Prefatory note:Those interested in the statistical issues raised in this case might find of interest “The Mismeasure of Discrimination,” a paper presented at a Faculty Workshop at the University of Kansas School of Law in September 2013, which addresses a number of issues pertinent to the Sears case.The role of evidence from the Sears case in leading to the articulation of the pattern whereby the rarer an outcome the greater tends to be the relative difference in experiencing it and the smaller tends to be the relative difference in avoiding it is discussed on the Sears Case Illustration sub-page of the Scanlan’s Rule page.Issues addressed in the Kansas Law paper have also recently been treated in my November 17, 2014 amicus curiae brief in Texas Department of Housing and Community Development, et al. v.The Inclusive Communities Project, Inc., Supreme Court No. 13-1371.
EEOC v. Sears, Roebuck and Co. is likely the largest employment discrimination ever fully tried. It was tried in the United States District Court for the Northern District of Illinois over a ten-month period from September 1984 to June 1985. The case, which involved allegations of gender discrimination in hire and promotion into commission sales position at over 800 Sears stores and allegations of gender discrimination in pay of exempt managerial and professional employees, consumed 131 trial days, with a transcript that ran over 19,000 pages. The trial court decision, which found for Sears on all issues, may be found at 504 F. Supp. 241 (N.D. Ill. 1980, and the court of appeals opinion upholding the district court’s decision, one judge dissenting, may be found at 839 F.2d 302 (7th Cir. 1988). The case received considerable attention in scholarly literature in some part because of the testimony (for Sears) of Rosalind Rosenberg, a prominent authority on women’s intellectual history, and (for EEOC) Alice Kessler-Harris, a prominent authority on women’s labor history (as is discussed, for example, in Ruth Milkman’s “Women’s History and the Sears Case,” Feminist Studies, 12:2 (1986:Summer), and various articles in a 1986 issue of Signs: Journal of Women in Culture and Society (Vol. II, No.4). The case is also treated, among a variety of other places, in Susan Faludi’s 1991 book Backlash: The Undeclared War on American Women. The case continues to receive occasional attention in various circles.[i]
My involvement in the case ran from the end of February 1979 until 1993 and entailed appraising the merits of the EEOC administrative decision underlying the case and recommending whether and how a case or cases based on the decision should be brought; signing the complaints in the case and four related race discrimination cases; serving as lead trial counsel for most of the period the case was pending, including pretrial litigation of certain procedural challenges, the trial itself, and the drafting the commission sales part of the EEOC appellate brief and arguing that portion of the appeal; and litigating certain costs and attorneys’ fees issues between the issuance of the initial district court decision in January 1986 and the resolution of the costs and fees issues in 1993. My initial role in the matter, and the one that led to my being placed in charge of the litigation, involve the drafting of certain memoranda critical of the analysis in the administrative decision, which memoranda were leaked to, and reported in, the press,[ii] complicating efforts to resolve the case short of litigation and creating various problems for EEOC in the litigation itself.
A key part of such criticism involved questioning the perception that the far greater female representation in non-commission than commission sales positions was evidence that women were discriminatorily excluded from the latter. The views then expressed would be later expressed in “Illusions of Job Segregation” (Public Interest 1988) and later articles raising similar issues with regard to the various job segregation cases that were prosecuted with considerable success in the 1990s.[iii] Table IV of Illusions, which showed the fallacy of a claim that lower qualifications among hires from a particular group was not evidence that the employer favored the group (a point also discussed, using data from the Sears case, with regard to perceptions of the implications of higher default rates among minority borrowers, in the unpublished “Confusion over Credit Discrimination” (1997)).
