productive. Cutting back the other way is the possibility that new faculty are more mobile than old — another reason why one might expect the new (who generally are younger) to receive higher raises. A further consideration is that tenure itself is a form of compensation, a substitute for money; this may be another reason why salary increases are higher on a percentage basis for nontenured than for tenured faculty. Given plausible (we do not suggest compelling) explanations unrelated to age discrimination for the results of Dr. Tagatz's statistical comparisons, we can hardly say that Judge Warren committed a clear error in refusing to infer, from evidence unusually meager, that Marquette was discriminating against Dr. Tagatz because of his age.
Dr. Tagatz attempted to bolster his weak statistical evidence of religious and age discrimination with shards of testimonial evidence. There was testimony that the dean had suggested to one of the non-Catholic faculty members that she not wear her Masonic ring (the Catholic Church once regarded the Masons as dangerous heretics, and the Masons returned the compliment); that he told the faculty that Catholic doctrine should be taught wherever possible; and that the younger members of the faculty were more productive than the older ones. Judge Warren was not required to give controlling weight to such evidence, especially since Dr. Tagatz teaches statistics and there is no suggestion that the dean wanted statistics taught with a Catholic slant (e.g., regression of angels on the heads of pins).
Finally, Dr. Tagatz complains that Marquette discriminates in favor of Jesuit members of the faculty by giving them free housing and other (free) benefits. It is a nice question, not necessary to answer, whether discrimination in favor not of a religion but of a religious order is within the scope of Title VII. See Pime v. Loyola University, supra, 803 F.2d at 353-54 (concurring opinion). The discrimination in favor of Jesuits, if that is what it is, is as costly to the Catholic members of the faculty, except those who are Jesuits, as it is to non-Catholics such as Dr. Tagatz. In any event, he failed to prove discrimination. It is true that the Jesuits receive various benefits that the other faculty members do not receive. But this is not the whole picture. Jesuits do not receive their salaries. Instead the university pays those salaries to an association which maintains the residence where the Jesuits live. The association deducts the cost of feeding and housing the Jesuits and various other expenses and returns the balance to the university. The amount returned is about $4,000 per year per Jesuit. So while the Jesuits do get free housing, it is not at the expense of the university; and while they get health insurance and other fringe benefits without contributing (as other faculty members do) to the cost of those benefits, there is no evidence that these benefits cost the university more than the amount that it receives from the association. So far as appears, far from discriminating in favor of its Jesuit faculty members Marquette is obtaining their services at lower cost than the services of its other faculty members.
Despite his use of the expert of his choice, Dr. Tagatz presented a weak case which was properly dismissed. Compare Laborde v. Regents of University of California, 686 F.2d 715 (9th Cir. 1982); Ottaviani v. State University of New York, 679 F. Supp. 288 (S.D.N.Y. 1988).