Plaintiff EEOC's claim in this Title VII case1 was based on both disparate treatment and disparate impact, the former requiring an intentional racial discrimination finding. The district court found no basis for concluding that Atlas had intentionally discriminated against potential black office and clerical employee applicants, 680 F. Supp. 1184. The focus in this case is upon the use of the Wonderlic Personnel Test in the screening of such job applicants, and the emphasis on appeal is whether the district court committed error in rejecting EEOC's contentions regarding the disparate impact theory of liability in light of the evidence.

There are several significant factors that bear upon plaintiff's contention in the challenge to the Atlas employment practices in question. The first is that the district court found "some evidence that the test was administered in a biased fashion."2 The second factor is the district court's finding that "Atlas has never had a black person working in its office." Third, Atlas selected a cut-off score of 25, contrary to Wonderlic's recommendations, for the positions in question even though a score of 21 for typists, 19 for file clerks, and 18 for telephone operators was suggested. As a consequence of choosing this high initial cut-off point, almost three-fourths of white applicants and well over 90 percent of black applicants failed to meet this criterion. An expert who testified for Atlas stated that this test methodology favored the white candidate by at least a three to one ratio.

Finally, all of the experts who testified in the case agreed that statistical evidence, if evaluated from 1969, when the Wonderlic test was first utilized, until 1984, the date of hearing, would reflect a significant disparate impact. Atlas contended that the proper period for analysis was 1978 through 1984, and produced expert witnesses who opined that the data for this period lacked statistical significance.3

The district court concluded that whether or not these significant factors were taken into account, plaintiff had failed to produce adequate evidence that a pattern or practice of racial discrimination was shown. It is clear, however, that at least one black applicant did "pass" the test and was not hired. Furthermore, the district court noted that on July 20, 1971, a black applicant, Kendrick, scored a 22 and was not hired while a white employee, Hutson, "scored only 21 [on an August 16, 1971 test and] ... was actually hired." A temporary white clerk who also scored 22 was hired in October. In 1973, white employee Dickerson took the test "two weeks after she was hired." Two whites were hired in 1974 who had scores of 17 and 21, and a "black applicant was not tested." In 1975, three of the six whites hired scored less than 25 (two scored 20 or less). In 1976, the same thing happened; two of eight whites hired scored 20 or less on the Wonderlic test. In 1977, black applicant Davis scored a 23 but was not hired. Later, a white applicant who scored 23 was hired. (In 1978, a black applicant who scored 20 was not hired and a white applicant who scored 19 was hired). Two whites were hired in 1980 with scores less than 25, and the district court noted that "it is unfortunate that Atlas did not seize upon the opportunity [in that same year] to hire this highly intelligent black applicant," Beverly Wilson, who applied during July and scored 25, equal to another white who was hired. Two other white applicants who scored less than 25 were hired in 1981 and in 1982 respectively. Three of four black applicants rejected in 1983 were not even tested. All this yearly information was found as a fact by the district court, who summarized by finding that 29 whites were hired who scored less than 25 on the Wonderlic, four with scores less than 20, while seven blacks scored better than these four white employees but were not hired.

We believe that these factual findings and other proof presented were sufficient to show that EEOC made out a prima facie case of disparate impact as indicated by the district court, particularly in view of the fact that not a single black person was hired to work in the office. In considering further whether Atlas rebutted the case made by plaintiff, the district court concluded only that in a clerical workforce as small as the one at Atlas, "any attempt to validate the Wonderlic test ... would have been statistically meaningless."

We disagree with the conclusion reached by the district court that the relevant statistical data and other related information, noted by the court itself, was statistically meaningless whether or not the test in question was attempted to be validated by a "formal study." It is evident that on a number of occasions over many years Atlas did not even test a number of black applicants and, moreover, hired many white applicants who were not tested at the time of hiring or who scored less than the requisite minimum score allegedly required. Under the circumstances, we are left with a definite and firm conviction that the conclusion reached upon a weighing of the relevant factors was in error. See Anderson v. City of Bessemer City, 470 U.S. 564, 572, 105 S. Ct. 1504, 1511, 84 L. Ed. 2d 518 (1985); Yung v. Raymark Industries, Inc., 789 F.2d 397, 400 (6th Cir.1986).

There is no fixed and firm rule regarding criterion for analyzing studies related to the cognitive ability test relied upon by Atlas. They must generally be evaluated by examination of "important elements of work behavior that comprise or are relevant to the job." Albemarle Paper Co. v. Moody, 422 U.S. 405, 431, 95 S. Ct. 2362, 2378, 45 L. Ed. 2d 280 (1975). It is sufficient here to say that no study was made at this particular Atlas office by any of the experts Atlas presented. Thus, there was little basis for the district court's conclusion that the purported screening test device was "more reliable than any other single predictor of job success." In the instant case, we conclude that the statistical data for the entire 1969-84 period was relevant and material, and that this statistical information and comparison between hiring practices with regard to white and black applications cannot properly have been deemed "meaningless" due to the size of the Atlas office workforce. We find, moreover, the evidence based on the entire fifteen-year period rather than the six-year period used by Atlas to be relevant and meaningful in this regard.

Looking at the statistical data and the testing practices adopted by Atlas, as well as the other factors previously discussed, including the failure to test and the disregard of relatively high test scores achieved by black applicants, indicates to us some evidence of bias in the administration of the tests. We must therefore reverse and remand the case to the district court. There was, then, statistical and other proof of racial disparate impact in the employment practices utilized by Atlas. There was evidence which would indicate that the testing and use of a purported cut-off testing score of 25 may have been a mere pretext to exclude blacks from employment in the office.

