Dorothy Wheeler, on her own behalf and on behalf of a class, brought this Title VII sex discrimination action1 against the City of Columbus, Mississippi.2 The district court, after making findings of fact and conclusions of law, held for the City. Wheeler appeals the district court's decisions as to her personal claim and that of the class. We vacate and remand to the district court for supplemental findings of fact and conclusions of law. The class and individual claims are discussed separately below, after a brief review of the factual setting.

In January 1979, the City placed the following advertisement in a local newspaper:

Employment in a supervisory capacity of a small staff with responsibility for renting of auditorium. Requires high school diploma or equivalent. Some experience in supervising janitorial work, bookkeeping and public contact will be helpful.

The advertisement describes the job of city auditorium manager.

Dorothy Wheeler, a forty-four year old white female, applied for the position on January 22, 1979. Wheeler's educational background included an undergraduate degree and the completion of substantial work toward a masters degree in history. Her employment record included stints as a factory assembly-line worker, moving company sales manager, and cosmetics saleswoman. The district court found that Wheeler's only managerial work experience was as a beauty parlor manager.

Alton Lewis, a white male, was hired for the position Wheeler sought. Lewis had served for twenty years in the United States Air Force and retired with the rank of sergeant. Lewis' education included a high school equivalency certificate and the completion of sixty hours of work toward an undergraduate degree.

The City received twenty-six applications for the auditorium manager position. Those applications were screened by the City's personnel office, and then several of them were considered by the Mayor and City Council. Wheeler was never interviewed for the position. After interviewing numerous applicants for the job, the City hired Lewis. Subsequently, Wheeler brought suit, alleging that the City's action violated her rights under Title VII.

Wheeler's complaint against the City also purported to represent a class and sought class-wide injunctive and declaratory relief predicated on a pervasive pattern of gender-based discrimination. The district court conditionally certified a class composed of "all female employees of defendant, past and present, and all female applicants for employment with defendant, past and future."3

* In the words of the district court, Wheeler presented "a plethora of statistical evidence" at trial to demonstrate the existence of gender-based discrimination in the City's employment practices. Wheeler also presented evidence of specific instances of discrimination to bolster her statistical proof. Below, we outline the evidence presented by Wheeler.

Females account for fifty percent of the available workforce in Lowndes County, Mississippi, the relevant labor market for purposes of City hiring, and thirty percent of the recent applicants for City jobs. Only nine percent of those hired in recent years have been women, however, and at the time of trial, only seven percent of the City's employees were female. In 1973, seven years earlier, women accounted for ten percent of the City's workforce.4 The starting point, of course, for Wheeler's statistical case was the striking disparity in available workforce (50%) and recent applicants (30%), on the one hand, and actual employees (7%) and recent hirings (9%), on the other.

Wheeler also presented evidence that women employed by the City were concentrated in office and clerical positions. In 1980, ninety-five percent of the City's female employees occupied such positions. This concentration was shown to differ significantly from the general employment patterns in Lowndes County.5 Wheeler also contrasted the concentration of females in City office and clerical positions with the complete absence of women among the ranks of uniformed police officers, fire fighters, and streets and sanitation employees.

This segregation of females was shown to result in pay disparities. Eighty-five percent of the City's female employees earned less than $10,000 in 1980. Only fifty-two percent of the men in the City's work force earned less than $10,000 in that year.

In addition to introducing evidence of general quantitative and qualitative employment-related statistical disparities between men and women, Wheeler presented evidence of specific instances of discrimination and more narrowly drawn statistical evidence. Witnesses testified about their experiences with the City and discussed concrete examples of attitudes toward women employees, the channelling of women into office and clerical job slots, and pay disparities between men and women. Evidence was adduced to show that in thirteen male versus female job competitions the male prevailed in eleven. Wheeler narrowed the scope of her statistical attack in discussing the situations in three specific City departments: Streets and Sanitation, Police, and Fire.

The Streets and Sanitation Department consisted of over a hundred employees, all of whom were men. Wheeler offered evidence that women were qualified to hold many of the positions in that department, and that women constituted twenty-eight percent of the Lowndes County workforce in the job categories from which they were totally absent in this department. A pattern of recruitment and selection practices in the Department was alleged to perpetuate the exclusion of women. Openings in the Department are announced by word of mouth rather than public advertisement. Formal applications and personnel records are the exception rather than the rule. City officials who recalled referring to the Department men seeking any available job could not recall making such referrals for women. Although the City tried to justify the absence of women from Department positions by pointing to the low incidence of female applicants, the City did not explain either the barriers to females filing applications or the City's failure to hire the women who did indeed apply.

