This is a race discrimination case in which the individual plaintiffs and the United States appeal from the judgment of the district court which rejected their claims of discrimination by the North Carolina Agricultural Extension Service.1
The case was filed in 1971 based on alleged violations of the First, Fifth and Fourteenth Amendments to the Constitution, 42 U.S.C. §§ 1981 and 1983; Title VI of the Civil Rights Act of 1964, 42 U.S.C. § 2000d et seq. (prohibiting discrimination in programs receiving federal funds); and 7 U.S.C. § 341 et seq. (funding authority of the United States Department of Agriculture). The United States intervened on the same causes of action. The complaints were both subsequently amended to incorporate alleged violations of Title VII of the Civil Rights Act of 1964, 42 U.S.C. §§ 2000e et seq. (employment discrimination).
After a ten week trial, the district court rendered its decision in two separate opinions issued August 20 and September 17, 1982. In addition to an opinion on the broad aspects of the case such as the pattern and practice claims, the district court, by a separate opinion and order, dealt with and disposed of the individual claims.
On appeal, the individual appellants challenge the rejection of class certification and of their individual claims of race discrimination in their salaries and in the selection of County Extension Chairmen. They also challenge the district court’s refusal to order the integration of the local 4-H clubs and Extension Homemaker clubs. The government and the individual appellants argue that the district court improperly concluded that a pattern or practice of racial discrimination was not proved with respect to the salaries of county-level employees and State Specialists, or with respect to the selection of County Chairmen. In addition, they claim that the district court improperly concluded that the Extension Service quartile system, used for merit pay and promotion evaluations, was valid. We affirm.
The district court made extensive factual findings regarding the structure and practices of the Extension Service, so only a brief review of the facts pertinent to the claims on appeal is necessary.
The Extension Service is a division of the School of Agricultural and Life Sciences of North Carolina State University (NCSU) at Raleigh. Upper level administrative positions in the Extension Service are faculty appointments in the School. At the county level, a County Extension Chairman (County Chairman) is in charge of each county’s Extension Service office. In addition to substantive responsibilities among the Service’s four program areas (agriculture, home economics, 4-H, and community resource development), he oversees the county’s overall extension program and manages the county professional staff. The staff consists of assistant, associate, and full agents, who provide services in the program areas. Actual tasks are determined by each county’s specific needs based on primary crops, geography, population, etc. Above the county level, a District Extension Chairman is responsible for the overall administration of the several counties within his district. He is in regular contact with the county professional staff and participates in performance reviews biannually. The Extension Service also has 12 District Program Leaders (responsible for one program in 2 districts) and approximately 25 State Specialists (essentially researchers in particular subjects, such as soybeans, cattle, horticulture, adult education), who are available to assist *666county staff members with specific programs. Many of the State Specialists work at NCSU in Raleigh.
When a vacancy exists for a County Chairman position, it is noted in the Extension Service monthly announcements. Applicants who meet objective requirements (experience and education) are interviewed separately by the Director of the Extension Service, his Associate Director, the Assistant Director for County Operations, and the relevant District Extension Chairman. The group then meets and makes a recommendation on the appointment to the Board of County Commissioners where the vacancy exists. Under a Memorandum of Understanding in effect between the Extension Service and the counties of North Carolina, the county Boards have authority to reject Extension Service recommendations.
The Extension Service has been in existence since long before the passage of the Civil Rights Act of 1964. Prior to 1965 it was divided into two branches, a white branch and a Negro branch. Both branches reported ultimately to the Director of the Extension Service. On August 1, 1965 (shortly after the effective date of the 1964 Civil Rights Act), the two branches were merged into a single organization. Salary disparities which had existed between the two branches were not immediately eliminated. By 1972, however, when public employers such as the Extension Service were required to comply with Title VII,2 it had already been operating on a nondiscriminatory basis for about 6 years. Some pre-existing salary disparities continued to linger on nevertheless.
Since the merger, the Extension Service has set the minimum starting salary for all new county level agents. A salary differential is paid for a master’s degree, prior relevant experience, and special skills. Once in the employ of a particular county office, however, each agent’s salary is affected by a number of factors, not all within the control of the Extension Service. Each county contributes differing percentages of its local extension agents’ salaries, ranging from a low of 18% in Camden County to a high of 57% in Forsyth County. The balance of agents’ salaries is a combination of state and federal funds, paid by the Extension Service. Both the counties and the state legislature subsequently determine pay increases by their contributions to the salaries, sometimes in the form of a flat sum to each employee, sometimes as percentage raises. There is no uniform procedure for pay increases, and they are not always annual occurrences. In addition to regular salary increases, the State and about 10 counties provide funds for merit pay increases,3 which are awarded based on the past year’s job performance. The exact amount is determined by the amount of merit funds available in each county and the agents’ job performance during the previous year, as measured by their performance according to the Extension Service quartile system and the Performance Review Guide. Increases to keep more senior employees’ salaries above inflationary increases of entry level salaries are set by the Extension Service and also come out of merit funds. Thus, wide variations among salaries can and do develop with the passage of time.4
*667Under the quartile system, each employee is placed in one of four quartile groups annually by the District Extension Chairman. Placement is based upon an evaluation of the employee, using the Performance Review Guide. Under that system, employees of like grade and experience are placed in the first, second, third or fourth quartile according to their relative performance during the previous evaluation period, so that an employee in the first quartile has, in general terms, performed better than one in the second quartile, etc. Within each quartile, there is no ranking, however. The Performance Review Guide measures the performance of each employee against goals and objectives and performance of duties called for in his job description. The evaluation under the guide takes into account the observations of the County and District Extension Chairmen, program leaders and specialists. It also includes an interview or interviews with the agent himself who is required to sign the evaluation made of him according to the Performance Review Guide.
The Extension Service-sponsored local 4-H and Extension Homemaker clubs are established on a wholly voluntary basis at the community level. The Extension Service requires each club to be organized without regard to race and to certify that its membership is open to all races. Although voluntary membership may result in single-race clubs, all activities above the local community level are conducted on an integrated basis (including summer camps and area activities). The Extension Service provides services to local clubs equally, regardless of their racial makeup.
We proceed to the specific issues raised on appeal. Our review of the district court’s findings of fact is subject to the clearly erroneous rule. FRCP 52(a); Pullman-Standard v. Swint, 456 U.S. 273, 287, 102 S.Ct. 1781, 1789, 72 L.Ed.2d 66 (1982).
A. Class Certification
In the district court the claims for class certification were quite broad, embracing a wide variety of alleged wrongs, which could have been based on race, in the operation of the Extension Service. On appeal, however, the claims have been somewhat narrowed and are that the district court erred in its refusal to certify classes consisting of: (1) all black employees of ... [the Extension Service] on or after November 18, 1971; (2) all black members and potential members of ... [the 4-H and Extension Homemaker Clubs] on or after November 18, 1971; and (3) all County Commissioners in North Carolina who held that position on or after November 18, 1971. The first two classes mentioned are classes of plaintiffs, and the last a class of defendants.
