Shafer v. Commander, Army & Air Force Exchange Service

667 F. Supp. 414 (1985)

Neoma SHAFER, Plaintiff,
v.
COMMANDER, ARMY and AIR FORCE EXCHANGE SERVICE, et al., Defendants.

No. CA 3-76-1246-R.

United States District Court, N.D. Texas, Dallas Division.

June 26, 1985.

Steven B. Thorpe, Law Offices of James C. Barber, Genice A.G. Rabe, Mullinax, Wells, Baab & Cloutman, P.C., Brian Eberstein, Dallas, Tex., for plaintiff.

Elizabeth Pugh, Judith F. Ledbetter, Don J. Mros, Patrick Sorek, Sharon L. Reich, Richard Greenburgh, Brook Hedge, Alan L. Ferber, Dept. of Justice, Civ. Div., Washington, D.C., John Albach, Sp. Master, Dallas, Tex., for defendants.

MEMORANDUM OPINION

BUCHMEYER, District Judge.

This is an employment discrimination case brought by the plaintiff, Neoma Shafer, against the defendant, the Army and Air Force Exchange Service ("AAFES").^[1]

Shafer claims that AAFES has refused to promote her because she is female, and has also discriminated against her in training and job assignment. In addition, Shafer alleges that AAFES discriminates against all female employees at its headquarters in Dallas with respect to promotional advancement opportunities, training opportunities and job assignment.^[2] On February 3, 1982, following a class certification hearing, this Court certified Neoma Shafer as the representative, under Rule 23(b)(2), of the following class:

"All females who, since April 1, 1972, have been or who are now employed by the defendants at their AAFES Headquarters Facility, Dallas, Texas (including the Fashion Distribution Center), and who have been discriminated against on account of sex with respect to: promotional and advancement opportunities; training opportunities; and job assignment."

A "Phase I" trial of the class liability issues was held in June of 1983. The filing of post-hearing briefs was completed by August 1983, with some letter-briefs addressing recent case developments being filed by February 20, 1984. (See footnote 1.) The purpose of this opinion is to resolve the class liability issues and all pending motions.^[3] Specifically, it holds:

(i) that the AAFES "mobility requirement" for all UA positions above grade 12 is discriminatory because of its disparate impact upon females;

(ii) that AAFES has not discriminated against female employees with respect to training;^[4]

(iii) that AAFES has discriminated against females by treating them disparately, and less favorably than men, with respect to initial placement and promotion to UA jobs and to HPP jobs (grades 6-11);

(iv) the motion by Shafer to strike certain tables attached to the AAFES post-trial briefs is moot; and

(v) the motion of AAFES to decertify the class action is denied.

1. Factual Background

a. The Plaintiff

Neoma Shafer has been employed by AAFES, at its headquarters in Dallas, since May of 1968. She is currently a merchandising clerk — and has held this same job during most of her employment with AAFES (although, as discussed below, she has unsuccessfully applied for promotion on several occasions).

b. The Defendant

The Army and Air Force Exchange Service is an instrumentality of the Department of Defense. It supplies various goods and services to members of the Army and the Air Force (and their dependents) through a world-wide network of retail establishments known as post exchanges (or "PX's").

AAFES employs approximately 60,000 persons throughout the world. At its headquarters, which are located in Dallas, AAFES employs some 2,000 persons. Approximately 65% of the employees at the AAFES Headquarters are female.

c. Job Categories

There are three catagories of employees at AAFES: (i) UA-EMP — "Universal Annual Executive Management Program" employees (grades 13-17); (ii) UA — "Universal Annual" employees (grades 5-12); and (iii) HPP — "Hourly Pay Plan" employees (grades 1-19).^[5]

UA and UA-EMP employees are paid an annual salary; the pay scale is adapted from the federal civil service pay scale. HPP employees are paid by the hour; their wage is determined by the prevailing rates in the area of each facility.

UA and UA-EMP employees are considered to be employees of AAFES' worldwide organization, and their job assignments are determined by the AAFES headquarters in Dallas. HPP employees are considered to be employees of the particular AAFES facility where they work; they work in three broad job classifications: (i) Administrative Support (grades 1-7); (ii) Patron Services (grades 1-7); (iii) Craft and Trade (grades 1-19).

d. Performance Evaluation Reports

All employees of AAFES are evaluated periodically — usually this is done annually—by Performance Evaluation Reports ("PER"). These performance reviews are prepared by the employee's immediate supervisor, and then approved by the secondline supervisor.

An employee who objects to the rating on the PER may submit written objections to it. However, the approving supervisor is not required to discuss the objections or to afford the employer a hearing concerning the PER. The approving supervisor does furnish a final reply to the employee, and the employee's objections and the final decision are then filed with the PER in the individual's personnel folder. The employee has no right to appeal an unsatisfactory evaluation.

There is, necessarily, a large degree of subjective judgment used by the supervisors in completing and approving these Performance Evaluation Reports.^[6] An unsatisfactory or marginal evaluation does adversely affect the promotion of the employee.

e. Promotions

AAFES has a "promotion from within" policy; accordingly, job openings are filled through promotion unless there are no AAFES employees qualified for the vacant position. There are no real lines of progression for promotion purposes; AAFES considers all qualified persons to be eligible for promotion.^[7] The evidence established that from 80-90% of the promotions at AAFES are filled by employees who are within 2 or 3 grade levels below the vacant position.

A vacancy posting procedure applies with respect to most of the HPP and nonsupervisory UA employees at the AAFES headquarters. Candidates for promotion must meet the minimum qualifications listed in the job description sheets for the posted vacancy. Employees who submit applications are then screened and ranked on the basis of three criteria: performance, potential, and length of service. Of these, performance — as reflected in the periodic Performance Evaluation Reviews — is most important. The top three candidates are then referred to the supervisor with the vacancy for final selection. There are no grievance rights for employees who were not referred for the final interviews or who were not selected for the open position.

A promotion roster is used with respect to all other promotions at the AAFES headquarters.^[8] An employee is placed upon the promotion roster only upon the recommendation of the immediate supervisor. These employees are also ranked according to three criteria of performance, potential and length of service. Specifically:

(i) Promotions into the HPP and nonsupervisory UA positions specifically excluded from the vacancy posting procedure are filled from the promotion roster composed by the Headquarters Personnel Division, with the final decision made by the supervisor with the job opening.

(ii) Promotions involving the supervisory and professional UA positions are handled by the Headquarters AAFES Personnel Division, Career Management Branch, which conducts a "worldwide search" by computer of qualified employees within 2-3 grades below the open position^[9] and which creates a promotion roster ranking them according to performance, potential and length of service — and, if appropriate, mobility.^[10] Final decisions are made by the Director of the Personnel Division.

(iii) A promotion board reviews all promotions involving UA-EMP positions, and makes recommendations concerning these high-level employees to the Commander of AAFES, who makes all final decisions concerning these promotions.