In the City's Police Department, all uniformed officers were male. In order to be eligible for such positions, an applicant must pass both a physical agility test and a written examination. In challenging the specific employment practices of this Department, Wheeler's general theory of disparate treatment merged with disparate impact theory. The disparate impact analysis of course centered on the propriety of the physical and written tests, while further evidence of the City's generally biased approach to women was presented. This evidence was bolstered by the testimony of a successful female test-taker who was not offered a job.

Wheeler's challenge to the Fire Department relied most heavily on the absence of female firefighters.

The district court held for the City and concluded that "the statistical evidence adduced at trial is insufficient to meet plaintiff's ultimate burden of persuasion as to the class action allegations of gender-based discrimination." (Emphasis supplied.) Significantly, the court departed from the manner in which the statistical evidence was presented in analyzing that evidence.

The court noted that for administrative purposes the City's work force is divided into twenty-three categories.6 In seventeen of those categories, there were eight or fewer employees. The court concluded "that statistics alone are insufficient to sustain plaintiff's burden of proof as to these seventeen categories of employment." The court opined that the small number of employees in those groups diminished the probity of the statistical evidence on them. The court went on to hold that "(s)ince plaintiff adduced no direct evidence to bolster her statistics in support of her class action allegations as to these seventeen categories of employment, the court finds that plaintiff has failed to sustain her burden of proof as to these seventeen of defendant's twenty-three categories of employment."

After eliminating the seventeen categories, the district court narrowed its focus. Two of the remaining six departments, Supervision and Finance and Police Civilian, were then eliminated because female employees significantly dominated their ranks.7 This process of elimination resulted in the district court's conclusion that Wheeler had established a prima facie case of discrimination based on statistical evidence in only four departments: Street Garbage, Street Cleaning, Fire, and Police. After noting that the statistical disparities in those departments were "striking," the court focused its analysis on those departments.

The district court collapsed the Street Garbage and Street Cleaning categories into one. The court found that the City, through the testimony of Casey Freeman, a supervisory employee with hiring responsibilities in the department, successfully rebutted Wheeler's prima facie case. Freeman testified that he had received no applications from women, would welcome applications from females, and would hire without regard to gender. The district court found that testimony credible. In addition, the district court found the application information both scanty and inconclusive. Wheeler failed to prove that the dearth of female applications for department positions was a pretext for unlawful discrimination.

The court's analysis of the Fire Department's hiring practices likewise focused on the lack of female applicants. The court found that 1.2 percent of the applicants for department positions over the past eight years were women, while the current workforce was 1.8 percent female.8 Because Wheeler failed to adduce evidence of a "chill" on the willingness of females to apply for department positions or direct evidence of specific discriminatory incidents, the district court concluded that Wheeler did not satisfy her burden of persuasion on the issue of discrimination.

The district court's analysis of Police Department hiring practices initially noted that the department employed less than two percent female and that fifteen percent of the department's job applicants over the previous eight years were female. The court treated Wheeler's case against the police department as being predicated on disparate impact theory and concentrated most of its attention on the physical agility test and the written examination that all applicants were required to pass in order to become eligible for a position. The physical agility test consisted of a one-hundred yard sprint, a standing broad jump, and pull-ups.9 The written test was not introduced into evidence, nor was proof adduced regarding the number or rate of success of female and male test-takers. The court concluded that Wheeler's disparate impact theory as to the written examination failed for lack of relevant evidence.

As for the physical agility test, one of the two females who took it passed. No comparative evidence of the male success rate was adduced. This weakened Wheeler's disparate impact case. Her case came down to Dana Medus, the only woman who succeeded in passing both tests and reaching the eligibility list, but who was not chosen for a position. The district court concluded that the evidence as to Medus, though sufficient to establish a prima facie case of individual discrimination, was insufficient to entitle Wheeler to class-wide relief on a disparate impact theory.

The district court, after analyzing department-by-department the units not culled out as too small, concluded that the City had successfully rebutted Wheeler's statistically-supported case.

Class action plaintiffs may assert a claim of disparate treatment or disparate impact. The disparate treatment class action plaintiff can prevail upon proof of a pattern and practice of gender-based discrimination by an employer. See generally International Brotherhood of Teamsters v. United States, 431 U.S. 324, 335, 97 S. Ct. 1843, 1854, 52 L. Ed. 2d 396 (1977). The proof must establish that gender-based discrimination is the employer's "standard operating procedure-the regular rather than the unusual practice." Id. at 336, 97 S.Ct. at 1855. It must also establish discriminatory motivation, either by direct or inferential proof. Id. at 335 n.15, 97 S. Ct. at 1854 n.15.