In arriving at our conclusions, we have had the benefit of, and have taken into account, the findings of the district court at trial, for the plaintiffs pressed the question of class certification from time to time throughout the proceeding as late as argument of the case on the merits. This is not always, or even usually, the case, for the certification of classes at earlier times during a proceeding is many times required and should be as early as practicable. FRCP 23(c)(1). In this connection, we note that the intervention of the government in the case, asserting its broad pattern or practice claims which would have required relief as broadly based as the claimed plaintiffs’ classes would have received, made it possible for the plaintiffs to introduce all evidence they would have been able to introduce had the classes been certified. All the questions on the merits were also presented to the court which would have been presented had the classes been certified. Thus, in a practical sense at least, even if not in a technical one, the complaint on appeal on account of failure of class certification may seem to be of little moment or was made moot by the intervention of the government.
The class of employees seeks to litigate the legality of Extension Service service to single race clubs, the existence of salary discrimination against black employees, and the existence of discrimination in the *668selection of employees for the position of County Chairman. While it is doubtful that the legality of Extension Service services to single race clubs is a proper subject of class representation by employees not members of the clubs, Hill v. Western Electric, 596 F.2d 99 (4th Cir.1979), it is discussed on its merits at the end of this opinion. The district court declined to certify the class of employees because there was no evidence that there existed any other potential class members, in addition to the named plaintiffs, who had been subjected to any alleged discriminatory employment practices of the defendants. While this finding is not clearly erroneous, it is qualified, as the district court found, by the fact that other witnesses testified on behalf of the pattern or practice claims who would have been members of the class had the pattern or practice claims succeeded. Thus, alone, it may not be sufficient to sustain the district court’s ruling on that point.
An alternate reason for denying the class certification of the employees given by the district court was the lack of typicality of their claims under FRCP 23(a)(3). The salaries paid to the employees are made up from federal monies which are not in question here, appropriations from the state legislature, and the county governments involved. Increases are either across-the-board, merit, or percentage increases and one county in North Carolina has nothing whatsoever to do with the amount of increase given by another county. Thus, the claim of a potential plaintiff against one county will not be typical of the claim of another potential plaintiff against a different county. The same reasoning applies to the selection of the position of County Chairman. That selection is subject to approval of the county involved, and the fact that one county may discriminate in the selection of its county chairman has nothing to do with whether another county discriminates. Thus, we feel that the claims are not typical in the sense that they should support a class certification and are more nearly like those asserted in Stastny v. Southern Bell Tel. & Tel. Co., 628 F.2d 267 (4th Cir.1980), in which we held that promotion and pay decisions subject to almost complete local autonomy in the various offices of Southern Bell throughout North Carolina would not support the typicality requirement under FRCP 23(a)(3) for a statewide class of employees.
The second class of black members and potential members of the 4-H and Extension Homemaker Clubs was not certified as a class because the district court found there was no evidence that anyone was denied services or provided inferior services because of their race; neither, it found, was there any evidence that any black female had been denied membership in any Extension Homemaker Club or had been denied services or provided inferior services because of her race. The district court further found that in only one case had a black child been denied membership in a 4-H club because of his race, and this resulted in the prompt removal of the offending volunteer worker by the Extension Service. The court found there was no evidence whatever that any other black child was ever denied membership in a 4-H Club, or denied 4-H Club services, or provided inferior services by the Extension Service. It thus denied certification of the class. The plaintiffs do not take exception either to the findings of fact of the district court or to its conclusion that the class should not have been certified on the basis relied upon by that court. They say, however, that the district court misunderstood the issue, which is not whether “all white clubs reject black applicants on account of race, but what affirmative steps, if any ..., the Extension Service must require as a condition of state services and materials.” Parenthetically, no issue is made of “materials” as contrasted to “services.” The plaintiffs’ statement of their proposition shows the flaw in their claim. While we do not think their present characterization of the issue is correct, but that the district court correctly defined the issue, even if the claim is properly made, it is at once apparent that if there has been no *669denial of membership or services on account of race, there has been no legally cognizable wrong.
The plaintiffs take exception to the district court’s ruling that, under General Telephone Company v. EEOC, 446 U.S. 318, 100 S.Ct. 1698, 64 L.Ed.2d 319 (1980), class action certification is inappropriate and unnecessary in pattern and practice suits brought by the EEOC and the government. We do not find it necessary and we do not pass upon the correctness of that ruling for the reasons stated just above, which we repeat for emphasis. The government’s intervention in this case asserting its broad pattern and practice claims enabled the plaintiffs to have the benefit in the case of all the evidence they would have been able to introduce had their classes been certified. The relief sought by the government was also as broad as that asked for by the plaintiffs. The questions presented were as broad. Thus, the district court’s alternate ruling that class certification was unnecessary and inappropriate in a pattern or practice case has no practical effect of any kind on the outcome of this case, and we think the resolution of that question is better left for another day.
The district court declined to certify the class of County Commissioner defendants because it found there “... was simply no evidence of any standardized practice among the 100 separate counties in the state to deprive anyone of any rights solely because of race.” Again, this finding is not clearly erroneous.
The only way in which the counties were alleged to have acted to deprive plaintiffs of any rights because of race was because of the failure to hire black County Chairmen and because of pay increases of whatever nature granted by the counties. Yet, each county acted individually in its approval or disapproval of County Chairmen recommended for employment by the Extension Service. They acted according to no uniform ruling, and there is no allegation that they did. There is no allegation that the counties had anything to do with the circulation of information about job openings, or that the circulation of job openings was discriminatory. Indeed, there were no black applicants for county chairmen for most of the vacancies for the job in the counties in the State. With respect to pay, there is only one tiny facet of the district court’s fact finding which could have favored the plaintiffs and the government, that counties designated to which individual employees raises were given, and that fact finding has been demonstrated by appellants to be clearly erroneous. Otherwise, pay increases granted by county governments were across-the-board, or for merit, or a percentage, and there is neither allegation nor proof that there was any discrimination by the counties in so awarding them.5 The plaintiffs press their claim for the certification of that class of defendants under FRCP 23(b)(1)(B) that adjudications with respect to one county would .as a practical matter be dispositive of the interests of other counties. Of course in this case that cannot be true. If Brunswick County in the southeastern part of the state, for example, had deprived a black applicant employment because of his race or deprived him of pay raises because of his race, that could have nothing to do with whether Ashe County in the northwestern part of the state had done the same thing. In this connection, two cases are persuasive, although decided under FRCP 23(b)(2). In Paxman v. Campbell, 612 F.2d 848 (en banc) (4th Cir.1980), we held that a class of defendants of 130 Virginia school boards was inappropriate when each was free to adopt maternity leave policies of entirely unknown differences or similarities, it being uricontradicted there was no statewide policy in force, centrally directed or otherwise. The Sixth Circuit followed Paxman in Thompson v. Board of Education, etc., 709 F.2d 1200 (6th Cir.1983), upon the same fact situation obtaining in Paxman. While those two cases were decided under a different part of the class action rule, the principle remains the same, *670that to have a proper class of defendants in a case such as this there must be either a statewide rule or practice so that relief is available if the rule or practice is invalid, or the adjudication with respect to a member of a defendant class must as a practical matter be dispositive of the interests of the other members of the class as provided in FRCP 23(b)(1)(B). Neither of those conditions is present here.