There are no grievance rights for employees who are not placed on a promotion roster or who are not selected for a vacancy.

f. The Mobility Policy

AAFES has a "mobility policy" which requires all UA employees to make a determination as to whether they are "mobile" or "nonmobile" by completing a "Statement of Mobility" (Def. Exh. 5). A person who is mobile agrees to "accept assignment and transfer to any location throughout the world-wide Exchange System." Furthermore, "this mobility obligation is unqualified" and a mobile employee may be required to serve at locations where dependents are not allowed. And, refusal to accept a transfer on the part of a mobile employee "could result in ... separation from AAFES employment for the reasons of declination of transfer."^[11]

In order to encourage its employees to be affirmatively mobile, AAFES provides certain benefits to mobile employees which are not available to non-mobile employees. These include preference with respect to vacations, preference in case of reductions in force, free travel to a place of residence upon retirement, and preference for promotions over non-mobile employees when the three criteria (performance, potential, and length of service) are equal.^[12] In addition, an employee who is mobile expands his own range of opportunities by being available for promotions worldwide — since no AAFES employee can be a UA 13 (or above) without being mobile.

g. Shafer's Applications for Promotion

In 1975, Shafer — who is a merchandising clerk, an HPP job — applied for promotion to a UA position, that of Quality Assurance Inspector.

There were two vacancies for this position. Of the 98 persons who applied, 82 were female and 16 were male. After the initial screening, 13 applicants — 10 females and 3 males — were referred for interview. Shafer's application was not referred. A female supervisor conducted the final interviews, and selected two males for promotion to the vacant positions.

Shafer has also applied for promotion to other HPP and UA jobs, both before and after 1975, but has never been successful.

2. The Disparate Impact Case

Shafer's "disparate impact" attack involves only the AAFES mobility policy. This is not "an attack upon the mobility system per se," since that policy applies to both male and female employees. Rather, Sbafer contends that "the use of mobility as a job requirement" for all jobs "UA 13 and above" is discriminatory because of its disparate impact upon females.

a. Shafer's Evidence

It is undisputed that affirmative mobility is a minimum qualification for promotion into any UA job which is grade 13 or above.^[13] AAFES witnesses conceded that a person who is not mobile cannot be promoted to a UA-EMP job, no matter how well qualified that person is otherwise. The "mobility policy" is, therefore, an absolute cut factor for UA 13 and above.

Moreover, the evidence established that the mobility requirement did have a disparate impact upon females. The expert witness for AAFES (Dr. O'Dell) testified that the requirement would disqualify a statistically significantly^[14] greater percentage of women than men in the AAFES workforce. And, AAFES witnesses attributed the relative absence of females from the EMP workforce to the mobility requirement.

In addition, the evidence demonstrated the obvious: the predominant form of family in our society is one in which the male is the primary wage earner and the female is the supplemental wage earner. It is, therefore, more difficult for a woman to make an open-ended commitment to mobility — and to being away from her family for extended periods — than for a man to make such a commitment. Where the male is the primary wage earner, he can typically expect his family to relocate with him when he is transferred; however, usually that same man cannot afford to give up his more lucrative job in order to relocate with his wife.^[15]

b. The Controlling Authorities

Since the "mobility policy" does have an adverse impact upon female employees, the next question is whether the requirement has a "manifest relationship to the employment in question" and a "demonstrable relationship to successful performance of the jobs for which it was used." Griggs v. Duke Power Co., 401 U.S. 424, 432, 91 S. Ct. 849, 854, 28 L. Ed. 2d 158 (1971). The mere assumption that the requirement "generally would improve the overall quality of the work force" is not sufficient — nor does "good intent or absence of discriminatory intent" redeem the challenged practice. Griggs, 401 U.S. at 432, 91 S. Ct. at 854.

In this case, since the mobility policy is used to screen promotions into UA-EMP positions, the issue is not whether mobile employees are potentially better performers; rather, it is whether those employees who are not mobile fail to possess the "minimal qualifications" for these jobs. Albemarle Paper Co. v. Moody, 422 U.S. 405, 426, 95 S. Ct. 2362, 2375, 45 L. Ed. 2d 280 (1975); Watkins v. Scott Paper Co., 530 F.2d 1159 (5th Cir.1976), cert. denied, 429 U.S. 861, 97 S. Ct. 163, 50 L. Ed. 2d 139 (1976). This requires a determination of whether the mobility policy is a "business necesity" for UA-EMP jobs. As stated by the Fifth Circuit in Parson v. Kaiser Alum. & Chem. Corp., 575 F.2d 1374 (5th Cir.1978), cert. denied, 441 U.S. 968, 99 S. Ct. 2417, 60 L. Ed. 2d 1073 (1979):

"[T]his doctrine is very narrow. a practice which is demonstrably discriminatory in impact must:

``Not only foster safety and efficiency, but must be essential to that goal. United States v. Bethlehem Steel Corp., 446 F.2d 652, 662, (2d Cir.1971); United States v. Jacksonville Terminal Co., 451 F.2d 418, (5th Cir.1971), cert. denied, 406 U.S. 906 [92 S. Ct. 1607, 31 L. Ed. 2d 815] (1972). In other words, there must be no acceptable alternative that will accomplish that goal ``equally well with a lesser differential racial impact.'" (575 F.2d at 1389) (emphasis added).

c. The AAFES Contentions

The basic claim by AAFES is that the mobility requirement is necessary to assure an adequate source of mobile employees with which to staff its facilities throughout the world. However, this claim is not supported by any study or any empirical evidence.^[16] Moreover, one AAFES witness (Smith) conceded that he was unable to say that opening the UA-EMP jobs to all employees would leave AAFES with an insufficient pool of mobile employees.

In addition, the claim that the mobility requirement is "necessary" to staff its world-wide facilities with UA-EMP employees is inconsistent with the fact that AAFES does not require mobility for UA positions below the level of UA 13. Despite this, AAFES is apparently able to generate a sufficient pool of mobile UA employees to staff its facilities with UA employees at grades 5-12 without requiring mobility at that level. Indeed, the evidence established that AAFES did not have a mobility policy for UA-EMP employees prior to 1958, but was still able to perform its mission satisfactorily.

Nor is the AAFES argument, based upon a period in 1968 when volunteers were solicited for duty at PX's in Viet Nam, a convincing one. Although the testimony was sketchy and far from specific, it did indicate that at least one time AAFES had an insufficient number of volunteers to staff its facilities in Viet Nam.^[17] However, there was absolutely no showing that the shortage existed only in UA grades 13 and above. Therefore, the testimony does not support the mobility policy — and, at best, it merely indicates that there may be certain duty for which even mobile employees will not volunteer.

AAFES also contends that the mobility requirement is necessary because it gives UA-EMP employees background and training outside of AAFES headquarters "to insure that [the] executive level employees gain the proper understanding of [its] worldwide mission in support of the United States Armed Forces." Contrary to this, the final witness for AAFES was Rosalee LeFleur, the highest graded female in the history of AAFES. Between 1969 and 1982, she rose in the UA-EMP ranks from UA 12 to her present grade of UA 16 — and during this entire period she was assigned exclusively to AAFES headquarters in Dallas and she performed in an exemplary manner. Therefore, it is obvious that "background and training" outside AAFES headquarters is not a necessary requirement for UA-EMP jobs.