We make no judgment on this record as to whether, in theory, the proper use of the Wonderlic test may not be demonstrated to be job related in the case of clerical hires. If a defendant uses such a test, it must show that the "procedure used measures important skills, abilities, and knowledge that are necessary for the successful performance of the job." Black Law Enforcement Officers Ass'n v. City of Akron, 824 F.2d 475, 480 (6th Cir.1987). In short, the employer must show that the testing in question is "related to job performance." Griggs v. Duke Power Co., 401 U.S. 424, 431, 91 S. Ct. 849, 853, 28 L. Ed. 2d 158 (1971). In Griggs, use of the Wonderlic test in general plant and maintenance jobs was in question, a different situation from application of the test to clerical employees. Similar Wonderlic testing of plant employees generally, not office employees, was also considered by the Court in Albemarle Paper Co. v. Moody, 422 U.S. 405, 95 S. Ct. 2362, 45 L. Ed. 2d 280 (1975).

In summary, we are convinced that the following circumstances of this case compel a reversal: (1) not a single black office worker was hired over a span of fifteen years when 121 white office workers were hired; (2) 29 of the 121 whites hired did not meet the purported required minimum score of 25 on the Wonderlic test; (3) white workers in some instances received favorable adjustments on their test scores and blacks did not; (4) recommendations as to minimum acceptable scores for typing, filing, and telephone answering skills were not used.

We note in respect to a remand in this case that the expert witness offered by defendant, John Hunter, failed to visit and inspect the Atlas office and never studied the nature and content of the Atlas clerical and office jobs involved. The validity of the generalization theory utilized by Atlas with respect to this expert testimony under these circumstances is not appropriate. Linkage or similarity of jobs in dispute in this case must be shown by such on site investigation to justify application of such a theory.

A test may be used in jobs other than those for which it has been professionally validated only if there are "no significant differences" between the studied and unstudied jobs. 29 CFR Sec. 1607.4(c)(2). The study in this case involved no analysis of the attributes of, or the particular skills needed in, the studied job groups. There is accordingly no basis for concluding that "no significant differences" exist among the lines of progression, or among distinct job groupings within the studied lines of progression. Indeed, the study's checkered results appear to compel the opposite conclusion.

Albemarle Paper Co. v. Moody, 422 U.S. 405, 432, 95 S. Ct. 2362, 2378-79, 45 L. Ed. 2d 280 (1974).

Upon remand we direct the district court to consider whether the plaintiff has made out a prima facie case of race discrimination on a basis of disparate impact or disparate treatment under the circumstances. The district court must then determine whether defendant Atlas has articulated a sound and reasonable business reason for its actions under the standards of Texas Department of Community Affairs v. Burdine, 450 U.S. 248, 101 S. Ct. 1089, 67 L. Ed. 2d 207 (1981). If the defendant has not demonstrated a justifiable business basis for its practices, particularly in light of the disparities herein pointed out in the testing procedures employed and the inapplicability of the validity generalization theory, then judgment should be entered for plaintiff EEOC and a proper remedy provided.

REVERSED AND REMANDED for further proceedings in conformity herewith.

JULIAN ABELE COOK, District Judge, concurring in part, and dissenting in part.

I concur with my distinguished colleagues, who have concluded that this cause should be reversed. However, it is my belief that a remand is inappropriate under the circumstances of this case. Moreover, I believe that the trial record mandates a much broader review and critical analysis of the Appellee's hiring practices and testing procedures than that which has been provided in the prevailing opinion.

* On January 30, 1987, the trial court rejected a contention by the Equal Employment Opportunity Commission (EEOC) that the Atlas Paper Box Company (Atlas) was guilty of racial discrimination in the hiring of office and clerical workers between 1969 and 1984, in violation of Title VII of the Civil Rights Act of 1964, 42 U.S.C. Sec. 2000e et seq. The trial court and the EEOC focused their attention on Atlas' use of the Wonderlic Personnel Test, a cognitive ability test, in the screening of job applicants. The trial court rejected the EEOC's disparate impact and disparate treatment theories of liability.

Atlas is a Chattanooga, Tennessee company that makes boxes, school paper, commercial envelopes, and related items. Atlas' plant workers are generally unionized. In contrast, its office workers are non-union. The clerical staff numbers from twenty-five to thirty workers. The EEOC's allegations relate solely to the clerical staff at Atlas whose members were expected to perform a variety of work assignments. As an example, a worker, who had been hired to type, was also expected to file or assist with shipping when necessary. Tr. 139. The trial judge noted that the procedures for hiring new employees were not always the same, in that:

Atlas rarely advertises clerical positions or uses employment agencies. As a rule, when extra help is needed, it brings in temporary help and offers them permanent positions if they work out well. Each applicant for a permanent position is expected to fill out a questionnaire and provide the typical information about education, past work experience, etc. Applicants are also given a typing or keypunch test, the Wonderlic test, and an interview. However, the record from the many years at issue indicates that, from time to time, a quick judgment is made either to hire or reject and the tests or interviews may be omitted. There is no question that the complete hiring procedure has been truncated from time to time for both black and white applicants. Some people have even been hired first and tested later. When Atlas decides it needs to hire somebody, it usually selects from those who have applied in the very recent past and does not review the earlier applications kept on file.

EEOC v. Atlas Paper Box Co., 680 F. Supp. 1184, 1185 (E.D.Tenn.1987).

The trial judge also found that Atlas began its use of the Wonderlic Test in 1969, and added:

The company did some unsystematic experimentation with the test, trying it out on management, family members and employees in the clerical staff whose relative performance was already known. No formal attempt was ever made to verify the validity of the Wonderlic test as a predictor of job performance.

Id. at 1186.

He also pointed out that "[t]he Atlas management decided that it would try to hire people who scored 25 or better for all of its positions." Id. Significantly, the trial judge observed that although the Wonderlic manual suggests a hiring score of 25 for secretarial help, it only recommends scores of 21 for typists, 19 for file clerks, and 18 for telephone operators. Atlas contended that it set a relatively high cut off score in order to secure a staff of qualified persons who would be extremely productive. As a consequence, the trial judge found no evidence that "this initial decision was racially motivated." Id.