The decision of the district court declining to certify the class of county commissioner defendants is thus without error.
B. Individual Salary Claims on Appeal
In addition to the denial of class certification, certain individual appellants appealed their claims of salary discrimination.
These salary claims were rejected by the district court. Simple salary comparisons were introduced to show some white agents who had no longer tenure or no more education earning more than the plaintiffs. Plaintiffs made no showing regarding their job qualifications and job performance which would raise their salary expectations to the level of an inference of discrimination.
In cases of individual disparate treatment, the showing of a simple disparity is not sufficient to make a prima facie case of discrimination. The rule of McDonnell Douglas Corp. v. Green, 411 U.S. 792, 802, 93 S.Ct. 1817, 1824, 36 L.Ed.2d 668 (1973), is that plaintiffs must show that they are qualified for an employment benefit which was denied them because of their race. Though evidence of salary disparities is relevant to the question at hand, the plaintiffs must show that they were qualified to be receiving a higher salary in order to shift the burden of production to the employer. Texas Department of Community Affairs v. Burdine, 450 U.S. 248, 252-253, 101 S.Ct. 1089, 1093, 67 L.Ed.2d 207 (1981); Larson, 2 Employment Discrimination §§ 50.22, 50.31(c) (1983). Lacking any data as to the relative qualifications of these plaintiffs compared to those who allegedly received higher salaries or as to the qualifications demanded by the jobs involved, the district court properly found no prima facie cases of disparate treatment had been made on behalf of these individual plaintiffs.
The individual appellants also raise on appeal the fact that they should be included in a class discriminated against which is the subject of the pattern and practice claims and discussed later in the opinion.
C. Wage Disparity Due to Pre-Act Discrimination
The plaintiffs claim that the pre-Act discriminatory difference in salaries should have been affirmatively eliminated but has not. We do not think this is the law. “A public employer who from that date forward [date of the Act] made all its employment decisions in a wholly non-discriminatory way did not violate Title VII even if it had maintained an all-white work force by purposefully excluding Negroes.” Hazelwood School District v. United States, 433 U.S. 299, 309, 97 S.Ct. 2736, 2742, 53 L.Ed.2d 768 (1977). Along the same line, in United Airlines v. Evans, 431 U.S. 553, 97 S.Ct. 1885, 52 L.Ed.2d 571 (1977), a present seniority and thus salary disparity which existed on account of a previous unlawful termination on account of sex which had resulted in loss of seniority was held not to be actionable when no complaint had been filed by the plaintiff on account of her unlawful termination, although the court acknowledged that “the seniority system gives present effect to a past act of discrimination,” 431 U.S. at 558, 97 S.Ct. at 1889 (1977). To the same effect are Fowler v. Birmingham News Co., 608 F.2d 1055 (5th Cir.1979) (present effect of uncomplained of racially discriminatory placement on seniority list held not actionable); Trabucco v. Delta Airlines, 590 F.2d 315 (6th Cir.1979) (present lower pay and fringe benefits resulting from past uncomplained of unlawful reclassification on account of sex held not actionable); Cates v. Trans World Airlines, Inc., 561 F.2d 1064 (2d Cir.1977) (present adverse seniority effects resulting from an uncomplained of racially discriminatory refusal to hire are not actionable). See also Schlei and Grossman, Employment Discrimination Law (2d Ed.), p. 1053.
*671D. Salaries
Much of the dispute in this case centers around salaries of the county' agents, whether full agent, associate agent, or assistant agent. The claim is that the salaries of black agents were lower because of their race.
The district court correctly declined to compare or include the salaries of County Chairmen with those of the agents. There is about a $4000 average salary difference between chairman and agent to begin with. The chairman is the head of the county based employees of the Extension Service. He is not routinely promoted from agent; the method of his selection is entirely different, and counting the salaries of the County Chairmen together with those of the agents for purposes of analysis could only distort the result.
At the outset of this part of the discussion, certain matters admitted or found by the district court and not appealed are important to remember. There is no discrimination in hiring. There has been no discrimination in entrance salaries since 1965. Neither is there discrimination in promotion from assistant to associate to full agent. Pay raises are of three kinds from whatever source, percentage raises, across-the board raises, and merit raises. There is no discrimination in across-the-board and percentage raises. Everyone gets them.6
This leaves only two sources of any discriminatory salary differential, the lingering effects of pre-Act discriminatory pay and the administration of the quartile system of ranking of performance of agents. Under the quartile system, each agent is placed in the first, second, third, or fourth quartile, according to his performance for the previous period. Approximately equal numbers of agents are supposed to be placed in each quartile. So, while the placing of an agent in a quartile follows a review of his performance under the Performance Review Guide, because of the necessary division of the agents into four quartiles, the performance of one agent relative to another is bound to be considered. The assignment to a quartile is made by the District Extension Chairman periodically, and merit pay increases, as distinguished from across-the-board and percentage increases, are ordinarily granted to those agents in the top three quartiles, and from time to time an agent’s higher quartile rank will result in a higher merit pay increase. Various counties contribute varying amounts of the agents’ pay, and, indeed, agents are paid by two checks, one from the county, the other from the Extension Service. Some counties pay higher salaries than do others.
The plaintiffs, as did the defendants, introduced the testimony of a statistical expert. Each of the statisticians had performed a multiple regression analysis with respect to the salaries of the county agents. The plaintiffs’ expert testified with respect to the years 1974, 1975 and 1981, while the defendants’ expert testified with respect to 1975 and 1981. A regression analysis is a method for examining the relationships among large numbers of variables. Such an analysis indicates the degree to which changes in the value of one variable corresponds to the value of the other. Barnes, Statistics as Proof (1983), p. 293 et seq.
The plaintiffs’ expert considered the variables of tenure, education, and race, and was of opinion there were statistically significant components of salaries attributable to race for the years 1974, 1975 and 1981, but he had not included job level or title (assistant, associate, or full agent) in his *672analysis. When job level was added, he did not find a statistically significant component in 1981. In the most general terms, the defendants’ expert agreed for the years 1975 and 1981. He offered isolated data analyses, however, which tended to show no discrimination. He had included job title in his analysis. The district court refused to accept the plaintiffs’ expert testimony as proof of discrimination by a preponderance of the evidence because the plaintiffs’ expert had not included a number of variable factors the court considered relevant, among them being the across-the board and percentage pay increases which varied from county to county. The district court was, of course, correct in this analysis. It defies logic to compare the salary of an agent in a county which has granted many or large increases or pays high salaries with one in a county which has granted few or small increases or pays lower salaries. The across-the board and percentage pay increases granted by various counties in varying amounts, as well as simply paying higher salaries, are bound to have an effect on the salaries of the agents in the various counties. In addition, as such increases accumulate through the years, greater disparity from county to county may certainly be expected rather than less. Percentage increases for all employees also would increase the dollar amount of existing disparities, and dollar amounts are what the statisticians dealt with.
