Full opinion text
CASSIBRY, District Judge: The plaintiff Andra Capaci filed this employment discrimination suit on October 8, 1974, alleging that the defendant Katz & Besthoff, Inc., [K&B] had discriminated against her by (1) refusing to promote her to a managerial position while it continually promoted less qualified male employees; (2) subjecting her to disparate terms and conditions of employment; (3) retaliating against her after she filed a charge with the Equal Employment Opportunity Commission [EEOC] through harassment and denial to her of a standard wage increase accorded to most other employees, and, finally, discharging her from her employment. The court has jurisdiction of this sex discrimination case under Title VII of the Civil Rights Act of 1964,42 U.S.C. § 2000e et seq. The EEOC which had determined as a result of the charges filed with it by the plaintiff in January, May and August 1973 that K&B had in fact discriminated against her, both by refusing to promote her and by retaliating against her for filing a charge, was permitted to intervene in the suit on July 19, 1977. Upon motion of the defendant the previously certified class was decertified on the condition that the EEOC would be permitted to prosecute the claims of the class. The EEOC took an active role in the trial of the case to prove that K&B had a policy and practice of discrimination against females by excluding them as a class from management positions. The trial of the case was limited to the liability issues. The plaintiff Andra Capaci, a female registered pharmacist, has been an employee of K&B in the New Orleans, Louisiana metropolitan area since 1963. K&B is a Louisiana Corporation domiciled in the City of New Orleans which operates a chain of drug stores in Louisiana, Mississippi and Alabama. The company, including its predecessor Katz & Besthoff, Ltd., has grown from one store opened in New Orleans in 1905 to eighty stores in 1979. The K&B organization was relatively small until 1971 when a period of rapid expansion began. In 1973, the year Capaci filed her charge, the work force of K&B was largely female, but its managerial force was predominately male. Until the time of her charge there existed basically two lines of progression to store-wide management positions at K&B. One line was from a manager trainee position to relief manager and progressively to assistant manager and manager. The other line prior to 1968 was from pharmacist to assistant manager to manager. The position of chief pharmacist was created in 1967 and thereafter this line was principally from pharmacist to chief pharmacist, and few pharmacists were promoted to store-wide management positions after that date. Capaci was never promoted to assistant manager or chief pharmacist during her 10-year period of employment with K&B. THE EEOC CASE The EEOC relies principally on its statistical evidence and the testimony of its expert Dr. Joseph L. Gastwirth to prove its allegation that K&B followed a policy and pattern of discrimination against females in its practice of excluding females from promotion to management positions after July 1965, the effective date of the Civil Rights Act of 1964. To buttress its statistical evidence it produced one present employee and four former employees who testified that their impression was that K&B did not treat males and females equally in appointments to management positions, and it presented evidence of advertising practices of K&B to show that it preferred males over females in its management positions. The EEOC also urges the court to draw an unfavorable inference from K&B’s failure to keep certain personnel records since the filing of the charge of discrimination in this case, allegedly in violation of the requirements of Title VII, § 709(c), and 29 C.F.R. 1602.14. STATISTICAL EVIDENCE The EEOC offered 79 statistical exhibits in three categories: (1) Referent Exhibits 1-26; (2) Manager Trainee Exhibits 1-17; and (3) Pharmacy Exhibits 1-36. The Referent Exhibits were for the most part presentations of data from the 1970 Census on the civilian labor force in Louisiana which Dr. Gastwirth considered relevant to the K&B managerial force. The Manager Trainee exhibits were directed to proof that K&B excluded females from the manager trainee position, and the Pharmacy Exhibits dealt with discrimination in the promotion of female pharmacists. Dr. Gastwirth made a variety of comparisons of the K&B managerial positions in his statistical studies. He compared the proportion of males and females in K&B’s managerial force at certain periods of time with the proportion of males and females in various segments of the civilian labor force in Louisiana in 1970. Theoretically in these comparisons he used the various labor force segments as labor force pools of potential applicants for managerial positions at K&B. For example, he tested the probability of observing 3 or fewer females out of 201 officials and managers, assuming they were selected like a random sample from various labor force manager groups in Louisiana, all of which had been weighted according to the number of stores in the various labor markets served by K&B, and the lower bound of that weighting was used for the testing. The manager groups used by Dr. Gastwirth for this test, designated by him as “Referents” and their respective fractions female are as follows: Referent Fraction Female All Managers .1607 Retail Trade Managers .1701 Retail General Merchandise Store Managers .2364 Department and Sales Managers (Retail Trade) .2207 He found the probability of 3 or fewer females occurring by chance was less than one in a billion as to each group which is statistically significant at the accepted .01 level of significance. Manager Trainees The only objective qualification for the position of manager trainee at K&B is a high school education. The highest annual salary for the period 1966-1973 was $7,488.00, or more when overtime was involved. The EEOC relies on three approaches in its statistical proof as to the manager trainee to prove that K&B discriminated against females in its hiring for the manager trainee position. First, it compared the proportion of males and females hired directly by K&B as manager trainees from July 1965 to January 1, 1973 with what it considered to be the relevant civilian labor force data for that position. Second, it compared the proportion of males and females hired as manager trainees during the years 1976 and 1977 with the proportion of males and females who applied for the position during those years. Third, it compared the distribution of males and females hired directly by K&B as manager trainees during the period July 1965 through December 1977. K&B hired directly, as distinguished from promotion from within the organization, 265 males and 0 females between July 1965 and January 1,1973. To test the probability of this proportion having occurred by chance, Dr. Gastwirth compared it with data for several groups of managers in the civilian labor force in Louisiana derived by weighting the labor market areas served by K&B according to the number of stores in them. Those groups of managers and the respective fractions female are as follows: Referent Fraction Female Managers Earning Less Than $7,000 (in 1969) .3387 Experienced Wholesale and Retail Managers Earning Less than $7,000 (in 1969) .2766 General Merchandise Retail Store Managers .244 Department and Sales Managers (Retail Trade) .2809 