Finally, as Shafer correctly argues, there was no evidence presented which related the mobility requirement to any specific UA-EMP job. There was no proof, therefore, that mobility was related to the performance of the UA positions above level 12. See Griggs, 401 U.S. at 432, 91 S. Ct. at 2378; Albemarle Paper, 422 U.S. 405, 95 S. Ct. 2362, 45 L. Ed. 2d 280. What was proven was the obvious fact that AAFES believes it is simpler and easier to require all UA-EMP employees to be mobile than (i) to create a volunteer pool of mobile UA-EMP employees by offering promotions and other incentives, or (ii) to create a volunteer pool of mobile UA-EMP employees for vacancies as it does other UA employees.^[18]

d. Conclusion

Under the evidence, the mobility policy is not a business necessity, and it is not essential to the mission of AAFES or to the successful performance of UA-EMP jobs.^[19] Accordingly, Shafer is entitled to a judgment of liability on this issue.

3. The Disparate Treatment Case

Shafer's "disparate treatment" case involves the claims that AAFES has treated females less favorably than males with respect to initial placement and promotion. This attack requires proof of "intent to discriminate." Wilkins v. University of Houston, 654 F.2d 388 (5th Cir.1981); Pouncy v. Prudential Insurance Co., 668 F.2d 795 (5th Cir.1982).

a. The Controlling Authorities

In a class action challenging an employer's entire promotion procedures, the plaintiff must establish that "discrimination was the employer's standard operating procedure — the regular rather than the unusual practice." International Brotherhood of Teamsters v. United States, 431 U.S. 324, 97 S. Ct. 1843, 52 L. Ed. 2d 396 (1977); Hazlewood School District v. United States, 433 U.S. 299, 97 S. Ct. 2736, 53 L. Ed. 2d 768 (1977).

Statistical evidence is very relevant in a disparate treatment case attacking promotion practices. In a proper case, a court may infer racial discrimination if gross statistical disparities are shown in analyzing relevant statistics. Lewis v. NLRB, 756 F.2d 882 (5th Cir.1985). In probing discriminatory intent, a trial court may examine history of the employer's practices, anecdotal evidence of class members, and the degree of opportunity to treat employees unfairly in the appraisal process. Carroll v. Sears, Roebuck & Co., 708 F.2d 183, 190 (5th Cir.1983); Pouncy v. Prudential Ins. Co. of America, 668 F.2d 795 (5th Cir.1982). In order to rebut a prima facie case of disparate treatment, the defendant must discredit the plaintiff's evidence or provide a non-discriminatory explanation for the apparently discriminatory result. Page v. U.S. Industries, Inc., 726 F.2d 1038 (5th Cir.1984).

The Fifth Circuit has repeatedly warned of the need for considered and refined statistical analysis in the context of Title VII actions. Pegues v. Mississippi State Employment Service, 699 F.2d 760 (5th Cir. 1983). It has also recognized that, in this analysis, "a difference between expected and observed values in excess of two or three standard deviations at the 5% significance level permits an inference of discrimination." Pegues, 699 F.2d 768, at footnote 9; Rivera v. City of Wichita Fallas, 665 F.2d 531 (5th Cir.1982).

b. The Data Base

Unlike the usual discrimination suit involving complex statistical evidence, the parties in this case worked together to create the data base which the expert witnesses for both sides analyzed for purposes of exhibits and expert opinions. Indeed, the pretrial order for the class liablity trial stated:

"The parties stipulate to the form, substance and accuracy of the computerized data base used in this action, which was jointly prepared by the experts and counsel for the parties from the official files and records compiled and maintained by the defendant."

This cooperation by the attorneys has greatly simplified this Court's obligation to make a "considered and refined statistical analysis" of the exhibits and expert testimony concerning class liability. Pegues, 699 F.2d 766.

c. Non Parametric Analyses

Both the plaintiff's expert, Dr. Michael, and the AAFES expert, Dr. Odell, performed non parametric analyses^[20] to test the hypotheses that males and females are selected for promotion by AAFES at the same rate. Proper analysis shows that, according to these studies by both experts, there is a gross statistical disparity adverse to females with respect to promotions by AAFES.

(i) Dr. Michael — UA Promotions

In plaintiff's exhibits 6 and 7, Dr. Michael analyzed promotions "using the grade prior to promotion as the pool for analysis." In doing so, he first determined the percentage of males and females in a given grade during a particular year — then compared these percentages with the percentages of males and females among those in the given grade who received promotion — and, by so doing, controlled for both year and grade of promotion.

By exhibit 6, Dr. Michael concluded that AAFES promotes women at a significantly lower rate than men. Specifically, the actual number of female promotions is 3.66 standard deviations below the number which would be expected if women were treated like men with respect to promotion.^[21] The probability of this result occuring by chance is 0 to 3 decimal places. Therefore, this is a grossly significant result adverse to females which, under controlling authority, establishes a prima facie case of discrimination. See Rivera, 665 F.2d at 545, footnote 22.

Moreover, exhibit 6 also demonstrates that the overall significance is not attributable to isolated instances — because it reflects that women received less than their proportional share of promotions in eight of the ten years examined and in nine of eleven UA grades. This indicates that the 3.66 standard deviations below the expected number of female promotions is not a product of a few years or a few job grades and, consequently, it adds weight to the summary result adverse to females.

Similarly, exhibit 7 — which differs from Exhibit 6 only in that it analyzes promotions from HPP grades 6-11 — demonstrates that women have received significantly fewer promotions than would be expected if they were treated as favorably as males. The result is 2.36 standard deviations low and has a probability of .009. Again, this cannot be attributed to isolated instances, instead of a pattern of consistent underpromotion, because women received less than their proportional share of promotions in each year analyzed and in each grade except one.

(ii) Dr. Odell — UA Promotions

In defendant's exhibit 7, Dr. Odell analyzed UA promotions using a methodology which differed from Dr. Michael's in one important respect: While Dr. Michael used grade prior to promotion as an availability, Dr. Odell chose to analyze promotion into rather than from grade. Therefore, Dr. Odell selected appropriate availabilities or "pools" of candidates for promotion into grades.^[22]

Although exhibit 7 contains eight different UA promotion analyses, Dr. Odell described page 8 as the "master analysis." This analysis shows (i) that AAFES promoted 81.6 less females than would be expected if defendant promoted women at the same rate as men; (ii) that the result adverse to women is statistically significant; and that this is not the result of isolated problems, because women were under promoted in 8 of 11 categories analyzed (or 73%).

On pages 9, 10, 18 and 19 of exhibit 7, Dr. Odell analyzed the same data presented on the "master analysis" (page 8), but he disaggregated it by mobility. Although Dr. Odell could not satisfactorily explain why he did these disaggregations by mobility^[23]—when asked this on cross-examination, Dr. Odell simply suggested that the Court look to page 8 as the "master analysis" — the results on pages 9, 10 and 19 are, overall, statistically significant and adverse to females. The result on page 18 is not but, on cross-examination, Dr. Odell admitted that this page was totally inaccurate.

Therefore, the UA promotion analyses by Dr. Odell actually support the expert opinion by Dr. Michael — that the non-parametric analyses demonstrate that AAFES has discriminated against UA female employees with respect to promotion.^[24]

(iii) HPP Promotions

By plaintiff's exhibit 6, Dr. Michael also concluded that AAFES promotes women into the higher level HPP jobs (grades 6-11) at statistically significantly lower rates than men, and that the significance could not be attributed to isolated instances. The actual number of female promotions is 2.36 standard deviations which would be expected if women and men were treated equally, and the probability of this occurring by chance is .009.^[25]

On page 6 of defendant's exhibit 7, Dr. Odell reached a different conclusion. However, this Court rejects Dr. Odell's conclusion for the following reasons:

... the Odell analysis required him to establish "availability pools"; however, unlike his other availability pools (see pages 2, 3, 4 of exhibit 7), page 6 uses a "mixed availability"; and Dr. Odell admitted that mixed availability was not supproted by any analysis.