Most of plaintiffs’ salary case is built upon the multiple regression analysis of their expert, and we think the district court was not required to accept that testimony as proof by a preponderance of the evidence. In the first place, the analysis contained salary figures which reflect the effect of pre-Act discrimination,7 a consideration not actionable under Title VII but permissible to show the general background of the case, or intent, or to support an inference that such discrimination continued. Compare Hazelwood, p. 809 thereof and n. 15 of 433 U.S. and p. 2742 thereof and n. 15 of 97 S.Ct. with Lehman v. Trout, — U.S. -, 104 S.Ct. 1404, 79 L.Ed.2d 732 (1984). Since the data used by the statisticians included pre-Act hires, there is bound to have been a racial component in each of their regression analyses, for the Extension Service admits that, while it had made some adjustments to try to get rid of the salary disparity resulting on account of pre-Act discrimination, it has not made all the adjustments necessary to get rid of all of such disparity. The across-the-board and percentage pay increases granted by a county, as well as counties which simply pay higher salaries, also are bound to flaw the regression analyses since they were not taken account of. It is also apparent that percentage pay raises from whatever source will magnify any existing disparity. Leaving aside for the moment the question of merit pay increases, we do not believe the district court was required to accept the conclusions of the plaintiffs’ expert for the reasons we have above set forth.
The conclusion of the district court that the multiple regression evidence was flawed is supported by authority. The purpose of multiple regression is to account for changes in a single variable through the examination of other variables which might be expected to influence that variable. In the present instance, that single variable is salary. An appropriate regression analysis of salary should therefore include all measurable variables thought to have an effect on salary level. Barnes, Statistics, as Proof (1983), p. 339; Fisher, Multiple Regression in Legal Proceedings, 80 Colum.L.Rev. 702, 714 (1980). However, both experts omitted from their respective analysis variables which ought to be reasonably viewed as determinants of salary. As a result, the regression analy-ses presented here must be considered unacceptable as evidence of discrimination.
The merit pay increases are another matter. They are granted to agents in the first three quartiles but not in the fourth, so that if there were discrimination and black employees were assigned to a lower *673quartile, especially the fourth quartile, because of their race, there would be a salary ... " disparity resulting on account of race.
On appeal, appellants rely upon quartile rankings by district and race for only one year, 1981, and we are told in the plaintiffs’ brief, p. 40, n. 40, that for the year 1975 the quartile scores of black employees were relatively better than those of white employees. The 1981 rankings appear in the table just below:
QUARTILE RANKINGS BY DISTRICT AND RACE
DISTRICT QUARTILE
I II III IV
W B w B w B w B
NC 17 5 16 7 14 12 12 9
NE 12 2 13 3 10 5 8 3
NW 19 1 21 1 17 3 9 4
SE 22 1 18 9 14 6 10 7
SW 20 3 21 2 18 2 12 4
w 16 — 16 — 15 — 13 __
TOTAL 106 12 105 22 88 28 64 27
Lumping together the first three quartiles, for employees in those quartiles are the ones who get the merit pay increases, a standard deviation analysis, with the expected number of black employees in the first three quartiles as 75%, indicates no statistically significant variation in the assignment of quartile rankings. Under the rule formulated in Hazelwood, “more than two or three standard deviations” would be required to undercut the presumption that employment decisions were being made without respect to race. Hazelwood, 433 U.S. p. 311, n. 17, 97 S.Ct. p. 2743, n. 17. The following table shows the results here to be well below this threshold.8
Total by District (Expected occurrence in first three quar-
tries ^75%)
Actual Ocurrence of Blacks
District in First 3 Quartiles
NC 24/33 = .7273 Std. Dev. = 0.3012
NE 10/13 = .7692 Std. Dev. =(-)0.1599
NW 5/9 = .5556 Std. Dev. = 1.3469
SE 16/23 = .6957 Std. Dev. = 0.6014 SW 7/11 = .6364 Std. Dev. = 0.8701
W No Blacks in 1981
Even if the expected occurrence in each district is instead the actual percentage of all employees in the district, black and white, which occurred in the first three quartiles in 1981, only one district could have any statistically significant variation under the Hazelwood analysis, as is shown in the table below. And in that one district (NW) a student’s adjustment shows the variation not to be of statistical significance under the Hazelwood rule.9
Total by District (Actual occurrences in each district used as norm)
NC Expected Occurrence = 71/92 = .7717
Actual Ratio Blacks = 24/33 = .7273
Std. Dev. from Expected Ratio = .6077
NE Expected Occurrence = 45/56 = .8036
Actual Ratio Blacks = 10/13 = .7692
Std. Dev. from Expected Ratio = .3122
NW Expected Occurrence = 62/75 = .8267
Actual Ratio Blacks = 5/9 = .5556
Std. Dev. from Expected Ratio = 2.1487
Student’s t std. dev. for sample size of 9 at 95% confidence level = 2.306
SE Expected Occurrence = 70/87 = .8046
Actual Ratio Blacks = 16/23 = .6957
Std. Dev. from Expected Ratio = 1.3172
SW Expected Occurrence = 66/82 = .8049
Actual Ratio Blacks = 7/11 = .6364
Std. Dev. from Expected Ratio = 1.4103
W No Blacks in 1981.
We are unable to say that these statistics support a finding of diserimina*674tion in the assignment of quartile rankings.10
The upshot of the matter is that there are only two ways in which the salaries of the county agents may reflect a racially discriminatory effect. The first is because of the lingering effect of pre-Act discrimination, and the second is because of the assignment of quartile rankings, the only aspect of salary computation in which the Extension Service exercised any discretion. The first is not separately actionable, as we have previously set forth, and the second we do not think is supported by the record. Thus, the judgment of the district court with respect to salaries was without error.
E. County Chairmen
The last subject we will dwell upon in any detail is that of selection of County Chairmen. The plaintiffs claim that this selection process has been flawed by racial discrimination, and the government as well as the named plaintiffs claim that the selection process has been flawed by a pattern or practice of discrimination.
Although there may be some repetition, some salient facts are worth repeating. The top administrative position in the Extension Service at the county level is that of County Chairman. Those holding that position have administrative and leadership responsibilities in addition to normally carrying a subject matter program work load depending on the needs of the county. The position is not within the promotional line from County Agent; it must be applied for separately, and the method of selection is neither akin to the hiring procedure for county agents nor to the process of their promotion. It is open to any qualified applicant. The minimum qualifications are six years (formerly seven) of service with the Extension Service or the equivalent, at least two years of which must have been in the service of the North Carolina Extension Service. A master’s degree is desirable but is not required, but a bachelor’s degree is required. While applicants are not required to be in the employ of the Extension Service, it is obvious that the requirement of two years’ employment in that service in *675North Carolina will have the effect of causing the vast majority of the applications to be from employees of the Extension Service, which is what happened here.