He was testing the probability that K&B would hire 265 males and 0 females into its manager trainee position between July 1965 and January 1,1973, if persons hired for the position were selected like a random sample from each of the above groups. His test results showed that the probability that 0 females out of 265 hired would occur by chance as to each group was less than one in a billion. Applicant flow data for the manager trainee position was available only for the years 1976 and 1977. The data included both outside applicants and those applying from within K&B, and showed the number of hires and the number of applicants for each year and for each sex, and the totals for both years were calculated. Males Females Percent Percent Applicants Hires Hired Applicants Hires Hired 1976 100 64 64.0 12 5 41.66 1977 296 84 29.38 82 10 12.20 Total 396 148 37.37 94 15 15.96 A chi-square test showed that the difference between the percentage of male applicants hired (37.37) compared to the percentage of female applicants hired (15.96) for the position of manager trainee during 1976 and 1977 is statistically significant. The probability of this difference occurring by chance is one in a thousand according to the test. During the period July 1965-December 1977, 633 males were appointed as manager trainee, and 20 females were so appointed. 97.95 percent of the males were hired directly and 30 percent of the females were hired directly. The remaining percent for each sex were promoted from within, of course. The remaining 70 percent female were all promoted from the position of cashier. A chi-square test, which tested whether the distribution of male and female outside hires was equal, showed that the difference in the percentage of females (30%) and males (97.95%) is statistically significant. This difference has a probability of less than one in a thousand of occurring by chance. K&B seeks first to discredit the statistical evidence of the EEOC. It urges the court to reject the EEOC statistical case because its expert, Dr. Joseph L. Gastwirth, was a theoretical statistician with little qualification for the problems of applied statistics in this case. It emphasizes Dr. Gastwirth’s complicated responses to some of the questioning, the great difficulty he had in explaining some of his statistical exhibits, his admitted failure to familiarize himself with the K&B operation, and his small expertise as a labor economist. These factors make the burden of the court more onerous, but they are not cause of themselves to reject his testimony and his exhibits as having no weight at all. K&B also urges that much of the evidence of the EEOC has no relevance to the issue of discrimination in this case. The statistical proof of the EEOC as to the manager trainee position deals largely with manager trainee hires from outside the K&B work force. K&B contends that these statistics are largely irrelevant because they ignore the established practice of K&B to look first to its existing work force for new manager trainee candidates before going outside. The EEOC statistics would be irrelevant if the EEOC had combined those hired from outside and those promoted from within and compared the total to groups in the Louisiana labor force on the theory that those groups were the labor source pools. In the recent case Johnson v. Uncle Ben’s, Inc., 628 F.2d 419 (1980), the Fifth Circuit Court of Appeals held that the two issues of access to jobs in an employer’s work force — promotion and lateral hiring— should be kept separate in scrutinizing statistical analyses. The real issue here is whether the statistics are deficient because no statistical studies were made comparing those promoted from within to manager trainee with K&B’s internal work force. The Fifth Circuit held in James v. Stockham Valves & Fittings Co., 559 F.2d 310 (1972), that for jobs filled by promotion, the relevant comparison is the company’s internal work force. According to the Fifth Circuit in Uncle Ben’s, supra, this statistical lack is not necessarily fatal: The difficulty is that most cases fall between these extremes of exclusive promotion and of exclusive non-promotion [hires from outside]. As a general matter, however, cases are dealt with in terms of the extreme to which they most closely accord. * * * 628 F.2d at 425. The extreme to which this case most closely accords is that the manager trainee position was filled by outside hires. When questioned about his failure to make studies of those promoted from within, Dr. Gastwirth testified that approximately 95% of those appointed to the manager trainee position were hired from outside the K&B work force. The groups .selected by Dr. Gastwirth based on Labor Market data from the 1970 Census are seriously questioned by K&B as not being appropriate for comparison to the manager trainee position or any management position at K&B. Dr. Gastwirth was equivocal at some points in his testimony as to whether his groups were source pools for actual applicants or comparison groups for the manager trainee position, but the explanatory language in his exhibits and his testimony as a whole indicate that in his studies he regarded the groups as source pools. K&B argues that these groups are not sufficiently representative of its source of actual applicants to have any relevance to its hiring practices. It accuses the EEOC of using any available data without regard to its relevance to the K&B work force. Dr. Gastwirth admitted that his groups include part time employees and the self employed, some of whom are not potential applicants for the manager trainee position, and that no data was available to him to make these refinements. The potential applicant pool will be further reduced in proportion female, according to K&B, by the lack of interest of females in applying for a job having a manager trainee schedule and work requirements. Dr. Gastwirth had not informed himself as to the nature of the job from these particular aspects of it. The job requires the manager trainee to unload supply trucks, to put up stock, straighten up the store, and the work schedule includes night work, week end and holiday work. Dr. Charles J. Cranny, testifying for K&B as a labor economist with some expertise in industrial organizational psychology (the psychology of people at work), expressed the opinion that the requirements of the job, including the schedule, would substantially retard females from applying as compared to males. He offered no studies to support his opinion, and admitted that no data was available, but defended his opinion as being based on a reasonable explanation for the reason that the K&B manager trainee force had a smaller proportion female than the labor market groups. To lend practical support to Dr. Cranny’s opinion K&B offered two female witnesses who resigned from the manager trainee position because the schedule was too rigorous. There was also testimony from a female employee who had refused to enter the management program when approached by management, and from another who had made known to management that she was not interested because of the heavy manual labor involved and the week end work. There was additional testimony from males in management naming female employees who were sought out for the manager trainee program and who refused to enter it. Dr. Gastwirth specifically questioned Dr. Cranny’s opinion as to the effect of night work in reducing the number of potential female