OPINION
McCUNE, District Judge.Plaintiffs, Rebecca Hopewell, Eunice G. Stewart, Gloria Proctor and Carol E. Wise have brought this action against defendants, the University of Pittsburgh (University) and the Board of Trustees of the University, asserting individual and class claims arising out of alleged violations of the Civil Rights Act of 1964, 42 U.S.C. § 2000e, et seq. and 42 U.S.C. § 1981. They now move for certification of this suit as a class action pursuant to Rule 23 of the Federal Rules of Civil Procedure (Rule 23), subsequent to extensive discovery on class issues. The class asserted includes a wide range of non-faculty, black employees employed at the main campus of the University, located in the Oakland section of Pittsburgh, Pennsylvania, who have allegedly suffered a wide range of discriminatory employment practices. After consideration of the arguments of the parties and the already voluminous record at this juncture, it is apparent that plaintiffs have not shown that common questions of law or fact predominate among the class members of a University-wide class, as required by Rule 23. Further, the named plaintiffs do not present claims typical of all class members asserted, due to the wide divergence of jobs held by the class which plaintiffs seek to represent. Because the claims they assert relate to violations directed largely at named plaintiffs, individually there is no evidence that these violations result from a general policy, custom, or practice of the University, or a failure to discourage individual acts of discrimination by the University, plaintiffs’ motion to certify a University-wide class is denied.
I
The Class Asserted
The Class which plaintiffs presently1 assert is composed of black persons who have been employed in three basic groups of job categories at the University: professional non-faculty jobs, technical or paraprofessional jobs, and secretarial or clerical jobs. These three categories contain 42 separate job descriptions as categorized in the Uni*692versity employment records. Each of these 42 job descriptions call for a separate pay schedule. In excess of 300 people are involved. The jobs range from computer programmers being paid an average base monthly salary over the period in question of $902.65, to assistant coaches being paid a base monthly salary of $1302.24 over the period in question, to clerical aides being paid a base monthly salary of $406.71 over the period in question. There is no indication of which offices or divisions (hiring units) of the University employ class members. It can be assumed, however, that persons holding secretarial or clerical positions would be employed in almost every office or division of the University. The wide dispersion of class members creates unique class difficulties due to the manner in which the University is organized.
The University of Pittsburgh is a large metropolitan University with branch campuses in several regional communities. The University educates approximately 35,000 students each year in its various schools. Administration of the University, due to its size and diversity, is decentralized. Each of 125 Departments, Divisions, Projects, and Special Centers constitutes an administrative unit of the University for purposes of day to day governance. Within these large units are approximately 300 sub-units referred to as budget units. A budget unit may be as small as one faculty member working on a limited project funded by an institutional grant. Hiring and personnel decisions and policies arise from these administrative and budget units. Insofar as individual employees are concerned, employment practices (including hiring, assignment of work, promotion, transfer into the department, merit salary increases, discipline, discharge and layoffs) emanate from these small units and effect only persons employed by these units.
The University maintains a personnel office which deals with personnel practices on a broader scale than the budget units. The personnel office is responsible for the preliminary recruitment and interviewing of job applicants. A pool of applicants is created from which the hiring unit selects the applicants it wishes to hire. The supervisor of a potential employee approves the applicant and directs the personnel office to extend a formal offer of employment. As a result, employment is a two level process. General University-wide employment practices are limited to the manner in which the applicant pool is created in the first step of the process. The second step, the actual hiring and general employment decisions, is directed by numerous autonomous administrative units. The class plaintiffs assert is employed by many, if not all, of these small units. As a result, the practices plaintiffs complain of are instituted by as many as a hundred or more unit administrators.
Plaintiffs assert that the members of the class as now defined have suffered a wide range of salary and hiring deprivations due to employment discrimination by these administrators. Plaintiffs allege that the administrators have restricted the hiring, classification and assignment of blacks to “low-paying, low-opportunity” jobs. They argue that in promotion, transfer, training and seniority policies, the University administrators impermissably discriminate on the basis of race. Finally, they allege generally that the University discriminates against blacks in “other terms and conditions of employment.” None of these practices complained of by plaintiffs arise from the personnel office in that plaintiffs do not challenge the manner in which the pool of applicants is created. All decisions relating to the practices plaintiffs question as discriminatory emanate from administrative units.