... Dr. Odell conceded that he attempted to analyze the promotions in these grades based upon availability pools of 1 and 2 grades below the one in question; it was only after he determined that each of these methods produced results favorable to the plaintiff that his notion of mixed availability was used.

As shown by plaintiff's exhibit 11, any consistent use of either 1 or 2 grade availability produces results that are grossly significant and adverse to females regarding promotions to the higher level HPP jobs. Therefore, this Court rejects Dr. Odell's HPP analyses and concludes, based upon Dr. Michael's analysis, that the plaintiff has established a prima facie case of discrimination concerning promotions of females to HPP grades 6-11.

(iv) Practical Significance

AAFES attacks the plaintiff's statistical analysis on the basis that, although they may show "statistical significance," they have no "practical significance." In this regard, AAFES argues that the plaintiff's evidence merely shows (i) that it "promoted 8.6 too few females" in HPP categories 6-11 during the ten-year period analyzed, an average of less than 1 per year (see footnote 25), and (ii) that it promoted 44.3 too few UA females in ten years, an average of only 4.43 per year.

This argument misconstrues the purpose of Title VII and the "statistical evidence" decisions cited above. No case has been cited by AAFES, and this Court has found none, which holds that results which are statistically significant — and which establish a prima facie case of discrimination because of gross statistical disparities — will not support a finding of liability because the absolute shortfall is too small. Instead, liability is founded upon levels of statistical significance which compel the inference that sex was a factor in decision making. Thus, while the absolute shortfall may affect remedy, it does not defeat liability.^[26]

More importantly, the absolute shortfall of female promotions to UA positions or HPP grades 6-11 is not small, when it is properly understood. The relevant question is not "how do women fare?" It is "how do women fare relative to men?" This requires a comparison of expected female promotions to female promotions actually received.

For example, Dr. Odell's "master analysis" (exhibit 7, page 8) shows an absolute shortfall of 81.6 UA female promotions over a ten-year period. However, the percentage of promotions expected for females is 635/716.6 = 88.6%. Thus, women receive only 88.6% of those promotions they should receive. In contrast, there were 906.4 expected male promotions. Thus, the percentage of expected male promotions is 988/906.4 = 109%. Therefore, while women receive only 88.6% of the promotions they deserve, men receive 109% of the promotions they deserve.

The difference, calculated in this manner, is 20.4%. Thus, a randomly selected female employee at AAFES is at a 20.4% disadvantage in the promotion process when compared with a randomly selected male, controlling for grade and year. Such a difference represents a significant disadvantage for the female employees of AAFES, and is obviously one to which Title VII is surely addressed.

(v) Claimed Defects in Analyses

AAFES makes three basic attacks upon the statistical analyses (including some of those done by its own expert, Dr. Odell).^[27] None of these are valid.

Qualifications. AAFES first claims that the plaintiff's UA promotion analysis "failed to control for differing qualifications" of the employees. This attack is baseless. Neither expert assumed that there were uniform qualifications throughout the workforce; instead, each used reasonable methods to control for differences in qualifications. Specifically:

... Dr. Michael analyzed the promotion rates of males and females within job grades. Rather than assuming that all employees in the work force were equally qualified,^[28] Dr. Michael specifically allowed for different grades — and assumed that the distribution of qualifications within UA grade is the same for males and females. This means that Dr. Michael recognized that some employees may be better qualified than others within a particular grade, he also assumed that there was no reason to suppose that men were disproportionately more qualified than women.^[29]

... Dr. Odell, the AAFES expert, also controlled for qualifications. Instead of assuming that all employees were equally qualified, he analyzed promotions into rather than from grades, and selected appropriate "pools" of candidates for promotion into each particular grade. In this manner, Dr. Odell allowed the defendant's own promotion practices to identify the actual pool of qualified candidates for promotions to each grade. And, his analysis is consistent with Dr. Michael's in that it detected a grossly significant underpromotion of females.

Worldwide Workforce. AAFES next claims that the statistical analyses "fail to properly account for the relationships between the defendant's worldwide and Headquarters workforces and failed to account for the impact of that relationship upon the promotion data assembled for the Headquarters workforce." This argument is baseless. The class certified includes only females at the AAFES headquarters. Therefore, both experts properly confined their analyses to headquarters personnel and their relative success in achieving promotions at headquarters. To do otherwise would have been inaccurate since the evidence demonstrated that considerations concerning promotion of an AAFES employee overseas would not necessarily apply to a headquarters employee.

Aggregation of Data. AAFES also claims that the plaintiff's analyses created the appearance of discrimination by improperly aggretating statistics — and that the promotion analyses should be considered on a disaggregated basis.^[30] This argument, too, is baseless. Neither expert, either Dr. Michael or Dr. Odell, criticized the use of aggregation in statistical analysis. Instead, both presented analyses which summarized the experience of females at AAFES headquarters with respect to promotions over a period of ten years — and which demonstrate that females received less than their proportional share of promotions, that this result was statistically significant, and that it could not be attributed to isolated instances in a few years or a few job grades.^[31] Moreover, the identical argument that AAFES makes here concerning aggregation of data was rejected by the Fifth Circuit in Capaci v. Katz & Besthoff, Inc., 711 F.2d 647 (5th Cir.1983):

"Finally, K & B attempted to demonstrate that there was no statistically significant evidence of discrimination when the data was broken down by city or year or both. In our view, this was unfair and obviously attempted to disaggregate that data to the point where it was difficult to demonstrate statistical significance. By fragmenting data into small sample groups, the statistical tests become less probitive. Wheeler v. City of Columbus, Mississippi, 686 F.2d 1144, 1151 (5th Cir.1982). Indeed, it became impossible [to] demonstrate significance with such small numbers in many instances, since even a record of hiring or promoting zero women would not yield statistically significant results. There was no reason to fragment the data geographically, since all hiring decisions regarding the manager trainees are made at the firm's New Orleans' office, and liability is premised on ``across the board' practices, occurring at all locations. Likewise, liability is premised on actions taken within the entire 1965 to 1970 period, and there was no reason to fragment the data by year. Aggregating the data as Plaintiff did was a much more reasonable approach, since liability under Title VII depends on whether the EEOC demonstrated a system-wide pattern or practice of disparate treatment, rather than ``the occurrence of isolated or accidental or sporadic discriminatory acts. It had to establish ... (that) discrimination was the company's standard operating procedure — the regular rather than unusual practice.' International Brotherhood of Teamsters v. United States, 431 U.S. 324, 336 [97 S. Ct. 1843, 1855, 52 L. Ed. 2d 396] (1977)."

Accordingly, the attempts made by AAFES to discredit the statistical analyses are baseless, and they are rejected.^[32]

d. The Regression Analyses^[33]

Both of the experts, Dr. Michael and Dr. Odell, performed regression analyses.^[34] Both also testified that, while regressions are done in terms of pay, they also contain information about other issues — since a person's pay at any give time is equivalent to initial placement plus promotion. Accordingly, the regression analyses also show that there is a gross statistical disparity adverse to UA females^[35] with respect to promotions by AAFES.