applicants. He admitted generally, however, that the data from the 1970 Census applicable to the managerial groups used as a basis for his studies was aggregated, that is, it needed refinement that he could not make, so that none of these groups would be the most appropriate to compare with any management job at K&B. K&B accuses Dr. Gastwirth of using an inappropriate statistical approach in his testing of the data for manager trainee applicants and hires for the years 1976-77. Whereas Dr. Gastwirth combined all applicants and all hires throughout the K&B chain for those two years, K&B contends that the data should have been broken down and tested as to each locality where applications were received. The justification for this is that the applicants for a position open at one location, for example, Alexandria, Louisiana, cannot logically be regarded as applicants for a position at another location, for example, Lake Charles, Louisiana, so that the data should not be combined. Dr. Cranny tested the data as to each location and found a result of statistical significance at only one — New Orleans. For the sake of argument, Dr. Cranny used five of Dr. Gastwirth’s groups taken from the 1970 Census data (and added two groups from EEO-1 summary reports for the State of Louisiana) and compared them with the manager trainee hire and promotion data broken down by year for the period July 1965 through 1978 to show that there was no pattern of disparate treatment of females at K&B. The Groups: 1. Civilian Labor Force (Percent Female 36.76) 2. All Managers (Percent Female 16.24) 3. All Retail Trade Managers (Percent Female 18.16) 4. All General Merchandise Retail Store Managers (Percent Female 24.40) 5. Department and Sales Managers— Retail Trade (Percent Female 28.09) 6. EEO-1 Report Summary by State: Officials and Managers, New Orleans SMSA (Percent Female 13.2) 7. EEO-1 Report Summary by State: Officials and Managers, Louisiana (Percent Female 11.4) (He considered the EEO-1 groups less inappropriate than Dr. Gastwirth’s groups for comparison purposes.) The Hire and Promotion Data: 7-12/65 1966 • 1967 1968 1969 1970 1971 1972 1973 M 12 20 28 31 32 42 47 54 71 F0 00000000 1974 1975 1976 1977 1978 M 93 55 67 85 1 F 1 3 5 8 11 Dr. Cranny reported his test results at .01 and .05 probability levels and N.S. (not significant). He found no pattern of disparate treatment over these years at the .01 level even after correcting several errors in his calculations which were called to his attention by the EEOC. Dr. Gastwirth objected to Dr. Cranny’s approach of testing for each year as being inefficient because it resulted in reducing the sample size, thus making more difficult a rejection of the null hypothesis that males and females were appointed in equal proportions. He was further of the opinion that it was not proper to separate the data by year for the period that the percentages were the same (1965-1972, 100% males and 0% females were appointed) because similar data should be pooled for testing. He did not demonstrate, however, how the lack of pattern would have changed if Dr. Cranny had pooled the data for those years and separated it only for the period 1973-1978. Dr. Gastwirth also suggested that a pattern may be hidden because the probabilities were reported at the .01 and .05 levels rather than the exact probabilities. To the serious and substantial objections K&B has made to the labor force groups which the EEOC used for comparison with the manager trainee group, and to the techniques the EEOC used in presenting its statistical evidence, the EEOC has made no argument to sustain its groups as relevant either as source groups or as comparison groups, or to sustain its questioned techniques. Statistics can serve as a useful tool in pattern and practice suits in the determination of whether defendants have engaged in employment discrimination. International Brotherhood of Teamsters v. United States, 431 U.S. 324, 97 S.Ct. 1843, 52 L.Ed.2d 396 (1977). In this case, for example, if appropriate data has been collected and properly interpreted, statistics can determine whether disparities in the form of underrepresentation of females in the K&B managerial work force can be attributed solely to chance factors. Where gross statistical disparities can be shown, statistics alone may in a proper case constitute prima facie proof of a pattern or practice of discrimination. Hazelwood School District v. United States, supra note 9; Equal Employment Opportunity Commission v. Datapoint Corporation, 570 F.2d 1264 (5th Cir. 1978); Davis v. Califano, 613 F.2d 957 (D.C.Cir.1979); Markey v. Tenneco Oil Company, 635 F.2d 497 (5th Cir. 1981); Little v. Master-Bilt Products, Inc., 506 F.Supp. 319 (N.D.Miss.W.D.1980). To have such force the statistics must be relevant, material and meaningful. Equal Employment Opportunity Commission v. Datapoint Corporation, supra; Johnson v. Uncle Ben’s, supra. To be relevant as source groups for comparison with K&B’s manager trainee force, the labor force groups used by the EEOC must be the proper population of potential applicants for the manager trainee job. All of the basic referent groups are admittedly aggregated, that is, they contain persons who are not truly potential applicants. Furthermore, all of the groups are aggregated because of the female self-selection factor. Dr. Cranny’s opinion on this factor as affecting the potential applicant pool has a reasonable basis that would remove it from the realm of mere surmise or conjecture. The data was aggregated, therefore, and there is no certainty as to how much it was aggregated. Dr. Gastwirth admitted that figures were not available for him to refine the data. As mentioned earlier, Dr. Gastwirth in his testimony at times appeared to be regarding the manager groups taken from census data as comparison groups, but he did not show how they could properly be considered as comparison groups, and the EEOC in its arguments does not indicate that it proposes any of them as comparison groups. All of the groups included data on businesses not comparable to the K&B operation, such as boutiques and dress shops. Dr. Cranny indicated that a comparable management level job in other chain pharmacies, or even in grocery chains, would provide an appropriate comparison group, but no data is available for either. The EEOC does not defend its statistical approach in testing the data for manager trainee applicants and hires for the years 1976-77 against the charge of K&B that the data combining all applicants and hires for the two years should have been broken down by locality. There was no testimony from Dr. Gastwirth explaining how his approach was preferable to Dr. Cranny’s. The value of testing the distribution of the percentages of male and female outside hires for the manager trainee position, as compared to the percentages of males and females promoted from within, for showing a pattern and practice of discrimination was never explained in testimony or otherwise by the EEOC. Statistics should be meaningful, and not fashioned to obtain a desired conclusion. Equal Employment Opportunity Commission v. Datapoint Corporation, supra. Pharmacists Statistically, the EEOC analyzed its allegation of discriminatory promotion practices from the standpoint of the pharmacist position. The purpose of the analysis was to test the validity of the hypothesis that male pharmacists and female pharmacists had an equal chance of being