When an impending vacancy becomes known, the District Chairman meets with the County Commissioners and reaches a decision as to the desired qualifications of the prospective applicants with respect to subject matter, knowledge, and competency. This is communicated to administrative officials in the Extension Service, and an announcement is made of the vacancy, not only to all the county agents but also to the various educational institutions from which the Extension Service recruits personnel. Each applicant meeting the minimum qualifications is interviewed by the district chairman, the assistant director for county operations, the associate director, and the director of the Extension Service. These people then meet and come to a decision as to whom they will recommend to the County Commissioners of the county to fill the vacancy. Sometimes only one applicant is recommended; sometimes more than one is recommended. The County Commissioners then must approve one of the recommended names, for the agreement of the Extension Service and the County Commissioners is required in order to fill the vacancy.
At the outset, we should say that the individual plaintiffs who testified in support of their claims that they had been discriminatorily denied selection to County Chairmen had their claims considered by the district court on evidence particularly applicable to each case. The district court found against each of them. These particular findings of fact and conclusions of law have not even been appealed except in the context of a refusal to find a pattern or practice of discrimination. So, there are no cases of actual discrimination to bolster a statistical case. The district court properly placed the burden of proof on the plaintiffs to prove their individual claims. Of course if the pattern and practice claims had been successful, then these plaintiffs, being in the class discriminated against, would have had the benefit of a presumption that they had been victims of the class discrimination, and the burden of proof would have been shifted to the defendants to prove that such claimants should not recover. Teamsters v. United States, 431 U.S. 324, 359, 97 S.Ct. 1843, 1866, 52 L.Ed.2d 396 (1977), construing Franks v. Bowman Transportation Co., 424 U.S. 747, 96 S.Ct. 1251, 47 L.Ed.2d 444 (1976).
We will next proceed to an analysis of the statistical information which is relied upon to prove the pattern or practice claims. We must keep in mind that it is the Extension Service whose actions are complained of here, as distinguished from the action of the counties, so it is the “employment decisions,” Hazelwood, 433 U.S. p. 309, 97 S.Ct. p. 2742, of the Extension Service which must be examined. In the context of this case, the employment decisions made by the Extension Service with respect to the selection of County Chairmen were made when the Service either recommended or did not recommend an applicant for an existing vacancy to the County Commissioners. Thus, it is the act of recommending or not recommending which we examine. The district court found that Title YII was made applicable to the Extension Service in 1972 and that claims under 42 U.S.C. §§ 1981 and 1983 arising prior to November 18, 1968 would be barred by the statute of limitations. In this narrow aspect of the case, the bar of limitations prior to November 18, 1968, the plaintiffs, defendants, and the district court are in agreement. The district court found there had been no intentional discrimination proved which was necessary to maintain the case under §§ 1981 and 1983, and analyzed the statistical evidence from 1972 on.
The appellants claim that, for 1962 through 1981, of 227 appointments to County Chairmen only 6 were black, and that, although blacks account for one-fourth of the full agents and one-fifth of the professional work force, they have received only 2.6% of the promotions to County Chairmen during that time. Although the claim is *676not specifically articulated, of course not more than 6% of the existing County Chairmen were black. Thus, they say these statistics only support a finding of discriminatory selection and thus a finding of a pattern or practice of discrimination.
The statistics as thus presented, however, do not tell the full story. There was no position of County Chairman prior to 1962-63, when it was created. A County Chairman was appointed in 1962-63 for each of the white county units, the Service then being segregated as previously related. Black county units did not have county chairmen but worked through the District Extension Chairmen. In 1964 one more, and in 1965, 6 more whites were appointed County Chairmen. In 1965, the black and white units of the Extension Service were merged and the then existing white chairmen became the County Chairmen of the county units of the integrated Extension Service. There were no appointments in 1966, and in 1967, 5 more white chairmen were appointed. Between 1968 and 1970, 31 more whites were appointed, and, in 1971, 3 more whites and one black were appointed. In March 1972, the Extension Service was placed under Title VII, and since that time the balance of the appointments to County Chairmen have been made. For the purpose of analysis, we assume that the naked figures presented above might support an inference of discrimination in the selection of County Chairmen, but, as Hazelwood, p. 309, 97 S.Ct. p. 2742, directs, an employer must be given an opportunity to prove that each of his post-Act employment decisions was made in a non-discriminatory way. In this respect, the gross statistics we have related above show that at least 104 of the appointments in question (1962-64) were made years prior to the time Title VII was effective and well prior to the time of the Civil Rights Act of 1964 which became effective in 1965. The six 1965 and five 1967 appointments are not actionable since the statute of limitations ran on all appointments before 1968, as the parties agree. Indeed, our examination of the record reveals little or no attempt to present precise data on the appointments prior to 1968. The data for the years from 1968 until the date of trial appears to be fairly complete. The holding of the district court that there was no intentional discrimination shown for the appointments prior to 1972 is not the subject of serious objection on appeal, but we have analyzed the data from 1968 forward so as not to rely upon that finding alone. In this pattern or practice case, we have not excluded post-Act 1965 data and 1967 data from our analyses because barred by limitations; it is just that we do not find it in the record.
The district court found that there had been 77 vacancies since the institution of statewide announcements of vacancies in 1972, and, of that number, blacks had not applied for 59 but had applied for 18 of the vacancies. It found that 18 individual blacks did apply for the 18 vacancies, that 5 of them were selected as County Chairmen by both the Extension Service and the county involved and that their rate of selection was 28% of the applicants. For the same 18 vacancies, the court found that there were 37 white applicants, of which 13 were selected as County Chairmen, or a selection rate of 35%. It held that since the black rate of selection was 80% of the white that it complied with the EEOC guideline in 29 C.F.R. § 1604D which provides that a selection rate of a minority group of more than 4/5ths of the majority rate will generally not be regarded by federal enforcement agencies as evidence of adverse impact.
Appellants complain of the method of analysis of the district court. They say that the district court considered black applicants, not applications, so that those applicants who applied for more than one vacancy are only counted as one instead of two or more, thus increasing the percentage rate of black appointments. They also say that considering the vacancies for which blacks only applied but excluding the vacancies for which whites only applied further flaws the analysis by treating as irrelevant the vacancies for which whites only applied, but finding evidence of non*677discrimination by counting the vacancies for which blacks only applied. They say that the data should be analyzed only for positions for which both blacks and whites applied.