Interestingly, in the very next sentence of his decision, he stated that "[t]here was some evidence that the test was administered in a biased fashion." Id. Several kinds of evidence support this statement. First, Atlas did not follow the Wonderlic testing manual recommendation that a score adjustment be made on the basis of race in order to eliminate possible cultural bias in testing. Second, although the Company awarded scoring adjustments to certain older white applicants on the basis of their age, it gave no such favorable adjustments to black applicants, id. Third, the trial evidence indicated that ten white applicants were allowed to take the typing test on more than one occasion. No black applicants were given a second opportunity. Fourth, there was evidence that twelve job applications from black candidates bore marks of "B" or "Black" on them.1 Fifth, a significant number of white candidates were hired for clerical positions despite not having achieved a score of 25 on the test.2

The trial judge expressed his view of the net results of Atlas' hiring practices and procedures when he stated that "[a]lthough there are many blacks employed in the plant, Atlas has never had a black person working in its office." Id. at 1185 (emphasis added). Moreover, a former President of Atlas, Harry T. Robinson, Jr., also conceded at trial that the Company had never hired a black employee to work in the office. Tr. 143.3 According to Atlas' statistics, 148 of 699 white persons who took the Wonderlic Test passed with the requisite score of 25 (27.4%) whereas only 3 of 52 black applicants passed (5.8%). Atlas' expert, John Hunter, acknowledged that the Wonderlic Test generally favored the white candidate by at least a 3 to 1 margin. Tr. 112.4

Much of the testimony at trial consisted of experts who offered their respective interpretations of the statistical evidence. All of them agreed that the results on the Wonderlic Test, if evaluated on the basis of the entire 1969 to 1984 period, reflected a statistically significant disparate impact. However, Atlas' experts pointed out that any difference in the passing rates would lack statistical significance if the period between 1978 and 1984 was analyzed. In reaching this conclusion, they included Virginia Cunningham and Ben Smith among the clerical staff.5

Atlas also introduced expert testimony which purportedly demonstrated a certain correlation between the Wonderlic Test and clerical jobs at other businesses. These experts opined that the small sample of the work force at Atlas precluded the compilation of any meaningful job relatedness study. However, the EEOC experts were critical of this assessment, asserting that these studies had been compiled by someone who had neither visited the Company plant nor analyzed the tasks involved in the jobs performed there.6

The gist of the trial judge's reasons for rejecting the EEOC claim is summarized by his assertion that there was no evidence that a black applicant would have "qualified" for employment at Atlas even if any biased administration of the test or any other area had been removed. 680 F. Supp. at 1189.

The decision of a district court as to whether a defendant has violated Title VII is generally subject to an appellate review under the clearly erroneous standard. Pullman-Standard v. Swint, 456 U.S. 273, 290, 102 S. Ct. 1781, 1791, 72 L. Ed. 2d 66 (1981). In Pullman, the Supreme Court held that the determination of whether a defendant acted with discriminatory intent deserved such deference. The clearly erroneous standard has been extended to disparate impact cases. See e.g. Atonio v. Wards Core Packing Co., Inc., 827 F.2d 439, 443 (9th Cir.1987). Under this test, the reviewing court must affirm the trial court unless it is "left with the definitive and firm conviction that a mistake has been committed." United States v. United States Gypsum Co., 333 U.S. 364, 395, 68 S. Ct. 525, 542, 92 L. Ed. 746 (1948). However, the legal views of the trial court are subject to a de novo review on appeal even under Pullman.

The Supreme Court in Griggs v. Duke Power Co., 401 U.S. 424, 91 S. Ct. 849, 28 L. Ed. 2d 158 (1971) and Albemarle Paper Co. v. Moody, 422 U.S. 405, 95 S. Ct. 2362, 45 L. Ed. 2d 280 (1974) set forth the burden of proof standard in a Title VII disparate impact case. Griggs and Albemarle are highly instructive because both of these cases involved the exact employment test at issue here--the Wonderlic Test. Moreover, the nature of the evidence, which has been proffered by Atlas to validate the tests, have distinct similarities here.

In Griggs, the Supreme Court unanimously ruled that Title VII prohibits the use of employment tests which have a discriminatory effect unless the employer meets "the burden of showing that any given requirement [has] ... a manifest relationship to the employment in question." 401 U.S. at 432, 91 S.Ct. at 854. In Albemarle, the Supreme Court explained that:

This burden arises, of course, only after the complaining party or class has made out a prima facie case of discrimination, i.e., has shown that the tests in question select applicants for hire or promotion in a racial pattern significantly different from that of the pool of applicants.... If an employer does then meet the burden of proving that its tests are "job related," it remains open to the complaining party to show that other tests or selection devices, without a similarly undesirable racial effect, would also serve the employer's legitimate interest in "efficient and trustworthy workmanship ..." Such a showing would be evidence that the employer was using its tests merely as a "pretext" for discrimination. In the present case, however, we are concerned only with the question whether Albemarle has shown its tests to be job related.

422 U.S. at 425, 95 S. Ct. at 2375 (citation omitted).7

The major difference between a plaintiff's burden of proof in a disparate impact case and a disparate treatment case is that proof of discriminatory intent need not be shown in the former context. With regard to the disparate impact issue, Judge Pierce Lively, in Lynch v. Freeman, 817 F.2d 380, 383 (6th Cir.1987), wrote:

In such a case, the trial court is concerned with "the consequences of employment practices, not simply the motivation." Griggs v. Duke Power Co. (citations omitted) (emphasis in original). Disparate impact cases typically are concerned with facially neutral practices or standards that in fact work to place a disproportionate burden on a discrete group of employees who are protected under Title VII.

Accord, Kent County Sheriff's Assn. v. Kent County, 826 F.2d at 1492-1493.8

In the instant case, the trial court made a somewhat ambiguous statement regarding whether the EEOC had met its prima facie burden of demonstrating disparate impact:

As the Court understands the statistical proof offered in this case, the EEOC has made a prima facie case that the use of the Wonderlic test for screening Atlas' clerical applicants had a statistically significant adverse impact on black applicants at least before 1978, although apparently not in recent years.