promoted to a management position at K&B. The promotion practices were surveyed from various viewpoints and for several periods. First, Dr. Gastwirth examined how female pharmacists fared in promotion to management positions prior to the effective date of the Civil Rights Act, July 1, 1965. Second, he examined the relative length of service between male and female pharmacists prior to promotion. Third, he compared the proportion of male and female pharmacists promoted during certain periods with the proportion of male and female pharmacists hired during the period. Fourth, he examined as of January 1973 (1) the distribution of jobs according to the level of desirability of all employees who at some stage of their employment had been pharmacists, and (2) the proportion of male and female pharmacists at K&B in managerial and staff positions in comparison to the proportion of males and females in those positions in community pharmacies, hospitals and nursing homes in Louisiana during 1973. Dr. Gastwirth tested the probability of equal promotion of male and female pharmacists in the K&B work force as of July 1965 who were appointed before Capaci was appointed pharmacist — those appointed between June 1949 and January 1963 — and who were promoted prior to July 1, 1965, the effective date of the Civil Rights Act. His data showed that out of 29 males and 6 females, 65.5 percent (19) of the males were promoted and 0 females. The Fisher exact test gave this comparative rate of promotion a probability of .0049, or less than one-half of one percent, — statistically significant. Dr. Gastwirth extended the appointment period — June 1949 to July 1, 1965 — and the data changed. Out of 39 males and 13 females, 51.28 percent (20) of the males were promoted and 0 females. The Fisher exact test showed this difference to be statistically significant with a probability of less than one in a thousand. Using the Wilcoxin-Mann-Whitney test Dr. Gastwirth determined that there was a statistically significant difference for males and females in the time from appointment as pharmacist to promotion (either to chief pharmacist or a store-wide management position) of the 41 pharmacists promoted during the period July 1,1965 through January 1973. The two females who were promoted had service times of 229 and 97 months from appointment to promotion. These were the largest service time (rank 41) and the third largest (rank 39). The probability of this extreme rank order for the females occurring by chance was less than one-half of one percent — statistically significant. When the test was limited to data on time from hire to promotion to chief pharmacist (females ranked 25 and 24 out of 26 rankings) the results showed that the probability of observing this difference between males and females in time to promotion was .0123, and statistically significant in the opinion of Dr. Gastwirth. The proportion of male and female pharmacists who had been hired in the period from July 1, 1965 to January 1, 1973 and who had received a promotion during that period, was compared. Twenty of the 196 pharmacists hired during that period were female, but of the 28 promoted during that period 0 were female. Using the Fisher exact test the probability of observing this difference is .049. A statistical comparison was made of the proportion of male and female pharmacists hired between June 1963, the date plaintiff was hired, and January 1,1973, just prior to the charge, who were promoted to chief pharmacist or store-wide management position during that period. Out of 220 hired (191 males, 29 females) 34 males (17.8%) and 0 females were promoted. The probability of observing so few females promoted is less than .0082, statistically significant at the .01 level of significance. How female pharmacists fared in promotion to assistant manager was tested. From the inception of the Civil Rights Act, July 1, 1965, to December 31, 1977, 273 males were hired, 24 of whom were promoted to assistant manager. Seventy-nine females were hired and none was promoted. This difference tested at a probability of .0022, which is statistically significant. The 145 positions held as of January 1973 by employees who were pharmacists at some stage of their employment with K&B were examined on the basis of their distribution among males and females according to their wage level of desirability (top level —supervisor, manager, buyer; middle level — assistant manager, chief pharmacist; lower level — pharmacist). Twenty-six males and 0 females held top level positions, twenty-four males and one female, middle level, and eighty-three males and 11 females, lower level. The Mann-WhitneyWilcoxin test rejected the hypothesis that males and females had the same distribution of positions at the .05 level of statistical significance, but not at the .01 level. A comparison was made of the managerial and staff positions held by male and female employees as pharmacists at K&B in January 1973 with those positions held by active resident pharmacists employed in community pharmacies, hospitals and nursing homes in Louisiana during 1973. At K&B 1 of 51 managers was female, and 11 of 94 staff pharmacists were female. In the community pharmacies, hospitals and nursing homes 30 of 430 managers were female and 96 of 519 staff employees were female. Dr. Gastwirth combined the probability of observing fewer females in manager and staff positions at K&B than in the other places of employment, and the overall probability was .007, statistically significant at the .01 level. K&B charges that only by ignoring relevant facts and juggling dates was the EEOC able to present statistical evidence against it in its promotion of pharmacists. Under particular attack is the testing of data for those pharmacists hired between July 1, 1965 and January 1, 1973 and promoted during the same period. The data was to the effect that 0 females out of 20 were promoted, but 28 out of 176 males were promoted. By this juggling two females who were promoted to the chief pharmacist position in 1968 were eliminated because both were hired prior to 1965. These two were among the first six promoted after the position of chief pharmacist was created. Females were actually over represented in this instance. Furthermore, in its assistant manager test the EEOC tested data through 1977, but none of its tests were structured so that three females promoted to chief pharmacist in 1973 were included in the data. K&B objects to the statistical evidence relating to time before the Civil Rights Act because K&B is not liable under the act for preact conduct. Preact conduct can be useful, however, in an examination of pattern and practice if the postact conduct is shown to be a continuation of an employer’s preact discriminatory practices. Those studies comparing the K&B pharmacists to those in community pharmacies, hospitals and nursing homes are questioned as not being relevant because hospitals and nursing homes inflate the percentage female since those businesses traditionally attract a greater proportion female and therefore are not similar enough to community chain pharmacies to be comparable for statistical analysis. When the data on hospitals and nursing homes were eliminated, the result was not statistically significant. Additionally, Dr. Cranny did not agree with Dr. Gastwirth’s statistical method of combining the probability of the managerial group and the staff pharmacist group and multiplying the