The same set of facts giving rise to a disparate treatment case can also be structured to make out a disparate impact case. Id. Disparate impact claims involve an employer's use of facially neutral employment standards that operate more harshly on one group than another and that are not based on valid employment-related considerations. Id. Proof of discriminatory intent, a requisite in the disparate treatment case, is dispensed with in a disparate impact case. Id.

Wheeler's case blended disparate treatment and impact theories. However, it is clear that the central thrust overall was disparate treatment in City employment practices with occasional attempts to bolster that theory with evidence of employment practices in certain departments which had a disparate impact on female job-seekers.

A disparate treatment plaintiff's prima facie case can be made out "solely with statistics if they show a sufficiently great disparity between the employer's treatment of (women) and (men)-a disparate result." Payne v. Travenol Laboratories, Inc., 673 F.2d 798, 817 (5th Cir. 1982). We stated in Travenol :

The statistical showing of disparate result may also be buttressed with evidence of a history of discrimination practiced by the employer, individual instances of discrimination, and opportunities to discriminate that exist in the employer's decision-making processes. If the statistical disparity is insufficient alone to establish a prima facie case, the plaintiff may get over his or her initial hurdle by combining statistics with historical, individual, or circumstantial evidence.

Id.

The Supreme Court has held that "(w)here gross statistical disparities can be shown, they alone may in a proper case constitute prima facie proof of a pattern or practice of discrimination." Hazelwood School District v. United States, 433 U.S. 299, 307-08, 97 S. Ct. 2736, 2741, 53 L. Ed. 2d 768 (1977). This is so because "absent explanation, it is ordinarily to be expected that nondiscriminatory hiring practices will in time result in a work force more or less representative of the racial and ethnic composition of the population of the community from which employees are hired." International Brotherhood of Teamsters v. United States, 431 U.S. at 339 n.20, 97 S. Ct. at 1856 n.20.

Wheeler presented statistical, historical, anecdotal, individual, and circumstantial evidence in attempting to make out a prima facie case of gender-based discrimination by the City. Wheeler presented evidence that women constituted fifty percent of the City's available workforce, thirty percent of the recent applicants for City jobs, seven percent of the City's employees, and nine percent of the recent hirings. Thus, at the outset, the district court was shown a striking disparity between the expected and actual presence of females in the City's employ. Wheeler also presented statistical evidence of: the concentration of women in office and clerical positions; their complete exclusion from other types of positions; salary disparities between men and women; and a decline in the percentage of female employees over the past several years. Wheeler bolstered her statistics with historical, anecdotal, individual, and circumstantial evidence of gender-based discrimination.

Wheeler's action and evidence were geared to show that the City was engaged in a pattern and practice of gender-based discrimination. While Wheeler used individual departments as examples, she challenged the City practices across-the-board. The district court misapprehended the thrust of the attack. It sought to analyze the statistical evidence on a department-by-department basis by using administrative classifications of City jobs. Neither Wheeler nor the City had attached any significance at trial to these twenty-three job classifications, which had been created by the City only for payroll purposes. These classifications did not in fact represent City departments. By its analysis the district court fragmented Wheeler's statistical case.

The court concluded that Wheeler's statistics, standing alone, were insufficient to establish a prima facie case as to seventeen of the classifications since eight or fewer employees were classified within each. This was based on the proper premise that statistical evidence of discrimination is often irrelevant or lacking in probative value when the sample group is small. See Turner v. Texas Instruments, Inc., 555 F.2d 1251, 1257 (5th Cir. 1977). The problem is that Wheeler had not attacked the employment practices of those individual classification groups. Where Wheeler did attack particular departmental practices (e.g., police and sanitation), her purpose was primarily to bolster her case against the City as discriminator.

After eliminating seventeen classifications for lack of a large enough sample group, the district court eliminated two others, Supervision and Finance and Police Civilian, because women constituted a majority of the employees in those departments. Again, we note that Wheeler had not attacked the hiring practices of those individual departments per se. Furthermore, an important component of her case was that the concentration, as opposed to exclusion, of women in low-paying and stereotypical job categories exemplified the City's discriminatory employment practices. The district court found that only three classifications, Streets and Sanitation, Police, and Fire, required analysis, and apparently concluded that Wheeler had made out her prima facie case based on statistical evidence that there were no women in those departments.