680 F.Supp. at 1188 (emphasis added). Because the trial judge's use of "apparently" suggests that an adverse impact had not been shown during the later period, Atlas seizes upon this point to argue that the trial judge found no adverse impact after 1978. In addition, the Company argues that it would have been improper for the trial court to have relied upon any evidence prior to 1978.

Assuming, arguendo, that Atlas can validly assert this point even though it did not make any formal cross appeal on this issue,9 this decision by the trial court must be construed as a finding of adverse impact. In my opinion, any other ruling would be in error.

There are several reasons why such a conclusion is justifiable. First, the trial judge never explicitly determined that the EEOC had failed to establish any adverse impact after 1978.10 Even though he held "apparent" doubts about the post 1978 period, the trial judge still analyzed the pretext issue. Thus, the most reasonable interpretation of his decision is that, although finding adverse impact prior to 1978 and retaining some doubts about the post-1978 period, he was willing to assume that proof of adverse impact for the entire 1969 to 1984 period had been established on the assumption that the defendant had satisfied the job relatedness requirement.

Second, if the trial judge meant to fragment the EEOC's proofs as Atlas suggests, such a ruling would have been totally without foundation and erroneous. An examination of Atlas' arguments for segregating the statistical period will demonstrate that each one of them amount to unjustified fragmentations of the data. See, Capaci v. Katz & Besthoff, 711 F.2d 647, 654-55 (5th Cir.1983), cert. denied, 466 U.S. 927, 104 S. Ct. 1709, 80 L. Ed. 2d 182 (1984). Hence, there is no basis for discussing the adverse impact for any period other than 1969 through 1984 in toto.

Atlas asserts that only evidence subsequent to 1978 or 1979 is probative to this controversy since the original charging party filed her claim in 1980 or 1981. In support of this position, Atlas argues that backpay awards in Title VII cases are limited by 42 U.S.C. Sec. 2000e-5(g) to covering the two year period immediately prior to the filing of the charge. Even if Atlas has preserved the issue for appeal, this argument must fail because the limitation of a plaintiff's damages to certain periods does not mean that evidence of discrimination for prior periods cannot be admitted and evaluated.

In Held v. Gulf Oil Co., 684 F.2d 427 (6th Cir.1982), this Court ruled that a plaintiff may maintain a Title VII action based on incidents which occurred prior to the 180 day limit for filing an EEOC charge if "subsequent identifiable acts of discrimination occur within the critical time period and are related to the time barred incident." Id. at 430 (emphasis added).

This doctrine was reaffirmed in the context of an age discrimination case in Janikowski v. Bendix Corp., 823 F.2d 945, 948 (6th Cir.1987) where an "ongoing discriminatory policy" was the same before and during the filing period. Evidence of the effect of that policy in the period beyond the 180 day filing limit was relevant. See also, Bazemore v. Friday, 478 U.S. 385, 402, 106 S. Ct. 3000, 3010, 92 L. Ed. 2d 315 (1985) (Brennan, J. concurring); Black Law Enforcement Officers Assn. v. City of Akron, 824 F.2d at 483; Walker v. Jefferson County Home, 726 F.2d 1554, 1557 (11th Cir.1984) ("In a disparate impact case, the court clearly may consider evidence of prior discriminatory acts if such evidence is relevant to show independently actionable conduct occurring within the statutory period"); Guardians Assn of New York City v. Civil Service Commission, 633 F.2d 232, 249 (2d Cir.1980), aff'd, 463 U.S. 582, 103 S. Ct. 3221, 77 L. Ed. 2d 866 (1983); Harless v. Duck, 619 F.2d 611, 615 (6th Cir.), cert. denied, 449 U.S. 872, 101 S. Ct. 212, 66 L. Ed. 2d 92 (1980); Police Officers for Equal Rights v. City of Columbus, 644 F. Supp. 393, 431 (S.D.Ohio 1985).

In the instant case, the same employment test was given from 1969 to 1978 as for the period after 1978. The results of this test are certainly a part of an "ongoing policy" for the entire period. Thus, the pre-1978 results are relevant, properly considerable, and fully admissible.

In another argument, Atlas posits that the Chattanooga school system no longer experienced the effects of segregation after 1978, and, therefore, any pre-1978 disparate impact evidence was due to the inferior schooling which had been previously provided to black students.

This argument must also fail because it is totally lacking in supporting documentation. No statistical studies regarding the Chattanooga school system were introduced into the record or examined by the trial court. Moreover, no statistics were produced which demonstrated the absence of any lingering effect. Finally, the Court was not provided with any studies which addressed the correlation between test scores and segregated schooling.

Significantly, and running counter to this argument, the following revealing cross examination of Thomas Geraghty, an expert witness for Atlas, took place:

QUESTION: Are you saying that by 1978, that blacks had achieved a level equivalent to those of their white counterparts and therefore testing would not be--their race would not be a factor in their tests, test scores?

ANSWER: I said that by that time, probably in my judgment, the adverse effect of prior education had worked its way through the system.

QUESTION: Has there been any studies made of this?

ANSWER: There may have been or may not have been.

QUESTION: Are you aware of any studies?

ANSWER: No.

QUESTION: Do you know of any publication that has shown that because of integration and the result of integration that the scores of blacks on scored tests have improved and reached parity with their white counterparts?

ANSWER: I have seen--I cannot cite specific citations. I have seen references in the newspaper to improvement of scores of black students. Whether they have reached parity or not I cannot testify. I haven't--that's not my field of interest.

Tr. 53-54 (emphasis added).

Notwithstanding the reasons which have been advanced by Atlas, there is evidence in the record that 1978 was picked as a cut off date by one expert at the direction of Atlas' attorney. Tr. 62. Moreover, the selection of this date ensured the inclusion of Cunningham in the sample. Tr. 57. Another Atlas expert, Mickey Petty, stated that his selection of cut off dates between 1969 and 1984 "is admittedly arbitrary." Tr. 212.