While there may be some justification for the method of analysis by the district court, if we should consider that discrimination is against people and not against applications, we think that, without deciding that either method is correct, a comparative analysis is worthwhile. We believe that a correct analysis of the data at this point in one respect, however, should not be either as the district court analyzed it or as appellants would have us do. The Extension Service is not separately responsible for the appointment to fill the vacancy, but it is separately responsible for the recommendation of one or more of the applicants. Thus, it is the action of the Extension Service upon each application that is its employment decision, and those are the employment decisions that we examine. The results of our examination are set out in the table below:
RECOMMENDATION RECORD OF THE NCAES FOR POSITIONS FOR WHICH BLACKS APPLIED, 1972-1981
Applications Recommendations
Year County B W B w
1972 No black applications
1973 Lenoir 2 7 1 3
1973 Iredell 1 3 0 1
1974 Northhampton 1 2 1 2
1975 Columbus 1 4 0 1
1975 Martin 1 3 0 1
1975 Wayne 1 4 0 1
1975 Iredell 2 2 0 1
1976 Johnston 2 6 1 2
1976 Union 1 1 0 1
1976 Cumberland 1 0 1 0
1977 Granville 1 2 1 1
1978 Pitt 1 . 0 1 0
1979 Warren 2 1 1 0
1980 Nash 1 2 0 1
1980 Anson 1 0 1 0
1980 Hertford 1 2 0 1
1981 Vance 1 2 0 1
1981 Hoke I 1 1 0
TOTALS 23 44 9 18
Results for all vacancies for which blacks applied:
Black Recommendation Rate 9/23 = 39.1%
White Recommendation Rate 18/44 = 40.9%
Black Rate as Percentage of White Rate = 95.6%
Removing positions for which only blacks applied:
Black Recommendation Rate 6/20 = 30%
White Recommendation Rate 18/44 = 40.9%
Black Rate as Percentage of White Rate = 73.3%
It is apparent from the table, if we consider all of the vacancies for which blacks applied, the Extension Service recommended essentially the same percentage of black applicants as it did white applicants, for a difference of one recommendation would have made the black rate 112.5% of the white rather than 95.6% as shown. But, in accordance with appellants’ contentions, if we remove the three vacancies for which blacks only applied and exclude also the vacancies for which whites only applied, the black recommendation rate of 30% of the applicants as opposed to the white recommendation rate of 40.8% shows a black rate of 73.3% of the white. While not within the EEOC guideline, a difference of one recommendation would have made the black rate 90.7% of the white because of the small numbers involved, again a figure well within the guideline.
As the table above indicates, we have used 19 vacancies which we find rather than 18 as used by the district court and the appellants. Using the district court’s method of calculation for those vacancies, but using applications instead of applicants as appellants suggest, we find that blacks received 6 of the 19 appointments for which they applied. Six of the 23 black applications were ultimately successful, or 26.3%, while 13 of the 44 white applications were, or 29.6%. Eighty percent of the white rate would be 23.7%, and the black rate of appointment (26.3%) as compared to the white was 89%, well within the EEOC guideline.
With respect to the parties and the district court, however, we think that the correct method of analysis of this important data has not been advanced. We do not *678think that the applicant flow data should be analyzed as appellants wish us to do, using only vacancies for which both blacks and whites applied. Appellants’ suggested method, instead of determining whether there was sufficient disparity in the approval rates of members of the class for a sufficient period of time from which discrimination might be inferred against the class, would narrow it down in each instance in the usual case to require an analysis of all the pending applications on the day the particular appointment was made. This would permit either a wide disparity in employment decisions or a wholly evenhanded employment policy to go either unnoticed or unconsidered by the observation of a small number of employment decisions. It also could, and most probably would, require a McDonnell-Douglas11 analysis of each vacancy to ascertain as a preliminary matter whether the data with respect to that vacancy should be included in an applicant flow analysis.
We are of the opinion and hold that, on the data presented here, an analysis of all of the vacancies and the action on all of the applications for all of the vacancies is the proper way to consider the applicant flow data. See Hester v. Southern Ry., 497 F.2d 1374 (5th Cir.1974). This data is set out in the next table which appears below. We have used Government Exhibit 69 as the starting place for this table of vacancies and the action of the Extension Service with respect to them. Where no information is shown on that exhibit for other applications, we have assumed the successful applicant was the only one unless others as noted were found in those parts of the record we were examining. Where we could find no other recommendations for the vacancy, we have assumed the successful applicant was the only one. All of the vacancies in our table except three appeared on Government Exhibit 69.
COUNTY YEAR APPLICATIONS RACE RECOMMENDATIONS
Hoke12 1981 Willie Featherstone B Willie Featherstone
Ellen Willis W
Chowan 13 1981 Mike Williams W Mike Williams
Zackie Harrell W
Henry Riddick w Henry Riddick
Rutherford14 1981 Steven West w
Dewey Hennessee w
Eugenia Ware w Eugenia Ware
Pamlico 15 1981 Fred May w Fred May
R. Ray Harris w
Rockingham16 1981 Frank Green w Frank Green
Avery17 1980 J. M. Pittman w J. M. Pittman
Johnnie Hensley w
18 & o 1980 John Reeves w John Reeves
Cabarrus 19 1980 Alvin M. Stanford w Alvin M. Stanford
William Triplett w
McDowell20 1980 James R. Mabe w James R. Mabe
*679COUNTY YEAR APPLICATIONS RACE RECOMMENDATIONS
Vance 21 1980 Donald Cobb W Donald Cobb
Phyllis Stainback W
Fred Belfield B
Ashe 22 1980 James Carey W James Carey
Robert F. Breland W
Dana Tugman w
Hertford23 1980 John W. Dunham w John W. Dunham
James Daughtry w
James E. Wright B
Perquimans24 1980 W. R. Jester w W. R. Jester
(Juster, Jessup?)