(i) Dr. Michael's Regression

Dr. Michael testified about a series of five regression analyses (plaintiff's exhibits 1-5). Plaintiff's exhibits 1 and 3^[36] were of the entire UA work force controlling for education, experience and tenure. The value of the female coefficient (or beta) is given under "b" and the significance level in standard deviations as "T."

A review of the betas on these exhibits from 1973 through 1982 clearly indicates that females are payed less at AAFES even after controlling for qualifications in terms of education, experience and tenure. For example, the analysis reveals that being female costs a UA employee from a low of 28.1% in 1973 to a high of 37.1% in 1978 and 1979 when compared to males with equal qualifications in terms of education, experience and tenure. The T scores indicate that the result is grossly significant, with all results more than 12 standard deviations from equality. Therefore, these regression analyses indicate that being female has a substantial, significant, and negative effect on a person's salary with AAFES.

To make certain that these results were not caused by pre-act discrimination,^[37] performed the analysis summarized on plaintiff's exhibit 4. It considered only employees hired by AAFES after January 1, 1973 — and demonstrates that any conduct responsible for the observed differences in male and female salary must have occurred during the class period. The observed beta is grossly significant at the level of 5 standard deviations, and the approximate measure of each female loss is 15.5% per year. Moreover, since differences in salary must be due to differences in initial job assignment or in promotion, this analaysis supports the plaintiff's claim of liability on theories of placement and promotion discrimination.

The next analysis performed by Dr. Michael (exhibit 5) was to detect differences in promotion isolated from placement discrimination. This regression analysis was similar to those summarized in exhibits 1 and 3, except that initial salary was added as an explanatory variable to explain final pay. This means that the betas can no longer be interpreted as percentage differences in pay, but rather as percentage differences in change of pay over time. Thus, the beta for 1974 on plaintiff's exhibit 5 indicates that the average UA female in 1974 was at a 3% disadvantage with respect to the increase in her salary since 1973; and the beta for 1982 indicates that the average UA female in 1982 was at a 7.4% disadvantage with respect to the increase in her salary since 1973. Moreover, since the evidence established that change in pay and promotion are virtually synonymous at AAFES, women in the 1982 work force were at an average disadvantage of 7.4% in promotions as compared to equally qualified men.^[38]

Finally, Dr. Michael compared the analysis in plaintiff's exhibit 4 with that in exhibit 5 "to allow us to understand the way discrimination actually works at AAFES." Exhibit 4 indicates that discrimination in initial assignment and promotion costs the average women in the UA work force 15.5% of her salary after controlling for education, experience and tenure. Then, reviewing the beta for 1982 on plaintiff's exhibit 5, Dr. Michael's opinion was that of this 15.5%, 8.1% was due to discrimination in job assignment and 7.4% was due to discrimination in promotion.

(ii) Dr. Odell's Regression

Dr. Odell, the AAFES expert, analyzed pay in a slightly different manner. Rather than treat male and female pay in a single equation, Dr. Odell first produced a male equation — and then sought to determine the percentage of predicted pay women received, as determined by the male equation. The extent to which women receive more or less than the amount predicted by the male equation was termed the "sex factor" by Dr. Odell. In his analyses, this "sex factor" indicates whether men and women are treated differently, who is treated less favorably, and the magnitude of the difference.

Dr. Odell's first regression is summarized on page 27 of defendant's exhibit 7. It utilizes, among other variables, final grade as a predictor for final salary; and it appears to indicate that women and men in the same grade are payed the same. However, this result is meaningless in this case — because inclusion of final grade necessarily means that differences in placement and promotion are excluded from the result. Therefore, as Dr. Odell admitted from the stand, this analysis contains no promotion information and no placement information, and is not relevant to the issues in this case.

However, the analyses on pages 28-31 of defendant's exhibit 7 are relevant to the claim of promotion discrimination. Here, Dr. Odell performed regressions using starting salary as a predictor of final salary. Thus, his equation controls for starting salary and analyzes the change in pay over time. Since both experts agree that change in pay can be used to model promotion, these are accurate promotion analyses.

Just as in Dr. Michael's regressions, Dr. Odell's analyses indicates that there is a significant difference in the way men and women are promoted, controlling for education, experience and tenure. For example, page 28 expresses the average female salary as a function of "expected female salary." And, it indicates that in every cell analyzed, except 1972-75, actual female salary is below the expected. Since these figures reflect change in salary based on the male equation, they reflect underpromotion of females as compared to equally qualified males. Moreover, careful review of page 30 of defendant's exhibit indicates that the longer a UA female employee is followed, the further she falls behind males. For example, in terms of 1981 salary, an average woman is $11.94 behind after three years, $23.34 behind after six years, and $66.47 behind after ten years. These results clearly indicate that women have been disadvantaged in the promotion process continuously throughout the class period.

(iii) The AAFES Attacks

AAFES argues that Dr. Michael's regression analyses are defective because the low R2's reduce the reliability of inference of discrimination. However, at this stage of the case, it is not necessary to measure the actual magnitude of the "sex effect" (using Dr. Odell's term). Instead, in a Phase I liability trial the question is whether any sex effect exists, since the defendant is required by law to treat males and females equally. Moreover, there was no testimony at trial by either expert critizing the regressions of Dr. Michael because of the size of the R2. This was not the subject either of cross-examination or direct testimony by Dr. Odell.^[39]

AAFES also claims that the analyses by Dr. Michael did not take into account the fact that "defendant's UA employees receive periodic and virtually automatic changes in pay which are unrelated to promotion or placement decisions." This argument is baseless. Both experts used tenure (time with AAFES) as an independent variable to explain pay. Since the changes in pay to which AAFES refers are "periodic and virtually automatic," controlling for tenure also controls for such pay changes. Therefore, as Dr. Michael testified, the differences left after controlling for tenure must be due to changes which are not periodic and automatic — that is, promotions.^[40]

(iv) Summary

The regression analyses performed by both experts, Dr. Michael and Dr. Odell, reflect a consistent and persistent pattern. Specifically, they indicate that women at AAFES are paid less than men after controlling for those variables which should account for pay. Moreover, both experts agree that a substantial part of the difference is due to differences in placement (initial job assignment) and/or promotion. Therefore, the regression analyses support the conclusion that female employees at AAFES have been significantly disadvantaged with respect to placement and promotion.

4. Other Evidence

As discussed above, the class claims of liability are supported by the non-parametric and regression analyses of both parties' experts. In addition, there is other evidence supporting the class liability case.

a. Subjective Evaluations

The trial evidence established that the single most important criteria for any promotion at AAFES is supervisory evaluation (by the Periodic Evaluation Reports)— and that such evaluations are essentially subjective in nature. This, of course, calls for close scrutiny of the system since it provides "a ready mechanism for discrimination." See, e.g., Fisher v. Procter & Gamble, 613 F.2d 527, 546 (5th Cir.1980); Hamilton v. General Motors Corp., 606 F.2d 576 (5th Cir.1979); Swint v. Pullman-Standard, 539 F.2d 77 (5th Cir.1976); and Rowe v. General Motors Corp., 457 F.2d 348 (5th Cir.1972).