All of the factual allegations of discrimination made by plaintiffs can be broken down into two categories for purposes of analysis: hiring complaints and salary complaints. These complaints constitute the common questions of this action. Plaintiffs have given no indication of what might constitute their general complaints. Therefore, this category cannot be analyzed. Since plaintiffs assert two disparate factual patterns of discrimination, each type must be considered in relation to the *693class asserted in order that the question of whether these complaints are common to the members of the class can be answered.
II
Commonality
The four requirements of Rule 23(a) must be satisfied before an action can be certified as a class action, even in situations where certification would be otherwise salutory. East Texas Motor Freight System, Inc. v. Rodriguez, 431 U.S. 395, 97 S.Ct. 1891, 52 L.Ed.2d 453 (1977). Rule 23(a) requires, inter alia, that for an action to be certified as a class action, there must be “questions of law or fact common to the class.” In most Title VII actions this requirement of commonality is easily met, since impermissible discrimination is by its nature directed at a class of persons. In those instances where a commonality problem (or other Rule 23 deficiency) is not present, class certification is favored, since a class proceeding best provides remedies necessary for the vindication of discrimination. See, e. g., Wetzel v. Liberty Mutual Insurance Co., 508 F.2d 239 (3rd Cir.) cert. denied, 421 U.S. 1011, 95 S.Ct. 2415, 44 L.Ed.2d 679 (1975); Gay v. Waiters’ and Dairy Lunchmen’s Union, 549 F.2d 1330 (9th Cir. 1977) and Newman v. Piggie Park Enterprises, 390 U.S. 400, 88 S.Ct. 964, 19 L.Ed.2d 1263 (1968).
When plaintiffs seek to represent a class, the members of which hold jobs unrelated to those held by the representative plaintiffs, the existence of common factual questions cannot be assumed. Discrimination emanates from the source of decision making. When that source is on a low administrative level and is not centralized, it cannot be assumed that a broad class of employees are suffering from the discrimination suffered by individual employees. That discrimination may be the result of decisions made by low level administrators affecting only the named plaintiffs and a few others. For common questions to predominate in such a case, it is necessary that there be some affirmative showing beyond individual claims and general class allegations that discrimination has been suffered by the class. This required showing results solely from the requirements of Rule 23 and has no bearing on the larger question of whether a prima facie case has been shown. For purposes of this motion plaintiffs’ claims are assumed to be meritorious. See Eisen v. Carlisle & Jacquelin, 417 U.S. 156, 94 S.Ct. 2140, 40 L.Ed.2d 732 (1974).
Common questions of across the board discrimination can be shown in a number of ways. Most commonly, statistical evidence may indicate an overrepresentation of blacks in low paying positions. This could also indicate, as could deposition testimony or documentary evidence, that high level administrators may have known of discrimination on the part of supervisors and not discouraged these activities. In addition to statistical evidence, documents and depositions may indicate policies having class wide discriminatory effect. This evidence could also indicate the failure of high level administrators to discourage known acts of discrimination.
Discovery on the issue of common questions of fact was conducted over an 18 month period. During that time, plaintiffs were unable to produce documentary evidence or deposition testimony which would indicate that the class asserted is suffering from class wide discrimination. There is no indication that further discovery would produce such evidence. Plaintiffs have, however, produced a statistical profile of the employment records of the University which indicates that more blacks hold low paying, low opportunity jobs than high paying, high opportunity jobs. At question, therefore, is whether these statistics are sufficient to show, in and of themselves, the existence of common questions.
It is improper at this juncture to comment upon the sufficiency of plaintiffs’ statistical evidence in relation to the proof of a prima facie case. This is not to say, however, that the conclusions plaintiffs urge on the basis of this evidence must be accepted uncritically. For common questions to exist, plaintiffs’ statistical evidence *694must logically support the inference of discrimination against the class asserted. Employment statistics may suggest at first reading an impression of employment practices which do not comport with reality. It is necessary to analyze statistical evidence logically, particularly in an employment context, due to the wide range of significant variables at work in any employment decision. Certainly, “statistics are not irrefutable . . . their usefulness depends on all of the surrounding facts and circumstances.” International Brotherhood of Teamsters v. U. S., 431 U.S. 324, 97 S.Ct. 1843, 52 L.Ed.2d 396 (1977).2
The primary difficulty with statistical evidence arises when two unrelated groups are compared. In Hazelwood School District v. U. S., 433 U.S. 299, 97 S.Ct. 2736, 53 L.Ed.2d 768 (1977), the Supreme Court affirmed a reversal by the Eighth Circuit of a finding by the District Court of the absence of prima facie discrimination based on a comparison of the percentage of black teachers employed by the defendant school district and the number of black students in the district. The Circuit considered this comparison to “shed no light whatsoever upon the dispute,” since it gave no indication of the relationship between the black teachers hired and the available black teaching workforce. The comparison was irrelevant, since it compared statistical profiles of groups which were different in kind. 534 F.2d 805, 812 (8th Cir. 1976). See also Mayor of Philadelphia v. Educational Equality League, 415 U.S. 605, 94 S.Ct. 1323, 39 L.Ed.2d 630 (1974); Patterson v. American Tobacco Co., 535 F.2d 257 (4th Cir. 1976); James v. Wallace, 533 F.2d 963 (5th Cir. 1976).