two to obtain the overall probability, which was significant at the .01 level. Dr. Cranny admitted that it would be statistically correct to multiply the results for the two groups if they were independent of each other, and he was not able to explain clearly that they were not independent. Dr. Gastwirth’s justification for considering them independent was that one group was not taken from the other, that is, managers and staff pharmacists were not taken from the same pool, and Dr. Cranny was not able to demonstrate a basis for considering them dependent groups. K&B argues that Dr. Gastwirth shifted his choice of level of significance to fit the EEOC position. Dr. Gastwirth agreed that to avoid the possibility of error in assuming that a nonchance reason accounted for the difference in the data on promotion of males and females, the .01 level of significanee was proper. K&B points out that he departed from this opinion in several of his tests and found statistical significance even when the test results showed a probability above the .01 level either by choosing the .05 level or by using such terms as “marginally significant”. This occurred in his test of job level distribution as of January 1973 of all employees who had been pharmacists at some time in their employment at K&B. This test was also subject to the single date testing objection, and the further objection that the data included preact employment decisions. K&B offered its own statistical evidence for the purpose of showing that statistical evidence does not support an inference of discrimination against females in its promotion of pharmacists. Dr. Cranny performed a median test to test whether the median time for promotion to chief pharmacist for males and females differs in a statistically significant way. The results showed there was no significant difference in the distribution of time to promotion in months during the periods 1965-73, 1973-78, and the combined period 1965-1978 (chi-square tests for the two periods and combined were, respectively, 1.05,1.15 and .05). By utilizing this median test Dr. Cranny explained that he was looking at the typical time to promotion, whereas Dr. Gastwirth, by using the Wilcoxin-Mann-Whitney test to test the difference in the length of time to promotion, was testing the similarity of the shapes of the two distributions—male and female—, and statistical significance will be obtained if the two distributions do not match, regardless of the manner one distribution does not match the other. Dr. Cranny questioned the validity of Dr. Gastwirth’s reporting of mean or average length of time to promotion for males and females (163 months for females, 29.10 for males), while using the Wilcoxin test, as an inconsistency, because the Wilcoxin test makes no assumption about the symmetry of the distribution, but the average assumes a symmetrical distribution. Dr. Gastwirth denied that reporting the average time to promotion for male and female along with the results of the Wilcoxin test is misleading, but downgraded the average in importance as being “just a summary statistic.” Dr. Gastwirth agreed that the median test performed by Dr. Cranny would not result in statistical significance, but he considered it inappropriate as not being a statistical test powerful enough to test the hypothesis of equal distribution of time to promotion between males and females. K&B introduced the results of several statistical tests performed by it to show that raw data, produced by Dr. Gastwirth at his pre-trial deposition, when tested would produce nonsignificant results. K&B tested the comparison of males and females promoted to chief pharmacist from July 1965 through December 1973 with the total number of each sex employed by it as pharmacists during that period and the results were not significant. A total of 215 males and 33 females were in the K&B pharmacy work force during this period, and of these, 32 males and 5 females were promoted to chief pharmacist. Dr. Cranny made a chi-square test of the proportions male and female promoted and the results were not significant — .05 level. The test differs from the EEOC evidence testing promotions to chief pharmacist in that it has no cut-off hire date which would eliminate the females promoted during the period, and the test period extends to the end of 1973 rather than January 1973, the date of the charge. This extension of the test period allowed the three females promoted after April 1973 to be included in the data count. Raw data provided by Dr. Gastwirth before trial, but not tested, also prompted K&B to test a comparison of male and female promotions to chief pharmacist of those who were hired after the plaintiff Capaci (i. e. after June 1963) and who were pharmacists before January 1973. Twenty of 179 males hired were promoted compared to 0 females out of 29 hired. Dr. Cranny utilized the chi-square test and the result, as corrected at trial, was 2.41, not significant. The EEOC challenged the chi-square test as the proper one because it is not precise enough in this instance where the testing would show that the number of females expected to be promoted out of 29 is so few — 2.79, or about three. The challenge was based on the school of thought that the chi-square test should not be used when the expected value is less than five. Dr. Gastwirth preferred the Fisher Exact test here, and using a quick approximation of it, he concluded that females had a lower promotion rate than males, but he admitted that Dr. Cranny’s chi-square result was not a “terribly bad” approximation. Dr. Cranny defended his use of the chi-square test, and was of the opinion that the chi-square test should be rejected for the more precise Fisher Exact test only when the third or fourth decimal place is needed, and in testing for two decimal places, .01 or .05, the chi-square test is an adequate approximation. Dr. Cranny further explained that the actual number of females promoted will inevitably be fewer than the expected number shown in the testing when the male group from which the selection is made is much larger than the female group and the selection ratio is small. This is so because the larger group the more qualified persons will be found in it, and therefore the chance of the selection being made from that group, rather than the smaller one, increases. For separate periods — July 1, 1965 through January 1, 1973 and July 1, 1965 through December 31, 1973 — Dr. Cranny tested the comparison between proportions of male and female pharmacists promoted to a store-wide management position. In the first period, according to Dr. Gastwirth’s data, of 215 males at K&B during the period, 18 were promoted, and of 33 females, 0 were promoted. In the second period, of 242 males, 18 were promoted, and of 39 females, 0 was promoted. The results of the chi-square tests were not significant at levels 1.87 and 1.98 respectively. Since the chief pharmacists and store-wide managers who are pharmacists are drawn from the same pool — the pharmacy work force — Dr. Gastwirth disagreed with the technique of analyzing the promotion to these two categories separately as Dr. Cranny did. He disagreed because once a pharmacist is chosen from the pool for promotion to chief pharmacist, that same pool can no longer serve as the pool for promotion to assistant manager. This is so because the pharmacist promoted to chief pharmacist is no longer available for promotion to a store-wide management