Although AAFES correctly argues that women, as well as men, are supervisors at AAFES, it is also true that men are disproportionately more likely to become supervisors than females. Plaintiff's exhibit 8 shows that the percentage of male supervisors at AAFES is twice as large as the percentage of female supervisors. See also plaintiff's exhibit 9. Therefore, the evidence does establish that, on the whole, the net effect is to have men supervise and evaluate women under a subjective evaluation system. And, the fact that the PER forms have specific categories to guide an evaluating supervisor does not remove the subjective decision making which is necessarily involved. See Carroll v. Sears, Roebuck & Company, 708 F.2d 183 (5th Cir. 1983); Medina v. Reinhardt, 686 F.2d 997 (D.C.Cir.1982).

b. Anecdotal Witnesses

The plaintiff presented five female witnesses to testify as to their actual experiences at AAFES. Basically, they testified that: (i) they trained males for supervisory positions, but never received promotions to a supervisor's job themselves; (ii) they held higher positions on a temporary basis, but were never promoted to these jobs permanently; (iii) there were some direct statements of discrimination in promotions;^[41] and (iv) one of them (Ms. Merrit) was the victim of sexual harrassment by her supervisor.

AAFES also produced five female witnesses who testified that they knew of no policy of discrimination at AAFES and that AAFES had not intentionally discriminated against them or, to their knowledge, any other females. These witnesses included (i) women who had refused promotions by choice, (ii) women who had remained in lower level jobs for many years with little change of grade, and (iii) women who had received many promotions while employed by AAFES.

By its anecdotal witnesses, the plaintiff merely presented testimony which gave life to her statistical evidence; it is considered only for that purpose. The testimony of the AAFES anecdotal witnesses is understandable, and the Court accepts it to show that these females were not the subject of sex discrimination. However, since the plaintiff has established the class claims of liability through the statistical evidence, these "affirmations of good faith selection" by the AAFES witnesses does not discredit the plaintiff's evidence or explain the statistical results in a nondiscriminatory way. See Pegues v. Mississippi State Employment Serv., 699 F.2d 760, 766 (5th Cir. 1983); Wheeler v. City of Columbus, Miss., 686 F.2d 1144 (5th Cir.1982); and Williams v. New Orleans Steamship Ass'n.,, 673 F.2d 742 (5th Cir.1982).

5. Motion to Strike

The plaintiff's motion to strike certain tables attached to the AAFES post-trial briefs is moot. This opinion is based upon the evidence presented at trial, and does not consider the information contained in these tables.

6. Motion to Decertify

The only remaining motion is that filed by AAFES to decertify the class. It is, of course, based upon General Telephone Co. v. Falcon, 457 U.S. 147, 102 S. Ct. 2364, 72 L. Ed. 2d 740 (1982), which "significantly altered the pertinent law of this Circuit" concerning "across-the-board" certifications in employment discrimination cases.

In Falcon v. General Telephone Co., 611 F. Supp. 707 (N.D.Tex.1985), affirmed 815 F.2d 317 (5th Cir.1987) this Court analyzed these "significant changes" resulting from the Supreme Court's opinion in Falcon and the Fifth Circuit's opinion in Vuyanich v. Republic National Bank of Dallas, 723 F.2d 1195 (5th Cir.1984), cert. denied, 105 S. Ct. 567 (1984). That analysis^[42] identifies five Falcon-Vuyanich factors which trial courts must consider in resolving class certification issues in discrimination matters.

Although Shafer correctly argues that the certification order in this case was not an "across-the-board" certification,^[43] application of these five Falcon-Vuyanich factors demonstrates that Shafer is a proper representative of a class of females who claim discrimination in promotions and job assignment by AAFES. In particular:

(i) Class claims are "fairly encompassed." Shafer claims she was discriminated against because of sex in job assignment and in promotions to HPP and UA positions. The class claims are identical—alleged sex discrimination in job assignment and promotions to HPP and UA positions — and, therefore, are "fairly encompassed" by the individual claims.

(ii) Same interests and injury. Shafer does, under Falcon and Vuyanich, possess the same interests — and she suffered the same injuries — as the class because of the sex discrimination in promotions and job assignment.

(iii) Identical evidentiary approaches. Unlike Falcon and Vuyanich, the evidentiary approaches concerning the individual and class claims are identical. Shafer must prove her individual claim through by means of "disparate treatment" evidence, just as she has done in proving the class liability claims.

(iv) Standing. Shafer can assert injury as the result of her failure to obtain a promotion and her job assignment. Therefore, she does have standing to assert the class claims of discrimination in promotion and job assignment. Vuyanich, 723 F.2d at 1200; Blum v. Yaretsky, 457 U.S. 991 [102 S. Ct. 2777, 73 L. Ed. 2d 534] (1983).

(v) Footnote 15 exception. This factor is not relevant to the present case (except as discussed below in footnote 44).

The class attack upon the "mobility policy" must be considered separately. As discussed above, this requirement of mobility as a job requirement for all jobs UA 13 and above is discriminatory because of its adverse impact upon females. Although Shafer has never progressed high enough to apply for a UA 13 position, this policy still has the same impact upon her as it does other HPP and UA employees who may aspire to higher level jobs. Moreover, the claimed effect of this policy is to limit females to the lower-paying UA jobs. Therefore, Shafer may properly represent the class with respect to this claim.^[44]Carpenter v. Stephen F. Austin University, 706 F.2d 608, 617 (5th Cir.1983); Richardson v. Byrd, 709 F.2d 1016, 1020 (5th Cir. 1983), cert. denied, 464 U.S. 1009, 104 S. Ct. 527, 78 L. Ed. 2d 710 (1983).

In addition, even if Shafer could not serve as class representative in the attack upon the mobility policy, the decertification of this advanced class action would not be appropriate. Instead, this Court would permit the intervention of another class member to represent the class with respect to this claim.^[45] As stated in Carpenter v. Stephen F. Austin University:

"... However, if after the class has been certified and its claims heard and the representatives are found to be inadequate for some reason during the course of the class claims or during a bifurcated hearing with respect to individual claims, the appropriate step is appointment of new representatives from the existing class, not decertification." (706 F.2d at 617.)

7. Conclusion

The plaintiff, Neoma Shafer, has presented a very strong statistical case of employment discrimination, which was supported by anecdotal testimony from several class members. AAFES did not rebut the plaintiff's statistical case; indeed, the analyses of the expert for AAFES actually confirm those done by the plaintiff's expert. Moreover, there is no evidence to otherwise account for the gross statistical disparities which indicate employment discrimination.

Therefore, the plaintiff Shafer is entitled to judgment in favor of the class on Phase I liability that:

(i) the AAFES mobility requirement for all jobs UA 13 and above is discriminatory because of its disparate impact on females;

(ii) AAFES has treated females disparately, and less favorably than males, with respect to initial placement and promotion to UA jobs and to HPP jobs (grades 6-11).