Ken Bateman w Ken Bateman
Henry C. Riddick w
Anson 25 1980 Hoover Royals B Hoover Royals
Nash26 1980 James Stephenson W James Stephenson
Fred Belfield B
Bryan Page W
Buncombe 27 1980 George Biddix w George Biddix
J. B. Reeves w
Reagon Ammons w
Montgomery 28 19?? R. C. Reece w R. C. Reece
Surry29 19?? John Waddell w John Waddell
Stokes 30 19?? Susan Hilton w Susan Hilton
Cleveland 31 1979 B. F. Spencer w B. F. Spencer
Robeson 32 1979 John Richardson w John Richardson
M. K. Dennis w
Lee 33 1979 John Hall w John Hall
Fay T. Donnell w
Clarence J. Cameron w
Jones 34 1979 M. C. Small w M. C. Small
Charles Hammond w
Fletcher Barber B
Northhampton35 1979 Bill Rogister W Bill Rogister
*680COUNTY YEAR APPLICATIONS RACE RECOMMENDATIONS
Vance 36 1979 James Stephenson W James Stephenson
Phyllis Stainback W
Marshall Bowden W
New Hanover37 1979 C. E. Lewis W C. E. Lewis
Thomas Dail W
Harvey Morris W
James E. Warren W
Caswell38 1979 L. N. White W L. N. White
Donna Pointer W
Pamlico 39 1979 Reginald Piland W Reginald Piland
Warren 40 1979 L. C. Cooper B L. C. Cooper
George Koonce B
Emily Ballinger W
Onslow 41 1979 Dan Baucom W Dan Baucom
Charles Hammond w
Chase Padgett w
Bladen42 1979 M. K. Dennis w M. K. Dennis
Harvey Morris w
Moore 43 1979 C. E. Hammond w C. E. Hammond
Tom Colson w
Paul Seabolt w
Clarence Cameron w
Davie 44 1979 W. E. Mainous w W. E. Mainous
Judge A. Pierce w
Ronnie Thompson w
Yancey 45 1979 James Ray w James Ray
Carlos Bickford w
Johnny G. Hensley w
Forsyth 46 1979 J. D. Carroll w J. D. Carroll
W. E. Mainous w X
Worth Gurkin w Worth Gurkin
Al Stanford(?) w
Randolph 47 1979 Talmadge Baker w Talmadge Baker
Dick Peterson w Drue Finette w
Richard Freeman w
Doug Young w
Rodney Haines w Rodney Haines
Wayne 48 1978 Wesley Townsend w Wesley Townsend
John H. Wynne w
William Lamm w
*681COUNTY YEAR APPLICATIONS RACE RECOMMENDATIONS
Rowan 49 1978 James H. Caudill W James H. Caudill
Rodney Haines W
Duplin50 1978 Lois Britt w Lois Britt
Robert Swain w
S. B. Wilson w
Pitt51 1978 Leroy James B Leroy James
Polk 52 1977 0. R. Ammons w O. R. Ammons
Carlos P. Bickford w
Judy M. Groff w
Bertie 53 1977 W. J. Griffin w W. J. Griffin
Mitchell54 1977 David Terrell w David Terrell
Alleghany 55 1977 Helen Dosier w Helen Dosier
William Fowler w
Walden M. Hearn w
Jim Weaver w Jim Weaver
Granville 56 1977 Frank Baker w Frank Baker
I. W. Murfree B I. W. Murfree
Dorothy Wilkinson w
Watauga 57 1977 Gene Brewer w Gene Brewer
Bill Fowler w Bill Fowler .
Wilson 58 1977 C. H. Jernigan w C. H. Jernigan ■
Frank Baker w
Greene 59 1977 R. A. Hayes w R. A. Hayes
Cumberland 60 1976 B. T. McNeil B B. T. McNeil
Lincoln 61 1976 David Choate w David Choate
A. L. Smith w
Howard Waynick w
Wake 62 1976 Y. B. Lynn w V. B. Lynn
Union 63 1976 M. C. Howell w M. C. Howell
P. E. Bazemore B
Brunswick 64 1976 R. M. Coleman W R. M. Coleman
Dan E. Baucom w
M. K. Dennis w
*682COUNTY YEAR APPLICATIONS RACE RECOMMENDATIONS
Swain 65 1976 C. D. Bunn W C. D. Bunn
Roger Hyatt W
Transylvania 66 1976 Robert R. Hyatt W Robert R. Hyatt
Roger Hyatt W
James H. Ray W
Dennis Winters W
Linda Best W
Cabarrus 67 1976 E. E. Bishop W E. E. Bishop
J. P. Bowles W
Johnston 68 1976 Bruce Woodard W Bruce Woodard
David Stanaland W
Frank Baker W Frank Baker
Harold Lloyd W
Peter Westerbeck W
H. F. Palmer B
James Stephenson W
Leroy James B Leroy James
Alamance 69 1975 Millis B. Wright W Millis B. Wright
Richard J. Freeman W
Buncombe 70 1975 D. R. Burnette W D. R. Burnette
C. P. Bickford W X
J. D. Brooks W
Nancy Stancil W
Robert Hyatt W
Stanly 71 1975 J. F. Simpson W J. F. Simpson
Carroll Baker W Carroll Baker
Ray Kiser W
Frederick J. Rivers W
Pender72 1975 W. F. Walker W W. F. Walker
Dan E. Baucom W
J. G. Richardson W
John Wynne W
Iredell73 1975 Kenneth Vaughn W Kenneth Vaughn
D. O. Ivey B
S. D. Little B
Jack Smith, Jr. W
Beaufort74 1975 M. C. Griffin W M. C. Griffin
H. C. Riddick W
Jack Cullipher W
R. M. Pilch W
W. M. Hearn W
Gordon Sawyer W
Steve Warrick W
Gaylon Ambrose W
Harvey Morris W
*683COUNTY YEAR APPLICATIONS RACE RECOMMENDATIONS
Person 75 1975 F. J. Rivers W F. J. Rivers
M. B. Wright W
Tyrrell76 1975 F. S. Voliva W F. S. Voliva
Wayne 77 1975 T. S. Godwin W T. S. Godwin
Frank Baker W
William Lamm w
Wesley Townsend w
B. T. McNeil B
Martin78 1975 L. L. Allen w L. L. Allen
R. M. K. Edwards B
C. J. Cameron W
P. C. Bryant w
Columbus 79 1975 G. D. McCullen w G. D. McCullen
M. K. Dennis w
Harold Lloyd w
Harvey Morris w
John Spaulding B
Pasquotank 80 1974 Donald Baker w Donald Baker
W. M. Hearn, Sr. w
J. R. Vaughn w
Harvey Morris w
Victor Lynn w
Burke 81 1974 H. L. Miller w H. L. Miller
R. M. Bowden w
Roger Hyatt w
Robert Hyatt w
Latham Smith w Latham Smith
Alexander 82 1974 B. G. Westbrook w B. G. Westbrook
Roger Hyatt w
Kenneth Patterson w
Caswell83 1974 M. C. Small w M. C. Small
Northhampton 84 1974 E. J. Long w E. J. Long
Fred Belfield B Fred Belfield
Doug Eason w Doug Eason
Wilkes 85 1974 H. M. Ramseur w H. M. Ramseur
M. B. Wright w
Watauga86 1974 W. C. Richardson w W. C. Richardson
Gene R. Brewer w
Rodney Haines w
Agnes A. Greene w
Henry B. Hagwood w
Harold Lloyd w
*684COUNTY YEAR APPLICATIONS RACE RECOMMENDATIONS
C. D. Bunn W
James W. Gentry W
Patrick Guyer W
Iredell87 1973 Alvin Stanford W Alvin Stanford
D. 0. Ivey B
H. M. Stoney W
H. W. Myers W
Haywood 88 1973 Herman McCall w Herman McCall
H. Leslier Miller w
Lenoir 89 1973 W. S. Lamm w W. S. Lamm
John H. Wynne w
W. N. Payton B
G. E. McDaniel B G. E. McDaniel
R. L. Stroud w
Walter C. Johnson w Walter C. Johnson
T. S. Godwin w
C. H. Jernigan w C. H. Jernigan
F. H. Baker w
Montgomery 90 1973 Walter Bowers w Walter Bowers
Stokes 91 1973 Jack Barnes w Jack Barnes
M. B. Wright w
C. J. Cameron w
Catawba 92 1972 Edwin Nolley w Edwin Nolley
Davidson 93 1972 W. C. Holtzman w W. C. Holtzman
Wade Johnson w
Latham Smith w
Tyrrell94 1971 Fred Rivers w Fred Rivers
Nash 95 1971 William Shackelford w William Shackelford
Durham 96 1971 Carl Hodges B Carl Hodges
Harnett97 1971 James Goff w James Goff
Connie Jernigan w Connie Jernigan
Gates 98 1970 Zackie Harrell w Zackie Harrell
C. M. Grimes B
Mr. Baker w Baker
Mecklenburg 99 1970 Phil Haas w Phil Haas
Max Irving w
Bill Bowers w
Wallace Flynt w
*685COUNTY YEAR APPLICATIONS RACE RECOMMENDATIONS
Edgecombe 100 1970 Joe Perry W Joe Perry