position — assistant manager. That promoted pharmacist is removed from the pool. He disapproved therefore of comparing separately one promotion category — store-wide management position — with the total pharmacy pool. The proper analysis in his opinion, where two categories are promoted from the same pool, is a two-step analysis. The first step is to look at those pharmacists who are not promoted and those promoted, and the second step is to look at those who were promoted to ascertain the category to which they were assigned. By this analysis the total number of pharmacists is classified into three groups — those not promoted, those promoted to chief pharmacist, and those promoted to assistant manager. The first question to be answered is, were the promotion rates of males and females equal, and the second is, of those promoted, were the assignment rates equal? The purpose of this analysis is to examine the equality of distribution of assignment of males and females to each of the two categories, that is, to determine if the distributions are the same — equal percentage male and female promoted to chief pharmacist, equal percentage male and female promoted to store-wide management. Dr. Gastwirth demonstrated his analysis by the example of a total of 100, 70 not promoted, 15 promoted to chief pharmacist, 15 promoted to assistant manager. If 15 of the males were promoted to assistant manager, 15 of the females would be expected to be promoted to assistant manager. If 15 males were promoted to chief pharmacist, the expectation female would be 15. He represented that the mathematical equivalent of this analysis was made by him when he combined those pharmacists promoted to chief pharmacist and store-wide management position between June 1963 and January 1973, who were hired during the same period, to compare statistically the proportion of males and females promoted. His explanation of this test on his direct examination and the explanation of the test in the exhibit introduced to demonstrate the data and test results do not reveal any such technique was followed. According to him, the second step of the analysis was eliminated as not being essential when his data showed 0 females were promoted, thus he had no distribution of males and females to compare in the chief pharmacist and store-wide management categories. He concerned himself only with the first question, therefore, — were the promotion rates of male and female equal? His criticism of Dr. Cranny’s technique of testing separately the two categories did not include his own application of his preferred technique to data used by Dr. Cranny to show the statistical result. Dr. Gastwirth further objected to separating the promotion categories, chief pharmacist and store-wide management for testing because separating them reduced the sample size and made more difficult any meaningful statistical testing of the issue of whether the female promotion rate was equal to the male. To show generally that K&B did not discriminate against females in its pharmacy work force, Dr. Cranny compared the proportion female from July 1, 1965 through January 31, 1973 (of 248 pharmacists, .133% were female) with three labor force groups: 1. Proportion female (.11) of active resident pharmacists in Louisiana; 2. Proportion female (.09) of pharmacists practicing in community chain pharmacies in Louisiana. 3. Proportion female (.1441), according to the 1970 census of employed pharmacists in the civilian labor force in markets in Louisiana served by the defendant. K&B has a higher proportion female than the first two groups, and the result of a statistical test comparing the proportion female at K&B with the third group is not significant. The EEOC has not questioned these comparisons. The EEOC statistical tests of the data relating to K&B’s promotion of pharmacists before the passage of the Civil Rights Act are reliable tests to raise an inference of pattern and practice discrimination against female pharmacists. Less reliable are the tests relating to pharmacists’ promotions after passage, of the Act. Particularly suspect is the EEOC’s juggling of dates to eliminate from the data in some of its tests female pharmacists who were promoted after the effective date of the Act and before the date of the filing of the Capaci charge, so that 0 females was the consistent data tested. All actual promotions of pharmacists after the Civil Rights Act, particularly those for the period 1965-73, were relevant to the issue of discrimination in this case. Dr. Gastwirth was not able to explain to the court the relevancy of the hire date to this issue in several of his studies in which the hire date had the effect of eliminating data on female pharmacists who had been promoted after the Act. The inclusion of pharmacists in hospitals and nursing homes with the data on community pharmacies makes this group less appropriate as a comparison group to the K&B operation than a group limited to chain pharmacies. Furthermore, comparisons of these groups have a limited value because the tests involve only the year 1973. NON STATISTICAL EVIDENCE To enhance its statistical proof the EEOC relies on the testimony of several females, some of whom applied for management positions at K&B and were allegedly rejected because of their gender, and some of whom were allegedly discouraged from seeking management positions. Marsha Lee Abrams applied for a manager trainee job in Baton Rouge in September 1977. She was never interviewed. Her education — she had worked towards her doctorate at Columbia University — and her extensive work experience in the publishing business clearly supports K&B’s contention that she was in fact over qualified for the manager trainee job for which a high school education was the only qualification and which involved a substantial amount of manual labor. Two positions were open in Baton Rouge at the time Abrams applied. One was filled by a male, and one was filled by a female salesperson, Madeline Hopwood, who was promoted to the manager trainee position. Furthermore, according to the applicant-hire data which is available in this case and which includes the period of Abrams application, of 11 female applicants in Baton Rouge, four were hired. Patricia McAuley testified that she applied for a manager trainee job in June or July 1974 and was told by William Serda, K&B personnel director, that she was not qualified for the management program and that it was not then open for women. She had been hired as a drug clerk in 1971 and, after making two more applications to become a manager trainee, she was appointed to that position in January 1976. The weight of her testimony in recalling that Serda remarked that the management program was not open to women is lessened by her further recall that he told her to try again, and by the fact that a female, Joyce Innerarity was at that time in the program, having been promoted from a salesperson to manager trainee in February 1974. Viewing McAuley’s testimony from the standpoint of her ultimate success in being accepted into the management program, it can be interpreted as evidence that, if a female wanted to be in management, and could convince the personnel director that she was qualified, being female would not exclude her. Carol Baughman, the head cosmetician in the Bogalusa, Louisiana store is an employee allegedly discouraged from entering management. She made an application at the insistence of her