The plaintiff's attorney is directed to submit to this Court an appropriate judgment—and, after conference with the defendant's attorney, a proposed schedule for a Phase II hearing on class relief.^[46]

NOTES

[1] The acronym for the A rmy and A ir F orce E xchange S ervice—AAFES—rhymes, albeit very poorly and Ogden-Nashedly, with the words "gave us" (or, more correctly, "gaf e us"). Therefore, it is somewhat understandable that a party distraught with the delay of this hard working, but overburdened Court, might well — at least during the Holiday Season, when spirits are high — send an attractive female to these chambers, carrying numerous balloons of festive colors, with this message in verse (sung, of course, to the tune of "Let It Snow, Let It Snow, Let It Snow": "Oh the case is Shafer v. AAFES We recall the trial you gave us Do you remember, yes or no? Let us know Let us know Let us know "Oh the age of this case is gallful Your procrastination awful Our impatience we must show Let us know Let us know Let us know "Would you finally give the word Now we're down on our knees, pretty please? And we heard from a little bird You'll even add on attorney's fees "Oh we've spoken as long as we dare to One final question have we for you Are we shafted, yes or no? Let us know Let us know Let us know"

However, this "motion" was not accompanied by a certificate of service, a brief, an order or a certificate of conference, as required by Local Rules 2.1(e) and 5.1(a)-(c). Accordingly, it will not be considered for purposes of this opinion. See, however, footnote 46. Please!

[2] Shafer did not seek an "across-the-board" certification attacking every employment practice at AAFES on allegations of sex discrimination. See General Telephone Co. v. Falcon, 457 U.S. 147, 102 S. Ct. 2364, 72 L. Ed. 2d 740 (1982); Vuyanich v. Republic National Bank of Dallas, 723 F.2d 1195 (5th Cir.1984), cert. denied, 469 U.S. 1073, 105 S. Ct. 567, 83 L. Ed. 2d 507 (1984).

[3] The opinion constitutes this Court's findings of fact and conclusions of law concerning the class liability issues. Rule 52(a), Fed.R.Civ.P.

[4] In closing arguments following the class liability trial, Shafer's counsel conceded that there had been no proof of sex discrimination by AAFES in "training opportunities."

[5] Each category is divided into "grades" for purposes of pay and promotion, and there are "steps" of differing pay levels within each grade. As the grade level increases, so does the pay and the responsibility of the particular job.

[6] This was conceded by the AAFES witnesses; it is also evident from the face of the PER forms (Def. Exhs. 2, 3, 4).

[7] AAFES has 39 career "cones" or fields for UA employees. There was some testimony that not every UA employee of the same grade would be eligible for promotion to a particular vacancy if they were in a different career cone. However, this Court credits other testimony — that employees are transferred and promoted between career cones — particularly since it is consistent with AAFES policies of "promotion from within" and of affording training courses to permit career changes.

[8] This applies to all supervisory UA employees and to some HPP and non-supervisory UA positions that are excluded from the vacancy posting procedure.

[9] The search first focuses on "surplus employees" (i.e., those who no longer have a job for some reason) and "lateral transfers" (i.e., employees returning from overseas).

[10] As discussed below, "mobility" is a required qualification for promotion into any UA job above grade 12.

[11] It appears that AAFES does "excuse" refusals to transfer, and does not terminate the employee, where there are extenuating circumstances (such as health problems of the employee or family member, financial reasons, etc.). Moreover, AAFES employees who testified indicated that they believed a refusal to transfer would not result in termination, but in denial of a promotion or possibly a demotion.

[12] Testimony at trial established that AAFES employees who are transferred are paid relocation allowances and given additional pay or quarters allowance.

[13] These are all of the UA-EMP jobs (grades 13-17) at AAFES.

[14] That is, a percentage significant at the ".05 level."

[15] The mobility requirement is not, therefore, one which an employee can "readily comply with." Accordingly, the decision in Garcia v. Gloor, 618 F.2d 264 (5th Cir.1980), upon which AAFES places great reliance, is clearly distinguishable from this case.

[16] On no less than three occasions, the Supreme Court has found job requirements to be discriminatory where there was no proof of "a meaningful study of their relationship to job performance ability." Griggs, 401 U.S. at 431, 91 S. Ct. at 853; Albermarle Paper, 422 U.S. 405, 95 S. Ct. 2362, 45 L. Ed. 2d 280; Dothard v. Rawlinson, 433 U.S. 321, 97 S. Ct. 2720, 53 L. Ed. 2d 786 (1977).

[17] The plaintiff argues that anyone "with knowledge of recent American history will realize that the disinclination to serve in Viet Nam was hardly limited to AAFES employees."

[18] This Court discounts the testimony of one witness (Smith) that it would be "too costly" to search out volunteers — since this is done for UA employees below grade 13 and since other evidence establishes that, under regulations, AAFES normally has four months to replace a UA-EMP employee in a particular job. Any "administrative inconvenience" cannot justify a practice with a disparate impact upon females. See Gunther v. Iowa State Men's Reformatory, 612 F.2d 1079 (8th Cir.1980).

[19] The parties are in dispute as to who has the burden of proof to show that "suitable alternatives exist which would have less impact" than the mobility policy. However, since this case has been fully tried on class liability issues, the ultimate factual issue is whether or not the mobility policy is discriminatory because of its disparate effect upon females. Williams v. Southwestern Bell Telephone Co., 718 F.2d, 715, 718 (5th Cir.1983). The plaintiff Shafer has met her burden of persuasion on this issue; Lewis v. NLRB, 756 F.2d 882 (5th Cir.1985).

[20] Although the term "non parametric" is subject to interpretation among experts, for purposes of this opinion it means the analyses of the experts other than multiple regression.

[21] In summary, UA females lost 44.3 promotions over a ten-year period, or 4.43 per year, according to exhibit 6.

[22] The plaintiff concedes that, "to the extent that Dr. Odell supported his availability conclusions with statistical analysis, they appear to be correct."

[23] Dr. Odell did concede that disaggregation "always reduces significance." It is for this reason that there must be a good reason to disaggregate data in statistical analyses. See Capaci v. Katz & Besthoff, Inc., 711 F.2d 647 (5th Cir.1983).

[24] Pages 11 and 12 of exhibit 7 are meaningless since, without explanation or justification, they disaggregate by mobility and college degree (even though testimony established that a college degree was not required for entry into any job grade). And, Dr. Odell characterized his final UA promotion analysis (page 17 of exhibit 7) as statistically meaningless.

[25] In summary, females lost 8.6 promotions over ten years in HPP grades 6-11, or less than 1 per year, according to exhibit 6.

[26] The plaintiff argues that "to hold otherwise would be to hold that an employer is free to discriminate against women so long as he does not discriminate ``too much.'"

[27] In its brief, AAFES concedes "at the outset ... that its own UA promotion analyses my possess some of the same defects as those of the Plaintiff. However, the fact that both sides may have performed statistical studies which produced questionable results does not change the fact that the burden of proving that Defendant has engaged in intentional discrimination on the basis of sex against its female employees is on the Plaintiff. That burden cannot be met with an improper statistical study no matter who has generated it."

[28] As was done in Pouncy v. Prudential Ins. Co. of America, 668 F.2d 795 (5th Cir.1982).

[29] Dr. Michael's assumption was supported by the evidence at trial. AAFES claimed that people were assigned to job grades on the basis of qualifications. Therefore, it was not unreasonable to assume that employees within a particular job grade were similarly qualified and should be treated similarly with respect to promotion.

[30] For example, AAFES claims that only two UA grades, UA 9 and 10, showed statistically significant results and that one-half of all UA promotions were observed in these two grades. And, it claims that promotions from only two HPP grades, HPP 6 and 7, dominate the HPP promotion analysis because they "accounted for 8 of the 8.6 shortfall for the ten-year period." Therefore, AAFS objects to the summary statistics for both UA and HPP promotions because they supposedly "totally altered the nature and meaning of the component numbers."