Plaintiffs’ comparison of the number of low salary black employees to the number of high salary black employees is a comparison of two groups which are different in kind, thus it is irrelevant evidence.3 Higher salary positions generally require greater skills than lower salary positions. The number of qualified black applicants for each group is different. The proper comparison, which would indicate class wide discrimination, is of the number of black persons employed by the University in a given position with the number of qualified black applicants for the position or with the number of qualified blacks in the workforce.
The University has provided these statistical comparisons, which indicate that in both skilled and unskilled positions,4 the University employs a greater percentage of blacks than are in the skilled and unskilled workforces, respectively, in the area. On *695this basis, therefore, there is no statistical indication of a common scheme of discrimination by the University. In light of these statistics, the ambiguous statistics of the plaintiffs regarding hiring complaints must be considered irrelevant.
Plaintiffs also present salary claims based on statistical evidence. This evidence is also too ambiguous to support a finding of common questions in relation to salary claims. Plaintiffs submit Table 5 through 13 of their statistical profile to support their allegations of common questions of salary discrimination. Each table, however, is structured in such a way as to exclude significant variables from consideration. All the tables are consistent with a proposition that there are no common questions of discrimination present, and all are irrelevant under the Hazelwood standard.
Table 5, for example, lists base monthly salaries for the years 1972 through 1976 of blacks and others for the class as a whole. The others are paid more as a group. This observation results from the fact that more whites in the job categories included in the class have higher paying jobs than blacks. This is evident from plaintiffs hiring statistics which find more blacks in lower paying positions. Since this comparison has been found to be irrelevant, the salary table is equally suspect.
Table 6 lists the average date of hire for the years 1972 through 1976 of the entire class. This table is submitted to prove that black and other employees in the class are equal in seniority. Again, in attempting to prove equal seniority, plaintiffs have compared dissimilar groups. The significant comparison of seniority is by job. Employees who have served many years in low paying jobs will be paid less than employees who have little seniority in high paying positions. When these groups are compared, the average seniority across the board of blacks and others may be approximately equal (others, by and large, having high seniority in high positions, low seniority in low positions, and blacks having high seniority in low positions and low seniority in high positions), but since others are employed in higher paying positions, the statistics will show a disparity in pay. These results do not prove commonality because the jobs in mass are compared to show that seniority is about equal. The proper comparison would be to compare seniority in the same jobs. When this is done in the University statistics, the black employees with seniority equal to the others, are paid equally.
Table 7 is again a restatement of the same logical error made throughout plaintiffs’ statistics. Table 7 lists employees for the years in question by race and salary. Blacks were found to have a large number of lower paying positions. This is a restatement of the statistical allegation that more blacks are employed in low paying jobs than in high paying jobs. Since this statement has been shown to be consistent with an absence of discrimination, it is irrelevant to show the existence of common questions of discrimination. The comparison is between two dissimilar groups, and is thus inadequate as a basis for any conclusions.
The same error is repeated in Table 8, which lists average base monthly salaries for the year by race and hire year, tabulating the discrepancy between black and other salaries in each hire year. This table makes no allowance for different jobs held by individuals. It compares laborers to computer programmers. Since it is apparent that more blacks hold lower paying jobs than higher paying jobs, it is clear that when all jobs are taken as a mass, there will be a discrepancy between white and black salaries.