male assistant manager, Mr. Sheler, for a relief manager opening in that store in 1976. She made the application on the last day K&B was receiving applications, and she heard the next day that a male had got the job. A relief manager position became available again in August 1977. This time her manager Mr. Hopper asked her if she intended to apply again, and in response she asked Hopper if he thought she would make a good relief manager. He told her he thought she would make a good one and she made the application. On the last night that applications were being taken her assistant manager Mr. Sheler told her he thought she had the job if she wanted it, and asked her how much did she want the job. After some conversation she told Sheler she was withdrawing her application. According to her, she withdrew her application because she knew she was about to have surgery and she did not want to inconvenience the other managers because of it, although she did not disclose to Sheler any information about her impending surgery. The EEOC was not making its point for calling this witness and the Court took over the interrogation to ask the witness to search her mind for anything discouraging said to her in the conversation. At this prodding she recalled that Mr. Sheler had told her that, although he was the one who had talked her into making her first application, she was a lady and this was not the kind of job he thought she could handle. The crucial point of whether this remark was made before she told him to withdraw her application or afterwards was not established by her- testimony. She testified that Mr. Sheler never suggested that she should withdraw her application, that its withdrawal was wholly voluntary, and that she believed she would have got the job if she had not withdrawn it. However, if the remark of Sheler was made before she told him to withdraw her application, it could be construed only as encouraging her to withdraw it. If his remark was made after the withdrawal, it could be construed as an effort to make her comfortable with her decision and to indicate that perhaps he had been wrong to encourage her to apply. Linda McBride, a former employee of K&B from 1973 to 1976, serving first as cashier and then as drug clerk, and a former roommate of Patricia McAuley, felt that she had been discouraged from applying for a management position because her managers would not give her information as to what the qualifications were for manager and because her roommate Patricia McAuley had so much trouble entering management. She testified that she wanted to know if a high school diploma, or a college degree or managerial experience was required. When she inquired as to these possible qualifications of one relief manager, Robert Williams, he did not tell her the qualifications, but told her instead that she would not be able to unload stock trucks, catch shoplifters and do other heavy work. She recalled talking to her manager Tom Gereighty about the requirements, but she could not recall what he said to her. She admitted that she never asked her roommate what the qualifications were, and that she did not apply for a manager trainee job because she did not want to go through the “hassle” of proving herself to the company. The testimony of Grace Taylor Miller, a pharmacist who commenced her employment with K&B in 1931 as a cashier, is relied upon by the EEOC to show that she performed the duties of managers but was never officially promoted. The EEOC does not state the basis for its contention as to lack of official promotion. Miller’s employment record which is not disputed, shows that she was promoted from pharmacist to assistant manager on November 11, 1955 and that she was promoted from pharmacist to chief pharmacist on July 1, 1968. Miller did not clarify what she meant when she testified merely that she never received the official title. The further contention is made by the EEOC that Miller was never treated as a manager because, whenever she attempted to discipline or correct the behavior of a male subordinate, she was not supported by her male superiors. Any of her testimony that supports this contention is not identified or discussed. She testified at length about an incident in 1978 near the end of her employment with K&B when she considered that a subordinate pharmacist was infringing upon her authority when he took over the ordering for the pharmacy department with an explanation that the store supervisor had so instructed him. By her own testimony she did not attempt to discipline him, but instead waited to discuss the matter with the store supervisor. When her supervisor told her that the subordinate was in error as to the instruction, reminded her that she was supposed to be the chief pharmacist and suggested that under the circumstances it would be better for her to get out, she construed the suggestion as termination and she did not return to work. All efforts by the store operations manager, Walter Feltman, to convince her that she had misconstrued the conversation and to get her to stay on were unsuccessful. On an earlier occasion, when she considered that one of her subordinates was out of line, she left the store and went home. K&B transferred the subordinate to another store. Her testimony as a whole does not show that she was not treated as a manager by K&B. The charge that she did not receive the bonuses that other K&B managers were paid is not substantiated by any evidence other than her testimony that she did not receive bonuses. Chief pharmacists have never been included in the bonus program, and there is no evidence to show the policy as to bonuses when she was assistant manager in the 1950’s. K&B undertook to show affirmatively by nonstatistical evidence that it has never either before or after the Civil Rights Act, discriminated against females on the basis of their gender. Sidney J. Besthoff, III, president of K&B, joined the organization in 1949 when he finished college. His grandfather was one of the founders of the original predecessor organization and his father was an owner of this closely held company. He worked at a number of positions in the company including assistant manager in 1949 or 1950, personnel manager in the early 1950’s, and later on store operations manager before he became president in 1965. Besthoff regards himself as having been a leader in efforts to end racial discrimination. In the early 1960’s K&B was the first local chain to integrate voluntarily its food service facilities. As a result of this action Besthoff, as a representative of the chain drug industry, was invited to attend a conference at the White House while John F. Kennedy was President on the problem of racial discrimination in transportation and public accommodations. He supported the civil rights legislation which was later adopted. Although the “early thrust” of K&B under Title VII of the Civil Rights Act was a program directed to end racial discrimination in employment, Besthoff testified that the company had no real need for a recruitment program for females in management because females had been in management for many years before the Civil Rights Act, and the need for recruitment of females in management arose only after the expansion program for the company began in the early 1970’s. From 1947 through May 1965 nine females were promoted to assistant manager. Of these Genevieve Ronquette was further promoted to personnel director in 