The views expressed in criticizing the findings of job segregation or assignment discrimination in the EEOC’s administrative decision were a product of my experience since approximately 1975 in reviewing for purposes of appeal the district court decision in EEOC v. du Pont (which case is discussed in Illusions), representing the EEOC and other government plaintiffs in resisting challenges to the affirmative action consent decree aimed at addressing perceived job segregation at AT&T and the Bell System Operating Companies (discussed in Illusions),[iv] including the Beal challenge to the decree (discussed in Illusions),[v] reviewing compliance with the consent decree, and writing the Addendum to the government’s Final Report in the case, as well as in representing the EEOC in the court of appeals in EEOC v. Datapoint Corp., 570 F.2d 1264, 1269 (5th Cir.1978) (which case will eventually be treated in a page concerning the inordinate weight given to presence or absence of statistical significance in law as well as the social and medical sciences, see Statistical Significance sub-page of Vignettes page).[vi]
The memoranda that were leaked to the press also addressed a number of other issues that may eventually receive treatment here. One of the memoranda, which urged the agency not to bring suit if there were other options for resolving the matter, made the statement that “Sears has been in the vanguard of voluntary affirmative action since 1974.” That language was sufficiently appealing to Sears that Sears paraphrased the language in advertisements maintaining that it would not yield to government pressures to reduce consumer credit, noting that “Sears has been in the vanguard of the voluntary extension of consumer credit since 1899 …”
The Sears case raised a host of interesting issues well before it went to trial. And the trial itself involved the presentation of statistical evidence by both sides of a level of sophistication not seen in employment discrimination litigation prior to the Sears trial or, to my knowledge, since. I probably will not ever give the case the full attention it deserves. But this page is a start in giving the Sears issues some of that attention. And gradually, in addition to providing narrative materials regarding some of those issues, I will be posting on this page various documents from the case (some of which are available on the Sears Documents sub-page of the Document Storage page of this site).[vii]
The issues that I hope eventually to address here will vary in technical content. Most of what I shall say will bear little resemblance to points made by either those supporting the EEOC’s case or those condemning it. The subjects on which I have views or knowledge that I regard as eventually worth recording include:
(1) the significance of the historian testimony;
(2) the perception that the EEOC should have presented testimony of putative victims of the alleged discrimination;
(3) the mistakes of the EEOC that provided Sears and the courts avenues for failing to address the strengths of the case (i.e., EEOC’s presenting any analyses unadjusted for characteristics and presenting any exhibits suggesting that the case could be evaluated without regard to the different patterns observed in different time frames);
(4) the initial analyses of the EEOC’s experts that followed the same approach that my 1979 memoranda had criticized;
(5) the EEOC’s applicant sample that reflected an effort to prove the commission sales claim in a manner not subject to the aforementioned criticisms;
(6) the relevance of the evidence in the case to Scanlan’s Rule (i.e., that the rarer an outcome the greater tends to be the relative difference in experiencing it and the smaller tends to be the relative differences in avoiding it), which, as mentioned in the prefatory note, is discussed on the Sears Case Illustration sub-page of the Scanlan’s Rule page;
(7) the sophistication of certain evidence offered by defense experts and the flaws of other evidence by those witnesses (such as that discussed with regard to Table IV of “Illusions of Job Segregation”);
(8) the EEOC’s presentation of the results of its logistic regression as an indicator of the expected female representation among hires when such figures were in fact the ratio of the odds that a female applicant would be hired to the odds that a male applicant would be hired, something utterly different, and the failure of experts of either side to recognize such error throughout the case;
[v]See Telephone Workers Union of New Jersey Local International Brotherhood of Electrical Workers v. New Jersey Bell Telephone Company, 584 F.2d 31 (3rd Cir. 1978).
[vi]As it may be some time before I create that page or other pages that specifically address aspects of the Datapoint case, I note the following here: EEOC lost the case in the district court and was assessed substantial attorneys’ fees by the Honorable John H. Wood, Jr. EEOC appealed the decision as to the merits and as to the attorneys’ fees. Datapoint’s appellate brief noted that the percentage of minority employees at Datapoint had grown over the period at issue, pointing out that such change was statistically significant, several times citing J. Johnston, Econometric Methods, McGraw-Hill Book Company, Second Edition, but without references to particular pages. Whatever pages Datapoint relied upon, however, it should be clear to an intelligent observer that the reference could only support the methodology for calculating statistical significance and that the fact of changes, whether statistically significant or not, had negligible bearing on the issue of whether discrimination occurred during any part of the period at issue, save possibly to suggest that the company was more likely have been discriminating during the earlier part of the period at issue than during the later part. (Such possibility, of course, should be appraised with an understanding that companies hire employees with different skills when they are just starting up and when they are fully operations.) The court of appeals, however, relied on the point in upholding a decision in favor of Datapoint, noting: “In the categories of officials and managers, professionals, technicians, office and clerical, and operators, the minority growth percentage was statistically significant, based upon the percentage of such employees at the beginning of the period. See, J. Johnston, Econometric Methods, McGraw-Hill Book Company, Second Edition, 1972, pages 24, 26, 27.” That the court of appeals opinion included page references not provided in Datapoint’s brief suggests that the court in fact examined the work to explore its pertinence to Datapoint’s argument. The Table of Authorities in Datapoint’s brief, however, reveals that the pages the court of appeals cited are not pertinent pages in J. Johnson Econometric Methods, but pages in the Datapoint brief where Datapoint had cited the work. Such is one example of the way courts deals with things they do not understand. .
[vii] The catalogue referenced in the first note indicates that a large volume of material on the case is maintained in the Harvard University Library. I have a great deal of material from the case. Anyone interested in a particular piece of evidence from the case may contact me by means of the information on the home page of this site. If I have the item, I can probably scan it and email or post it on the site.