39 worthiness. depending on the standard regarded appropriate for a specific situation. To allow for the limited information and unrefined data generally present at the initial stage, for example, a ten percent significance level (.10) could be used. As and when appropriate, the criterion level could be reduced without having to conduct repeated analyses of the kind that would be necessary following an initial adverse impact determination using the 80 percent rule. V. Every article or study of the 80 percent rule as a criterion for determining adverse impact that I am familiar with concludes that the rule is unsatisfactory. Legal commentators and scholars regard the rule as flawed. See e.g., Shoben, "Differential Pass-Fail Rates in Employment Testing: Statistical Proof Under Title VII." 91 Hary. L. Rev. 793 (1978) ("an ill-conceived resolution of the problem . . . will produce anomalous results in certain cases"); Van Bowen & Riggins. "A Technical Look at the Eighty Percent Rule as Applied to Employee Selection Procedures", 12 U. Richmond L. Rev. 647 (1973) ("Eighty percent rule not statistically valid and should not be used"); Greenberg, "An Analysis of the EEOC Four-Fifths Rule", 21 Management Science 762 (1979) ("A large number of actual cases of discrimination will go undetected and truly nondiscriminatory practice will be erroneously deemed discriminatory a substantial fraction of the time"); Burns & Ray, "A Comparison of the 80% Rule With a Statistical Test", unpublished paper (1980). VI. Judicial approval of the 80 percent rule has been limited to selection cases, and premised on deference to the administrative regulations contained in GESP, 229 CFR § 1607.4 (D). It is devoid of judicial acceptance in the context of utilization, or the standard set forth in OFCCP's final rules. Courts that have considered the 80 percent rule on the merits of its analytical value have rejected it. See, e.g., Rich v. Martin Marietta Corp., 167 F.Supp. 587 (D. Colo. 1979) (Test for statistical significance more appropriate than 4/5 rule); Hameed v. Ironworkers, 637 F. 2d 506 (Sth Cir. 1980) (passing score difference statistically significant although no 80 percent rule violation); Jones v. First Federal Savings & Loan Ass'n., 546 F.Supp. 762 (M.D. N.C. 1982) (Statistical methods more reliable than four fifths rule); Reynolds v. Sheet Metal Workers Local 102, 498 F.Supp. 952 (D.D.C. 1980) (Although EEOC guidelines entitled to deference, four-fifths rule fails to account for sample size), ait'd., 702 F.2d 221 (D.C. Cir. 1981); Brown v. Delta Airlines, 522 F.Supp 1218 (S.D. Tex. 1980) (adverse impact may be established by statistical evidence [80 percent rule not a statistical test]; Cormier v. P.P.G. Industries, Inc., 519 F.Supp. 211 (W.D. La. 1981) ("The Court finds that the 4/5ths rule is not an appropriate statistical test to use in examining the question of whether blacks have been discriminated against by being denied entrance into specific job categories"), aff'd., 702 F.2d 567 (5th Cir. 1983). Even where the 80 percent rule has been recognized by courts as a basis for adverse impact determinations, several decisions specifically indicate that the holding was confirmed by statistically significant disparities, rather than reliance on the rule itself. See, e.g., Guardians Ass'n. of New York City Police Dept. v. Civil Service Commission. 630 F.2d 79 (2d Cir. 1980); Chisholm v. U.S. Postal Service, 665 F.2d 482 (4th Cir. 1981); Gilbert v. City of Little Rock, 544 F.Supp. 1231 (E.D. Ark. 1982). In other instances, the disparities were so facially great so that statistical significance could be inferred without actual calculation, confirming the adverse impact finding of the 80 percent rule. Firefighters Institute v. City of St. Louis, 616 F.2d 350 (8th Cir. 1980). Conclusion The OFCCP final rules should not be accepted by EEOC so long as OFCCP employs the 80 percent rule for determining adverse impact in utilization. The Commission should seriously consider the broader question of the best measure for determining adverse impact in all contexts. Alternatives to the 80 percent rule are presently available that are acceptable to the courts, to comimentators, and consistent with professional analytical standards. U.S. DEPARTMENT OF LABOR SECRETARY OF LABOR December 22, 1986 The Honorable Augustus F. Hawkins and Labor U.S. House of Representatives Dear Mr. Chairman: Thank you for your letter of November 24 concerning the proposed Memorandum of Agreement between the Employment Standards Administration's Office of Federal Contract Compliance Programs (OFCCP) and the Bell Companies. I understand that OFCCP has had extensive and continuing discussions with your staff, going back to March of this year, on the conceptual framework of this agreement. In this connection, we appreciate the time and effort your staff has devoted to reviewing, analyzing and commenting on the proposed Memorandum of Agreement. Those comments raise a number of concerns you have with the proposed agreement. As a result of the concerns expressed in your letter, I have looked into this matter. Before addressing the particulars of the agreement I would like to offer several observations with respect to national reporting agreements. Initially, some misunderstanding seems to exist between officials of the Department of Labor and the Committee regarding the commitment the Department of Labor made during the Committee's June 1984 hearing on national reporting systems. At the time of the hearing, agreements had been developed with AT&T, IBM, Hewlett-Packard and General Motors Corporation. Departmental officials stated at the hearing that if a decision were made to enter into agreements with companies other than AT&T, IBM, Hewlett-Packard and General Motors Corporation, they would do so only after developing written criteria. In short, departmental officials felt that they had agreed to publish criteria before making new agreements with new companies, but did not intend to extend the commitment to include the renewal of any of the |