1950, then made personnel manager in 1959 and retired in 1969; Lillian Hennessey was further promoted to manager in 1954 and retired in 1973; Blanche Callihan was further promoted to manager, but had an accident which prevented her from assuming the promotion; and Grace Miller’s progression has been set out at length heretofore. What is notable about this documentation is that it reveals an alternative route to the pharmacy route to management for females before July 1965. Only one pharmacist, Grace Miller, is among the nine females promoted to assistant manager, and the others were all salespersons. These promotions were made at a time in the company’s history when store-wide management positions became available usually only on retirement or death of a manager. The operation was small in comparison to the expanded operation after 1970 when the company grew from 24 stores to 82 stores at the time of trial, and before the expansion period the executive level personnel evaluated the employees from personal knowledge of their performance and their capabilities. The testimony of Genevieve Ronquette, Lillian Hennessey, Blanche Callihan and Inez Nungesser is to the effect that they had not been treated differently from males by K&B in promotion to management. Inez Nungesser commenced her employment in 1927 as a salesperson in the drug department of a high volume store on Canal Street in New Orleans. In 1953 the male assistant manager at her store died. She asked for the job and got it without experiencing any difficulty. She never aspired to become manager, even though the male manager retired before her retirement in 1972, and her testimony left no doubt that her entire employment at K&B had been a pleasurable experience. When the male manager left the store where Lillian Hennessey was assistant manager in 1954, she became the manager and remained the manager until her retirement in 1973. Blanche Callihan was approached on two occasions to become an assistant manager, and refused both times, before she finally accepted in 1965 a third offer to her for this position. She never applied for promotion to management and, when she finally consented to take the assistant manager position, her initial acceptance was only on a trial basis. She had the opinion that more women were not in management because they did not want the responsibility, and the fact that she was approached to be assistant manager rather than applying for a management position convinced her that women were not treated differently from men in promotion to management. Genevieve Ronquette replaced a male when she was further promoted to personnel director in 1950 and she remained in management until her retirement in 1969. Upon the effective date of the Civil Rights Act seven of the nine females still held management positions, but by the time Walter Feltman was hired as store operations manager in October 1966 two of them had retired and one had died leaving only four females in management at that time— Ronquette, Hennessey, Nungesser and Callihan. From that time until the Capaci charge was filed in 1973 five females were promoted to management, but none to store-wide positions. Three female salespersons were promoted to cosmetic supervisor, and two female pharmacists were promoted to chief pharmacist. There is also some evidence of isolated unsuccessful efforts about 1968 to recruit female employees into the manager trainee program, but no females entered that program until 1974. William Serda became personnel director in March 1971, and at that time there were still only four females in management,— one manager, one assistant manager, one chief pharmacist, one cosmetics supervisor. He realized that this situation made the company vulnerable to charges of violation of the federal law against discrimination, and for that reason and the additional reason that the company had an urgent need for manager trainees in the expansion program, which was within his area of responsibility, he made the recommendation to his superior Walter Feltman in 1971 or 1972 that females be encouraged to apply for managerial positions. The recommendation for affirmative action was taken under advisement, but was ultimately approved. The promotion record does not indicate that the implementation of Serda’s recommendation was immediate, and the written affirmative action policy was not published until 1976. Besthoff agreed that the Capaci charge was the impetus for the implementation of Serda’s affirmative action recommendation. Thereafter the promotion of females into management has continuously accelerated. Several of those who were promoted after Serda became personnel director made clear in their testimony that from their experience at K&B their impression was that gender was not a factor in the promotion process at K&B. The Advertising Evidence To further bolster its statistical evidence the EEOC relies on K&B’s manner of advertising for applicants for job vacancies during the period 1965-1971 as reflecting its policy of excluding females from managerial positions, and more particularly its policy of limiting the manager trainee positions to males. The EEOC researched the “Help Wanted” ads in the Sunday edition of The Times-Picayune, a New Orleans, Louisiana newspaper, from July 1965-July 1974, in the Morning Advocate, a Baton Rouge, Louisiana newspaper, at various times from August 1968-November 1974, in the Houma, Louisiana newspaper from October 1968-June 1975, and in the Lake Charles, Louisiana newspaper in December 1971 and January 1972. From this research it introduced 36 ads placed by K&B in the Times-Picayune from August 15,1965-January 10, 1971 and six ads placed in the Morning Advocate in August and September 1968. To buttress its policy manifestation argument it contends that these ads violated section 704(b) of the Civil Rights Act of 1964, as amended, 42 U.S.C. § 2000e-3(b) by indicating a preference for males in the manager trainee and certain other ads, and by indicating a preference for females in the cashier, counter personnel and salesperson ads. The Times-Picayune had columns designated “Help Wanted — Male”, “Help-Wanted — Female”, and “Help Wanted — Male or Female” which it did not discontinue until 1972. The manager trainee ads appeared in the male column and the cashier, counter girl, saleslady, sales clerk ads appeared in the female column. The Morning Advocate had columns designated “Male Help Wanted”, “Female Help Wanted” and “Help Wanted”. The manager trainee ads appeared in the male column and the pharmacist ads appeared in the neutral “Help Wanted” column. Genevieve Ronquette testified that she composed the ads for the Times-Picayune until her retirement in 1969, gave the order for the ads by telephone and designated the column in which they would appear. She had no instructions as to the column in which the ads should be placed, and she had a very simple method for making her placement decision, — she designated the column according to her experience as to whether males or females could be expected to apply. Females, not males, applied for cashier, counter and sales jobs; males, not females, applied for management jobs. In other words, the ads were placed to attract applicants and she placed them where she considered they would get the best results. Ronquette’s method of column designation was expressly approved by the EEOC in a guideline issued on April 22, 1966 to employers in the area of “Job Oppo