Finally, Atlas argues that it had only experimented in the Wonderlic Test prior to 1978 and, hence, any statistics which result therefrom are not reliable. There are several serious problems with this argument. First, the trial record does not support this conclusory assertion that almost a full decade of concededly unscientific experiments took place. Harry T. Robinson, Jr. testified that "in the early part of the 1970's, we were evaluating the test and we may have hired anybody to see what kind of a performance we got from that type of score." Tr. 160. Thus, even his testimony limits the testing to the "early part" of the 1970's. This suggests that the experimentation continued only until 1972 or 1973.

Second, evidence of black applicants' performance during that earlier period is relevant and gives evidence that Atlas continued to utilize and rely upon the Wonderlic Test despite the disparate results. Third, there is nothing in the record which would indicate that any cut off other than 25 was used by Atlas. Finally, if Atlas is asserting that the Wonderlic Test had no real cut off during nine years of its use, the arbitrary nature of its administration exceeds the factual findings of the trial court.

Thus, for all the above reasons, I believe that Atlas' arguments against measuring the impact of the Wonderlic Test for the 1969 to 1984 period must fail. Since experts from both sides of this controversy agree that the Wonderlic Test had statistically significant disparate impact for the 1969 to 1984 period, see e.g. Tr. 215, 218 (Atlas' expert admits adverse impact for entire period), the EEOC clearly established its prima facie case.11

In Chrisner v. Complete Auto Transit, Inc., 645 F.2d 1251 (6th Cir.1981), this Court addressed some conflicting rulings with regard to a defendant's burden once a prima facie case has been demonstrated. In that case, we adopted a sort of sliding scale that was based on "considerations which are a function of the jobs demands." Id. at 1262.

In particular, this Court held that it would be somewhat more lenient in scrutinizing employment practices for jobs which implicate public safety (e.g., airplane pilots) than for other positions. Regarding those jobs that did not require such guarantees of safety, we quoted from a Tenth Circuit Court of Appeals case and held that:

When a job requires a small amount of skill and training and the consequences of hiring an unqualified applicant are insignificant the courts should examine closely any pre-employment standard or criteria which discriminates against minorities. In such a case, the employer should have a heavy burden to demonstrate to the court's satisfaction that his employment criteria are job related.

Id., citing, Spurlock v. United Airlines, 475 F.2d 216, 219 (10th Cir.1972). Accord EEOC v. Ball Corp., 661 F.2d 531, 541 & n. 20 (6th Cir.1981) (Regarding jobs which do not affect public safety, defendant in rebuttal must show "that use of the employment condition or selection device is compelled by business necessity").

This approach is premised on the argument that there is less justification for difficult employment standards where the job involved is neither particularly complex nor likely to involve the public welfare. Thus, standards in these more basic jobs require even stricter scrutiny. Other circuits have also drawn this kind of distinction. See e.g. Davis v. City of Dallas, 777 F.2d 205, 210-11 (5th Cir.1985), cert. denied, 476 U.S. 1116, 106 S. Ct. 1972, 90 L. Ed. 2d 656 (1986) (discussing cases where diploma requirements were found not to be job related for factory workers but were job related for police officers); Walker v. Jefferson County Home, 726 F.2d at 1558 (court examines the risk and skill required); Guardians Assn of New York City v. Civil Service, 630 F.2d 79, 93 (2d Cir.1980), cert. denied, 452 U.S. 940, 101 S. Ct. 3083, 69 L. Ed. 2d 954 (1981). Clearly, the clerical positions at issue in this case require a stringent scrutiny of Atlas' job relatedness justifications.12

In Black Law Enforcement Officers Assn v. City of Akron, this Court spoke of a defendant's obligation:

The defendant may meet this burden by establishing that the procedure used measures important skills, abilities, and knowledge that are necessary for the successful performance of the job. See 29 C.F.R. Sec. 1607.5 (1986).

824 F.2d at 482. This case makes it clear that there must be a measurable correlation between how people perform on a test and their performance level at the job position.

In the instant case, the EEOC correctly argues that the trial judge never applied the correct legal standards when he concluded that Atlas had proven job relatedness. In a key section of his Opinion, he wrote:

While no formal validation study was performed at Atlas itself, it is obvious that any attempt to validate the Wonderlic test on a clerical workforce as small as the one at Atlas would have been statistically meaningless. Much evidence was offered to show that the use of a cognitive ability test as an applicant screening device is more reliable than any other single predictor of job success. The Court is satisfied that Atlas effectively rebutted the prima facie case and that there was nothing to suggest that the use of the Wonderlic test was a mere pretext for racial discrimination. A close review of the hiring decisions made over the many years at issue shows a concerted attempt to hire the most qualified employees. While there were a few black applicants who met Atlas' stringent standards, they were never passed over for less well qualified white applicants.

680 F.Supp. at 1189 (emphasis added).

The most obvious problem with this reasoning is that the trial judge failed to discuss whether the manner of Atlas using the Wonderlic test (not the "cognitive ability" tests) was manifestly related to a particular job (not "job success" generally). This is particularly important in a clerical job as discussed above.13 Given that the trial judge applied the wrong standards for job relatedness, the EEOC was not obliged to prove that alternative tests with less adverse impact could have been used by Atlas--the third stage of the burden of proof.14

In addition, it is my belief that the trial judge also erred in assuming that Atlas attempted to hire the most qualified people on the basis of a claim that the Company usually picked the person with the best test score. The very question at issue in this disparate impact controversy is whether the testing procedure, which was used by Atlas, is a permissible method of measuring job qualifications under Title VII. Since the legality and the legitimacy of the Wonderlic Test is at issue, the trial judge should not have used the scores, which were obtained by applicants on the test, to evaluate legitimacy. Such logic begs the question rather dramatically. This Court held in Williams v. Vukovich, 720 F.2d 909, 924 (6th Cir.1983), that, "Ranking candidates according to their performance on a non-job related examination does not necessarily result in the selection of qualified candidates."