[31] For example, contrary to the AAFES claim (see footnote 30), plaintiff's exhibit 6 shows that in 8 of the 10 years (80%), and in 9 of the 11 grades, women in UA grades received fewer promotions than expected. Indeed, in five of those years, the results were statistically significant and were clearly not the result of activity in UA grades 9 and 10. In 1980 the result is significantly adverse to women even though women in grade 9 received more than the expected number of promotions. And, the same result occurs with respect to grade 10 in 1977.

[32] AAFES has made several other attacks upon the statistical analyses in its briefs. These do not merit discussion. However, this Court has considered — and rejected — each of these arguments by AAFES.

[33] For a discussion of the use of the regression analysis in employment discrimination cases, see Vuyanich v. Republic National Bank of Dallas, 505 F. Supp. 224 (N.D.Tex.1980), reversed on other grounds, 723 F.2d 1195 (5th Cir.1984), cert. denied, 469 U.S. 1073, 105 S. Ct. 567, 583 L. Ed. 2d 507 (1984).

[34] It has been noted that "Most economists think God is working great multiple regressions in the sky" — and also that "To every Ph.D., there is an equal and opposite Ph.D." The Official Rules (Delta Publ.Co.1979—compiled by Paul Dickson).

[35] Unlike the non parametric analyses, the regression analyses do not show a gross statistical disparity concerning promotions of females in HPP grades 1-11. Plaintiff's exhibit 2 shows statistically significant results in only three years (1973, 1974 and 1975), and in five years (1978-82) women in the grades fared better than males.

[36] The only difference between these exhibits is that in exhibit 3 all variables, including sex, were added to the equation stepwise, whereas in exhibit 1 sex was not added stepwise. This variation was prepared to satisfy the criticism of the AAFES expert, Dr. Odell, and did not alter the results significantly.

[37] A review of the betas on plaintiff's exhibits 1 and 3 indicates that the position of women relative to men has gotten worse since 1973, with the very worst years being 1978 and 1979. This supports Dr. Michael's opinion that the adverse results cannot be attributed to pre-act conduct.

[38] Although the plaintiff's expert performed this analysis, she correctly argues that it is not necessary for her to distinguish between placement and promotion discrimnation—because "present placement has been used by courts both as proof of initial placement discrimination, and as evidence of promotion discrimination." Vuyanich, 505 F.Supp. at 338.

[39] Indeed, Dr. Odell had agreed in his testimony that an R2 of .6 was acceptable.

[40] AAFES has made several other attacks upon the regression analyses in its brief. These do not merit discussion. However, this Court has considered — and rejected — each of these arguments by AAFES.

[41] Ms. Freeman testified that her supervisor, in selecting a male for promotion, stated that "a man would be better in that job."

[42] See parts 2 and 3 of this Court's opinion in Falcon, 611 F.Supp. at 711-18.

[43] This Court did refer to the Falcon decision (then pending on certiorari) and "across-the-board" certifications in its Findings of Fact and Conclusions of Law concerning the class certification issues (January 18, 1982). This was incorrect. Here, the class representative, Neoma Shafer, has individual complaints concerning sex discrimination in promotion and job assignment—and the class claims are identical (promotion, job assignment). Shafer was not certified as the representative of a class of applicants or of a class attacking every employment practice of AAFES (e.g., recruitment, hiring, termination). Therefore, this case does not involve an "across-the-board" certification as that term is used in Falcon and Vuyanich.

[44] In addition, the attack upon the mobility policy — a particular employment practice which, like a "biased testing procedure [used] to evaluate both applicants and incumbent employees," has the same effect upon all female employees at AAFES — would qualify as a "footnote 15 exception" under Falcon (457 U.S. 147, at footnote 15, 102 S. Ct. 2364, at footnote 15) since, for this purpose "the class action is limited to the tainted practice." Fleming v. Travenol Laboratories, Inc., 707 F.2d 829 (5th Cir.1983).

[45] Intervention by another class member would be proper since the intervenor's claims attacking the mobility policy would be "like or related to" Shafer's allegations of promotion discrimination in UA and higher HPP positions. This is not a situation like Vuyanich, in which it was not proper to permit other class members to intervene to attack all employment practices of Republic Bank since the named plaintiffs "had standing to assert only sex discrimination in hiring or race discrimination in termination." Vuyanich, 723 F.2d at 1198-1200.

[46] Explanatory note: For years, the Fifth Circuit approved "across-the-board" class certifications in employment discrimination cases, permitting a single plaintiff — whether current employee, ex-employee, or someone not hired — to maintain a class action attacking all of the employment practices of an employer. In Falcon, a landmark decision in a Dallas case, the Supreme Court reversed, telling the Fifth Circuit that they just didn't think too much of "across-the-board" certifications. Then in Vuyanich, a later case from Dallas, the Fifth Circuit said, well(!), they now cared even less than the Supreme Court about "across-the-board" certifications. But what of this case, Shafer v. AAFES, where the class — represented by attorneys Thorpe and Barber (who had three other class actions pending before this Court) — had been certified before Falcon and Vuyanich? Well...

THE FALCON

Once upon a backlog dreary, while I wrote on, weak and weary,

Opinions in class actions filed long ago, in days of yore,

While I pondered, nearly napping, suddenly there came a tapping,

As of someone gently rapping, rapping at my chambers door.

"``Tis some lawyer," I muttered, "TRO'ing at my chambers door —

Only this and nothing more."

Back to the class actions turning, all my thoughts within me burning,

When in stepped a frayed, bedraggled Falcon once class-certified when it did soar.

But two Supreme and Circuit trips had made he, so not a minute stopped or stayed he,

And, with mien of class action plaintiffs, perched above my chambers door —

Perched upon the Scales of Justice right above my chambers door —

Perched, and sat, and nothing more.

Then this troubled bird beguiling my weary fancy into smiling,

By the grave and stern decorum of the countenance it wore,

"Though thy class be shorn and shaven, thou," I said, "art sure no craven

Ghastly grim and ancient Falcon wandering from the Vuyanich shore —

Tell me whether the Fifth, the Circuit, will approve class actions, do outpour."

And, quoth the Falcon, "Nevermore."

Sad the Falcon, sitting lonely on those trembling scales, spoke only

That one word, as if his soul by that one word he did rent sore.

Nothing farther then he uttered; not a feather then he fluttered —

Till I scarcely more than muttered: "What of the classes I have certified before —

What of Hall, of Amason, of Autry, and what of the many more."

But said the Falcon, "Nevermore."

Startled at the stillness broken by this reply so aptly spoken,

"Doubtless," said I, "What it utters is its only stock and store."

But what of Thorpe, of Barber? Are there more across-the-boards to harbor?

What of the Quixotic battles dared, with Ma Bell and Mrs. Bairds?

Will they just handle fender-benders, soothe the pain with their bartenders?

Spoke the Falcon: "Evermore."

"Wretch," I cried, "the Fifth has sent thee—by Vuyanich it has bent thee,

But whether Supremely sent, or though the Fifth has tossed thee here ashore,

Desolate, yet all undaunted, in these tortured chambers haunted —

Tell me truly, tell me truly, I implore

Are there—are there no class actions?—tell me — tell me, I implore!"

Quoth the Falcon, "Nevermore."

"Except in Shafer v. AAFES, and you should probably let them know."