Plaintiffs’ additional tables are merely variations on the same theme. Each excludes a significant variable which explains the variance between white and black salaries, whether that variable be seniority (Tables 9 and 13) or skill (Tables 10, 11, and 12), or a combination of both. Plaintiffs’ statistics prove only one thing—that more blacks are employed and have greater seniority in lower paying jobs than higher paying jobs. In light of the University’s unambiguous statistics, this conclusion does not support a finding that there is a consistent class wide problem with discrimination.
*696In conclusion, on the basis of the evidence produced, we are not convinced that there are common questions of fact presented by this action as applied to the class asserted. This holding is a limited one, however, which is based solely upon the unusual facts reflected in the record. The University is structured in such a way as to assign employment decisions to low level administrators, each having only a limited number of class members within their authority. The plaintiffs have produced no evidence of a policy or general practice of discrimination during extensive discovery. There is no indication that further discovery is needed to find commonality between persons working for different Administrators. The only evidence which suggests class-wide discrimination is uncorroborated statistics which are totally consistent with the proposition that there are no common questions of fact. There is no guarantee that the problems plaintiffs challenge are the same as those of other class members. Few other class members have complained of discrimination. In light of these facts, there is insufficient commonality between the class members.
In addition to problems of commonality, the asserted class presents problems of typicality and management. There is little similarity between the members of this large diverse class. It is difficult to see how any plaintiff could present claims which would be typical of even a majority of the class. See Gardner v. Westinghouse, C.A. No. 75-614 (W.D.Pa., filed Feb. 3, 1976) appeal dismissed 559 F.2d 209, dismissal affirmed, - U.S. -, 98 S.Ct. 2451, 57 L.Ed.2d 364 (1978). Further, given the fact that discrimination, if any, emanates from low level administrators in this action, this case would be unmanageable in the presently asserted class form, since all administrators would present specific factual questions of discrimination.
. At the commencement of this action, plaintiffs sought to represent a class composed of all non-faculty blacks who had been employed by the University at any time since September 1, 1972, who had allegedly suffered a wide range of discriminatory employment practices. This early definition of the class proved to be too broad to be manageable, however, and the class was reduced in size by stipulation of the parties. Unionized employees, persons employed at other campuses than the main Oakland Campus, and persons employed at the Western Psychiatric Institute and Clinic were excluded from the class. This reduction resulted in a more manageable, but still large, class composed of employees at the Oakland campus in widely divergent job classifications. This class definition was retained through the entire period of discovery.
Briefs were prepared and submitted on the issue of class certification. At oral argument, plaintiffs submitted an amended motion for class determination limiting still further the proposed class. The University did not object to the timing of this amendment, even though the briefs, experts’ reports, and depositions were based on the larger class definition.
Because the University has not objected to the amendment, and because it is possible to relate the arguments and record heretofore developed to the amended class, the amended class will be considered in determining the propriety of this lawsuit proceeding as a class action.
. The Teamsters case discusses statistical evidence in the context of proving a prima facie case of discrimination on the merits, as do other cases discussed herein. These cases criticize the use of statistical evidence due to its evidentiary infirmity as applied to support specific inferences. Since these cases are evidentiary in import, there is no reason to consider them only in relation to the proof of a prima facie case of discrimination. It is only proper that statistics be given the same rigorous analysis in a Rule 23 determination as would be given on the merits. The question here is not what the statistics prove, but rather what they mean. When purged of logical ambiguity, statistical evidence can assume the exegetical role it is frequently assigned. See Alabama v. U. S., 304 F.2d 583, 586 (5th Cir.) aff’d per curiam, 371 U.S. 37, 83 S.Ct. 145, 9 L.Ed.2d 112 (1962).
. The comparison of low and high salary blacks may suggest that discrimination has been practiced, i. e., a finding of more low salary blacks than high salary blacks is consistent with a finding of discrimination in hiring or promotion. Consistency, however, is not enough. Given the present constitution of the workforce, a finding of more low salary blacks is consistent with an active affirmative action program as well. Since the statistic alone is equally consistent with a finding of discrimination or no discrimination, it is too ambiguous to stand as relevant evidence. When the number of blacks qualified for various positions is known, the statistic can be made unambiguous by relation to the relevant workforce. Also, when there is other evidence of discrimination or common questions of discrimination, the statistic takes on greater reliability. Where, as here, however, the statistic is presented in vacuo (indeed, in opposition to more reliable evidence of no common questions) the statistic must be considered irrelevant.
. The University’s statistics show that black employees' on the average have earned the same or perhaps more than white employees in comparable jobs.