In Pullman-Standard v. Swint, 456 U.S. at 292, 102 S.Ct. at 1792, the Supreme Court held that "where findings are infirm because of an erroneous view of the law, a remand is the proper course unless the record permits only one resolution of the factual issue." In the instant case, we have already determined that the trial judge misapplied the legal standard regarding "job relatedness." However, a remand is not necessary because Atlas' validity generalization theory is sufficiently deficient that a trial court could not reasonably conclude that the Company's use of the tests was job related even if the correct legal standard had been applied.15 Several appellate courts have held that:

Where the plaintiff has established a prima facie case, and the defendant fails to show the job relatedness of an examination with discriminatory impact, the issues are not open for redetermination on remand.

Nash v. Consolidated City of Jacksonville, 837 F.2d 1534, 1539 (11th Cir.1988). See also Bunch v. Bullard, 795 F.2d 384, 395 (5th Cir.1986); Crawford v. Western Electric Co., Inc., 745 F.2d 1373, 1386 (11th Cir.1984).

In Harless v. Duck, 619 F.2d at 616 n. 5, this Court discussed test validation:

The employer may use any of three methods to validate an examination: criterion validity, construct validity, or content validity.... Criterion validity compares evaluations of employees' job performances with test scores. Construct validity determines the degree to which applicants possess traits which are considered important for the job. Content validity assesses the ability of the applicants to perform specific tasks which must be performed on the job.

Traditionally, general intelligence tests have been validated by construct validation methods because instrumentalities such as the Wonderlic Test "necessarily measure for an inferred ability, regardless of the context." Guardians Assn. of New York City v. Civil Service Commission, 630 F.2d at 94 n. 13 (specifically referencing the Wonderlic test). Gillespie v. State of Wisconsin, 771 F.2d 1035, 1040 n. 3 (7th Cir.1985). In this case, the validity generalization studies involved criterion related studies. The premise of the validity generalization theory, as advocated by Atlas' expert, John Hunter, is that intelligence tests are always valid.

The first major problem with a validity generalization approach is that it is radically at odds with Albemarle Paper Co. v. Moody, supra, Griggs v. Duke Power Co., supra, relevant case law within this circuit, and the EEOC Guidelines, all of which require a showing that a test is actually predictive of performance at a specific job. The validity generalization approach simply dispenses with that similarity or manifest relationship requirement. Albemarle and Griggs are particularly important precedents since each of them involved the Wonderlic Test.

A test may be used in jobs other than those for which it has been professionally validated only if there are "no significant differences" between the studied and unstudied jobs. 29 CFR Sec. 1607.4(c)(2). The study in this case involved no analysis of the attributes of, or the particular skills needed in, the studied job groups. There is accordingly no basis for concluding that "no significant differences" exist among the lines of progression, or among distinct job groupings within the studied lines of progression. Indeed the study's checkered results appear to compel the opposite conclusion.

422 U.S. at 432, 95 S. Ct. at 2378-79. Thus, the Supreme Court concluded that specific findings relating to the validity of one test cannot be generalized from that of others. The Albemarle result is particularly striking because the jobs at issue appeared to be very similar since they only involved different points of progression. Nevertheless, the absence of any job analyses made the studies inadequate. See also Watkins v. Scott Paper Co., 530 F.2d 1159, 1188 (5th Cir.), cert. denied, 429 U.S. 861, 97 S. Ct. 163, 50 L. Ed. 2d 139 (1976) (where expert produced no proof that previously studied jobs were similar to the one at issue, the court was "unwilling to presume that all maintenance crafts require the same job skills"); Dickerson v. United States Steel Corp., 472 F. Supp. 1304, 1335 (E.D.Pa.1978), vacated and remanded on other grounds, 616 F.2d 698 (3d Cir.1980) ("Proper job analyses are required if a company seeks to use tests for similar jobs not included in the validation study").16

Atlas' validity generalization theory ignores the teachings of Albemarle by implying that no linkage or similarity between those jobs which had been previously researched by Hunter and those at the Company need be shown. The similarity was simply assumed.17

In Griggs, the Supreme Court held that the defendants had not shown their use of the Wonderlic Test to have been job related. The plaintiffs in Griggs were employees of the "operating" departments of the Duke Power Company. 401 U.S. at 427, 91 S.Ct. at 851. The Court repeatedly described the Wonderlic Test as a purported "general intelligence" test. Id. at 426, 428, 91 S. Ct. at 850, 852. Moreover, the Court, although not disputing the view that the Wonderlic Test measured general intelligence, found it not to have been job related.

The evidence, however, shows that employees, who have not completed high school or taken the tests have continued to perform satisfactorily and make progress in departments for which the high school and test criteria are now used. The promotion record of present employees who would not be able to meet the new criteria thus suggests the possibility that the requirements may not be needed even for the limited purpose of preserving the avowed policy of advancement within the company.

Id. at 432, 91 S. Ct. at 854.

The Supreme Court noted that the abstract measures of general intelligence, which had supposedly been obtained through the test, did not correlate to job performance. It also contrasted the need for a "meaningful study" of the job relationship to the fact that "the requirements were instituted on the Company's judgment that they generally would improve the overall quality of the work force." Id. The similarity between the background of the tests which were adopted in Griggs and this case is striking.

The EEOC has also set forth regulations which permit the use of studies done elsewhere to validate tests at a different site but only on specific findings of distinct similarities in the work.18 The regulations read:

(2) Job similarity. The incumbents in the user's job and the incumbents in the job or group of jobs on which the validity study was conducted perform substantially the same major work behaviors, as shown by appropriate job analyses both on the job or group of jobs on which the validity study was performed and on the job for which the selection procedure is to be used.