The district court's approach to Wheeler's evidence requires that we vacate and remand the class action case for supplemental findings of fact and conclusions of law. Wheeler's suit alleged a pattern and practice of gender-based discrimination against the City as a whole. The district court on remand should reconsider her statistical and other evidence in that manner. The court must determine whether her statistical evidence in and of itself, or together with other evidence, establishes a prima facie case of discrimination by the City. Her other evidence should be considered in like manner. Although the fact that women are in the workforce majority in the Finance and Police Civilian departments might absolve those departments if they were attacked alone, that is not the issue in this case. Indeed, the concentration of women in those departments could be probative of gender-based discrimination when contrasted with the exclusion of women from certain other departments. In addition, we emphasize that much of Wheeler's statistical evidence went beyond mere hiring percentages. Those statistics should be considered in light of Wheeler's across-the-board approach. With respect to the seventeen classifications that the district court excluded, the district court should ascertain whether some of those classifications belong in bona fide departments and also consider the statistics in those classifications as part of Wheeler's overall prima facie case.

We agree that Wheeler failed to make out a prima facie case for disparate impact with respect to Police Department hiring. However, the district court should enter additional findings in that area. The district court found that Wheeler could not prevail on a disparate impact theory since she failed to show that the facially neutral employment practices (written and physical tests) fell more harshly on women than men. Thus, Wheeler's disparate impact attack on the Police Department's hiring practices must fail. See Pouncy v. Prudential Insurance Co., 668 F.2d 795, 800 (5th Cir. 1982). Nonetheless, Wheeler's evidence of the complete absence of women from the ranks of the uniformed police force and the testimony of Dana Medus would continue to be pertinent to a determination of whether Wheeler made out a prima facie case of across-the-board gender-based discrimination.

If the district court concludes on remand that Wheeler has established a prima facie case, it must then determine whether the City has successfully rebutted that prima facie case. At the outset, we note that an employer's "general statements that it hired only the best qualified applicants" are insufficient. International Brotherhood of Teamsters v. United States, 431 U.S. at 342 n.24, 97 S. Ct. at 1858 n.24. The rebuttal may take the form of an attack on the plaintiff's statistics as being either inaccurate or insignificant, or an assertion of "a nondiscriminatory explanation for the apparently discriminatory result." Id. at 360 n.46, 97 S. Ct. at 1867 n.46. See Payne v. Travenol Laboratories, Inc., 673 F.2d at 817.10 If the defendants fail to rebut the prima facie case, plaintiffs prevail on the issue." Williams v. New Orleans Steamship Ass'n, 673 F.2d 742, 750 (5th Cir. 1982).

The district court concluded that a prima facie case had been made out in the Streets and Sanitation, Police, and Fire departments. The court then relied principally on a finding of absence of qualified applicants in concluding that Wheeler's prima facie case had been rebutted. On remand, the district court must supplement these findings if it concludes the City has rebutted a prima facie case of across-the-board, and not merely departmental, discrimination.

Caution should be employed in relying on applicant flow data as effective rebuttal for a prima facie case. It has been pointed out that such information "is frequently unavailable, and, when available, it is often distorted by inadequate or excessive recruiting efforts, improper deterren(ce) of applicants, unqualified applicants, multiple applications by the same applicant, or lack of specificity or improper groupings." B. Schlei & P. Grossman, Employment Discrimination Law 320-21 (1979 Supp.). While applicant data "would be very relevant ... (t)he absence of such evidence, however, will not prevent plaintiffs from recovering." Wilkins v. University of Houston, 654 F.2d 388, 396 n.8 (5th Cir. 1981). The court should again consider the pertinence of the applicant data that were presented as well as the effect of the absence of applicant data in other areas.11

In analyzing the City's rebuttal case, department-by-department evidence is of probative value, just as it is relevant in the prima facie case, but the degree of its relevance depends on other variables. If the court determines that there is substantial hiring autonomy in individual departments, then they should be considered individually. So too, if the district court on remand examines the practices of the departments-not just the three classifications with large sample groups and no women-and concludes that no discrimination occurs in any department, then the prima facie case could be rebutted. However, the departments are parts of a whole and one untoward incident from a single department may be probative of the City's intention to discriminate. Just as we have held "that segmented statistical evidence of racial imbalances within a few departments cannot prove a prima facie case of employer-wide discrimination," Williams v. New Orleans Steamship Ass'n, 673 F.2d 742, 748 (5th Cir. 1982), so too the court must not permit the balance of sexes or the unassailable employment practices of an individual department to absolve the entire City.