29 C.F.R. Sec. 1607.7B(2).

In effect, job analyses must generally be done at both sites in order to ensure that the studies which are conducted elsewhere arise out of similar situations to the site at issue. No such similarity has been found here.

The EEOC regulations provide, in part, as follows:

New strategies for showing the validity of selection procedures will be evaluated as they become accepted by the psychological profession.

29 C.F.R. Sec. 1607.5(A).

The major test for determining whether an expert's theories have become accepted by his colleagues is whether they have received a "significant degree" of recognition or whether the professional literature is essentially supportive of the concept. United States v. Distler, 671 F.2d 954, 961-62 (6th Cir.1981). In this case, John Hunter's approach is fundamentally at odds with the APA's standards, which were generally endorsed in Albemarle, regarding validity generalization.

When adequate local validation evidence is not available, criterion related evidence of validity for a specified test use may be based on validity generalization from a set of prior studies, provided that the specified test-use situation can be considered to have been drawn from the same population of situations on which validity generalization was conducted. (Primary)

Several methods of validity generalization and simultaneous estimation have proven useful. In all methods, the integrity of the inference depends on the degree of similarity between the local situation and the prior set of situations. Present and prior situations can be judged to be similar, for example, according to factors such as the characteristics of the people and job functions involved. Relational measures (correlations, regressions, success rates, etc.) should be carefully selected to be appropriate for the inference to be made.19

In contrast, Hunter's theory requires no proof of "similarity between the local situation and the prior set of situations." The APA standards represent the professionally acceptable standards for statistical studies. He did not satisfy these standards.20 Hunter never visited Atlas, studied the jobs at Company or even read the studies which formed the basis for the validity generalization compilation of Mickey Petty. His credentials as a fellow of the APA does not sufficiently counter this evidence of unacceptability. It is my belief that he has given this Court far too thin a reed upon which to base a radical departure from Albemarle.

The kind of potentially Kafkaesque result, which would occur if intelligence tests were always assumed to be valid, was discussed in Van Aken v. Young, 541 F. Supp. 448, 454 (E.D.Mich.1982), aff'd 750 F.2d 43 (6th Cir.1984). Hunter, who testified as an expert witness for the Plaintiffs in that case, advanced his validity generalization theory in defending certain exams for a fire department. The trial judge then critiqued Hunter's testimony:

He admitted he knew nothing specific about the Detroit Fire Department Academy and its curriculum, and stated that the ability to be a fire fighter depended just as much upon fast reading ability as agility and ability to climb ladders and do other physical tasks that fire fighters perform. He was unable to explain how a person in a wheelchair, who could get the highest grade on a written exam, could be an active fire fighter. He further testified that there was a high correlation between cognitive ability and the ability to be a good lemon picker.

These potential absurdities were exactly what the Supreme Court in Griggs and Albemarle sought to avoid by requiring a detailed job analysis in validation studies. As a matter of law, Hunter's validity generalization theory is totally unacceptable under the relevant case law and professional standards.

There are additional reasons why Atlas cannot demonstrate job relatedness other than deficiencies in the validity generalization theory.21 As a general proposition, a plaintiff can support a disparate impact case through statistics or by demonstrating the existence of substantial disparities in those procedures that are used to make hiring decisions. In EEOC v. Ford Motor Co., 645 F.2d 183, 198 (4th Cir.1981), rev'd on other grounds, 458 U.S. 219, 102 S. Ct. 3057, 73 L. Ed. 2d 721 (1982), the Fourth Circuit Court of Appeals held that the applications of women were "held" for six months and that "loose employment procedures," such as hiring by word of mouth, contributed to the disparities in hiring.

In Senter v. General Motors Corp., 532 F.2d 511, 528 (6th Cir.), cert. denied, 429 U.S. 870, 97 S. Ct. 182, 50 L. Ed. 2d 150 (1976), we discussed five factors that readily contributed to fewer minority promotions such as vague promotional standards, lack of notice of opportunities, and no anti-discrimination safeguards. See also, Greenspan v. Automobile Club of Michigan, 495 F. Supp. 1021, 1033-37 (E.D.Mich.1980); B. Schlei & P. Grossman, Employment Discrimination Law at 1391 (1983).22

The Court of Appeals for the Second Circuit in Guardians Assn of New York City v. Civil Service, supra, held that:

when an exam produces disparate racial results, a cutoff score requires adequate justification and cannot be used at a point where its unreliability has such an extensive impact as occurred in this case.23

In the instant case, the test cutoff score was seven points higher than in Albemarle. The higher cutoff score here clearly disqualified more blacks. The Wonderlic Company recommended scores of 18 or 19 for some clerical jobs and up to 25 for a few other positions. Yet, Atlas, in establishing a score of 25 for every position, justified the figure by asserting that all of its clerical workers were expected to perform multiple functions which, in turn, would provide the Company with a more productive work force and great utility. However, utility is not the benchmark. The benchmark is business necessity. No evidence in the trial record has suggested that a score of 25 was necessary for, or manifestly "related" to, these positions.

Atlas does not support its argument by relying only on Hunter's theory that general intelligence tests are always valid.24 In its brief, Atlas asserts that:

Finally, there is no question that the use of the Wonderlic Test is in compliance with the Uniform Guidelines on Employee Selection Procedures. Even were one to reject completely the fact of validity generalization, Dr. Petty did review the job descriptions of the clerical positions at Atlas, and he concluded that they were sufficiently similar to the positions examined in the other validation studies of the Wonderlic test.25

This argument is flawed, in that the job descriptions, which were reviewed by Petty, fell way below accepted norms. The interviewed employees at Atlas were simply asked to describe their job responsibilities. In Albemarle Paper Co. v. Moody, 422 U.S. at 433, 95 S.Ct. at 2379, job analyses were found to be inadequate when conducted by supervisors because of vagueness and bias. The absence of a careful job analysis, which evaluates job descriptions as well as assessments of the relative importance of various job tasks, has been emphasized by several courts in rejecting validation studies. James v. Stockham Valves & Fittings Co., 559 F.2d 310, 339 (5th Cir.1977); U.S. v. City of Chicago, 549 F.2d 415, 431 (7th Cir.1977); Rogers v. Int'l Paper Company, 510 F.2d at 1351; Dozier v. Chupka, 395 F. Supp. 836, 852 (S.D.Ohio 1975). Those cases where criterion related studies have been accepted invariably include such a detailed job analysis. See e.g. Bernard v. Gulf Oil Corp., 841 F.2d 547, 565 (5th Cir.1988) (supervisors rated importance of 117 duties and 37 abilities when performing criteria analysis).

Nonetheless, I do not base my conclusion on the problems with these very informal job descriptions.26 Even if the job description is sufficient to show a similarity of the study jobs to Atlas, it still faces the overwhelming problem that nearly a full half (48%) of the employees, who were hired during the period in question, did not take the test or failed to achieve a score of 25 before being hired. Atlas explains this action by claiming that immediate hiring decisions were often necessary. Yet, this is an astoundingly large and unsubstantiated number of sudden decisions.

Assuming, arguendo, that the validity generalization studies have similarities to Atlas' jobs, it only has similarities to about 50% of the jobs because the remaining half were hired without regard to the test requirement. Such an arbitrary hiring method cannot be found to be similar or validated by this metastudy. The adverse impact of the method used cannot be shown to be job related when the job relationship study only applies to about 50% of the hired employees.

In Walston v. County School Board of Nansemond Cty., Va., 492 F.2d 919, 927 (4th Cir.1974), the Fourth Circuit Court of Appeals offered various criticisms on the use of a general intelligence test by the defendant:

it cannot be used as a tool of discrimination; it cannot be used without proper validation and job analyses ... Nor can it be administered capriciously in derogation of the guidelines promulgated by the ETS so as to have a facially disparate impact.

In the instant case, the capricious administration of the test makes validation impossible.27

In summation, it is my belief that no remand is necessary in this case because Atlas' efforts to show job relatedness fall far below the applicable Supreme Court, Sixth Circuit Court of Appeals and professional guidelines. Consequently, I believe that the conclusion of the trial judge, who determined that disparate impact had not been established, should be reversed without remand.

Unlike the disparate impact case, the EEOC must show discriminatory intent in order to prevail on this disparate treatment claim.28 Teamsters v. U.S., 431 U.S. at 335 n. 15, 97 S. Ct. at 1854 n. 15. The trial judge's decision is subject to the clearly erroneous level of review. A plaintiff bears the initial burden of demonstrating a prima facie case. In Teamsters, the Supreme Court held that:

because it alleged a systemwise pattern or practice of resistance to the full enjoyment of Title VII rights, the Government ultimately had to prove more than the mere occurrence of isolated or "accidental" or sporadic discriminatory act. It had to establish by a preponderance of the evidence that racial discrimination was the company's standard operating procedure--the regular rather than the unusual practice.

Id. at 336, 97 S.Ct. at 1855.

A common method, which has been utilized by a plaintiff to satisfy his prima facie burden of proof, is to offer statistical evidence of disparity. The Supreme Court in Hazelwood School District v. United States, 433 U.S. 299, 307-308, 97 S. Ct. 2736, 2741-2742, 53 L. Ed. 2d 768 (1977) held that "gross statistical disparities" need be shown in the disparate treatment context to show discriminatory motive. This requirement is higher than in the disparate impact context. Where other evidence of discriminatory circumstances, incidences, or history are shown in addition to the statistics, the grossness of the disparity need not be as great. Bernard v. Gulf Oil Corp., 841 F.2d at 568. In Teamsters v. U.S., 431 U.S. at 338, 97 S.Ct. at 1856, over forty specific other instances were in the record.

Once the EEOC has met its prima facie burden, the defendant must rebut it by attacking statistics or by providing a "non-discriminatory explanation for the apparently discriminatory result." Payne v. Travenol Laboratories, Inc., 673 F.2d 798, 817 (5th Cir.) cert. denied, 459 U.S. 1038, 103 S. Ct. 451, 74 L. Ed. 2d 605 (1982). In the instant case, Atlas' efforts to discredit the EEOC's statistics have already been rejected. Moreover, the EEOC has also enumerated many specific instances of disparate treatment to supplement the statistics:

1. 10 or 11 white applicants were able to take a second typing test.

2. Several older white applicants received age additions to their test scores.

3. Many exceptions to the test requirements were made and only white applicants benefited from the exceptions.

4. Racial notations existed on some black candidates applications.

The EEOC has clearly met its prima facie burden on the basis of the statistics and these particular instances.

In response, Atlas relies upon the trial judge's conclusion that no individual hiring decision was racially motivated (only the most "qualified" were hired) and by asserting that none of the individual incidents made a difference in the hiring determination.

Having carefully examined the trial judge's ruling, I believe it to be clearly erroneous. Atlas never offered a nondiscriminatory explanation for the individual incidents of disparate treatment. Although Atlas asserts that these incidents had no effect upon the racial composition of the work force, that explanation does nothing to remove the inference that racial discrimination was responsible for the differential treatment. No other reason has been offered. Although these incidents alone might not rise to the level of a racially discriminatory pattern or practice, they certainly do when combined with the testing practices, and the statistics.

Atlas' assertion that no racial motivation was involved since only the most "qualified" were hired is premised on the assumption that the best performance on the test is synonymous to being the most qualified. Even assuming, arguendo, that the highest scores were the governing criteria and established an applicant's qualification to work, Atlas departed from using the test in hiring on 48% of the hired workers, all of whom were white. Tr. 176-177. When the statistics on the test are combined with the strikingly different practices used for hiring black workers as compared to white applicants, it is my conclusion that the trial judge clearly erred in holding that the EEOC had not proven a pattern and practice of discriminatory hiring, and his determination on this claim should be reversed without remand, as well.