The factual background of Wheeler's individual claim was discussed at the outset of this opinion. After a complete trial, the district court entered judgment for the City and filed findings of fact and conclusions of law. The district court held that Wheeler had made out a prima facie case of gender-based discrimination applying the standards articulated by the Supreme Court in McDonnell Douglas Corp. v. Green, 411 U.S. 792, 802, 93 S. Ct. 1817, 1824, 36 L. Ed. 2d 668 (1973). Under the standard of shifting burdens articulated in Texas Department of Community Affairs v. Burdine, 450 U.S. 248, 254, 101 S. Ct. 1089, 1094, 67 L. Ed. 2d 207 (1981), the City offered, and the district court accepted, the superior qualifications of Lewis as a legitimate, nondiscriminatory reason for passing over Wheeler. The district court held that Wheeler failed to prove that the City's proffered reason for not hiring Wheeler was merely a pretext for what was actually intentional gender-based discrimination. The district court concluded "that Mr. Alton Lewis possessed better qualifications than plaintiff and that the (C)ity did not discriminate against plaintiff because of her sex in refusing to hire her for the position of auditorium manager." That conclusion, together with the City's evidence that Wheeler's application was duly considered, led the court to enter judgment for the City.

At trial Wheeler presented the following evidence to which the district court did not allude in its findings of fact. Wheeler testified that she requested an interview after filing her application. Such interviews were granted to all male applicants, but not to her.12 A member of the City Council, Glen Lollar, allegedly told her that the job of auditorium manager was not one for a woman. Wheeler claims that similar comments were made by other City officials. Furthermore, statistical evidence was adduced to demonstrate that gender-based discrimination in City hiring was a widespread phenomenon.

Wheeler appears to argue that the City's refusal to interview her barred her from consideration and of itself entitles her to recover under Title VII. This assertion, however, fails to recognize that the essence of her Title VII claim must be the refusal to hire based on impermissible gender-based discrimination. Thus, the failure to interview, standing alone, gives rise to no entitlement to recover. Nonetheless, the failure to interview may have independent significance even if it was not determinative of whether she was considered with other applicants.

In this case Wheeler had the burden of proving that Columbus' stated reason for not hiring her was pretextual. The City's failure even to interview her could help carry that burden. For example, that failure, together with Wheeler's evidence of sexist comments by City officials and a pervasive pattern of discriminatory hiring practices, would constitute proof probative of the existence of intentional discrimination. The Supreme Court in McDonnell Douglas noted that an employer's general employment policies and practices, demonstrated through statistical evidence or otherwise, are relevant in determining whether the employer's ostensible reason for not hiring an applicant is pretextual. 411 U.S. at 804-05, 93 S.Ct. at 1825.

In its Answer, the City asserted that Wheeler was not hired because "historically" the City and the Board of Supervisors of Lowndes County cooperated in a joint venture of hiring the same person to be both city auditorium manager and county veterans service officer. The City contended that Wheeler was not hired because of her lack of veteran status. The district court, in ruling for the City, relied only upon Alton's superior qualifications. On remand, the district court should also consider the requirement of veteran status and make appropriate findings on its relevance to the issue of individual gender-based discrimination in Wheeler's case. Cf. Personnel Administrator of Massachusetts v. Feeney, 442 U.S. 256, 279 n.25, 99 S. Ct. 2282, 2296 & n.25, 60 L. Ed. 2d 870 (1979).

The absence of a discussion of these factors, which, if properly established, could strengthen Wheeler's case on the ultimate burden, leaves this court without a confident base for review of the district court's conclusion that she was not the victim of sex discrimination.

Since we have concluded that the class certification and claim must again be reviewed by the court below, we remand Wheeler's individual claim also with directions to the district court to supplement its fact findings as indicated above and determine whether the supplemental facts found alter the outcome on the issue of intentional discrimination.

We vacate the judgment appealed from and remand for supplemental findings of fact and conclusions of law on both the individual and class claims. Our remand is a limited one. We are by no means convinced that the district judge's decision is in error. The foregoing opinion is intended to do no more than offer a proper analytical framework under which to consider this complex case. Our action connotes only that "we lack a basis for reviewing his (decision) because (he) has failed to support (it) by adequate evaluation of the facts and analysis of the law." In re Corrugated Container Antitrust Litigation, 643 F.2d 195, 218 (5th Cir.), after remand, 659 F.2d 1322 (5th Cir. 1981).13 We retain jurisdiction of this appeal pending the district court's compliance with our limited remand. Id. at 226.

The judgment appealed from is VACATED and the cause is REMANDED.