Madison101 1970 Earl Wise W Earl Wise
Macon102 1970 Robert Rollins W Robert Rollins
George Conrad
Earl Wise W
Greene103 1970 Walter Johnson W Walter Johnson
Daniel Godfrey B Daniel Godfrey
Frank Baker W Frank Baker
w X
Moore104 1970 Talmadge Baker w Talmadge Baker
Granville105 1970 Aubrey Hardee w Aubrey Hardee
Perquimans 106 1970 Richard Bryant w Richard Bryant
Transylvania107 1970 Jerry Purser w Jerry Purser
Yancey108 1970 William Bledsoe w William Bledsoe
Henderson109 1970 Grover Westmoreland w Grover Westmoreland
Avery110 1970 W. W. Avery w W. W. Avery
Leslie Miller
Woodie Richardson
Montgomery111 1970 Robert Wesson w Robert Wesson
Chowan112 1970 Raymond Thompson w Raymond Thompson
Craven113 1970 Ed Simpson w Ed Simpson
Yadkin114 1969 Samuel Young w Samuel Young
Pitt115 1969 Edwin Yancey w Edwin Yancey
Camden116 1969 Gordon Sawyer w Gordon Sawyer
Hertford117 1969 Enos Register w Enos Register
Alleghany118 1969 William Corbin w William Corbin
Alleghany119 1969 Roger Murdock w Roger Murdock
*686COUNTY YEAR APPLICATIONS RACE RECOMMENDATIONS
Orange120 1969 Ebert Pierce W Ebert Pierce
Prank Emory B Prank Emory(?)
S. N. Shelton B S. N. Shelton
Bruce Woodard W
Union121 1969 John Stacy w John Stacy
Cumberland122 1969 Bruce Woodard w Bruce Woodard
B. T. McNeil B B. T. McNeil
Dalton Proctor w Dalton Proctor
Alexander123 1968 Henry Ramseur w Henry Ramseur
George McDaniel B George McDaniel
X w X
Y Y
Carteret124 1968 James Bunce w James Bunce
Warren125 1968 Laverne Hardage w Laverne Hardage
L. C. Cooper B L. C. Cooper
Guilford126 1968 John Crawford w John Crawford
Tyrrell127 1968 Murray Goodwin w Murray Goodwin
The above data may be summarized as follows, leaving out of our computations the 3 employees whose race is not shown. Background data for the 11 vacancies occurring in 1965 and 1967 is sparse in the record, and we have, not included them for that reason. Although the uniform notice of vacancies was not commenced until November 1972, Title VII became applicable to the Extension Service in March of that year, so we have included 1972 with subsequent years for the purpose of grouping data in the table and analysis.
Applications Recommendations
B W ltd 13
1968-71 8 45 7 40
1972-81 23 213 9 91
1968-81 31 258 16 131
An analysis of the above data for the years 1968-71 shows 8 black applications with 7 approvals and 45 white applications with 40 approvals, a black approval rate of 87.5% and a white approval rate of 88.9%. The black approval rate was 98.4% of the white. For 1972-81 there were 23 black applications with 9 approvals and 213 white applications with 91 approvals, a black approval rate of 39.1% and a white approval rate of 42.7%. The black approval rate was 91.6% of the white. For the entire period 1968-81 there were 31 black applications with 16 approvals and 258 white applications with 131 approvals, a black approval rate of 51.6% and a white approval rate of 50.8%. The black approval rate was 101.6% of the white. Thus, considering all of the data which we have for all of the applica*687tions shows that for the entire period 1968-81 black applications were approved at a slightly higher rate than were the white applications, and in neither the 1968-71 period nor the 1972-81 period, breaking down the data, was the black approval rate lower than 91.6% of the white. Certainly all of this data for either of the two shorter periods, or for the entire period under consideration, is well within the EEOC guideline.
Hazelwood, 433 U.S. p. 308, n. 13, 97 S.Ct. p. 2742, n. 13, describes applicant flow data as “very relevant,” and New York Transit Authority v. Beazer, 440 U.S. 568, 99 S.Ct. 1355, 59 L.Ed.2d 587 (1979), at least expresses a preference for, and perhaps. requires, applicant flow data in preference to more general data. Applicant flow data has been described as the most direct route to proof of racial discrimination in hiring in Hester v. Southern Railway Co., 497 F.2d 1374 (5th Cir.1974), and in United States v. County of Fairfax, Va., 629 F.2d 932, 940 (4th Cir.1980), we reasoned that “applicant flow data are normally highly relevant.” In this case, we think the probative value of the applicant flow data far outweighs the evidentiary value of naked statistics consisting only of numbers of vacancies and the successful applicant by race. Other than the tables we have set forth above and our earlier discussion of vacancies and appointments, there is no other statistical evidence in the record. There is also no finding of any case of discrimination against an individual who sought the position of County Chairman.
When the complete applicant flow data is analyzed, far from showing any discrimination in the appointments to County Chairmen, the only conclusion it will permit is that the Extension Service made its employment decisions with respect to County Chairmen in as evenhanded a way as could be imagined or sought for within reason. The essence of the analysis of all the applications is that black applications for the position were approved at about the same rate as were white applications. If we assume that a prima facie case was made by the naked statistics consisting of vacancies and successful applicants from 1968 through 1981, it has been completely refuted by consideration of all of the applications, and even if the burden was upon the Extension Service to prove that its employment decisions were free from racial discrimination, it has done so.
We have considered the remaining assignments of error and are of opinion they are without merit.128
The judgment of the district court is accordingly
AFFIRMED.
. The Extension Service is administratively a part of North Carolina State University, whose chancellor and the Assistant Dean/Director of the Extension Service are also appellees. In addition, the President of the University of North Carolina system, William C. Friday, the Board of Governors of UNC, and Alamance, Edgecombe, and Mecklenburg Counties are ap-pellees.
. P.L. 92-261 modified the definition of “person" in Title VII to include governments and governmental agencies as of March 24, 1972.
. The United States notes that the district court, based on the findings of fact submitted by the defendants, found that some counties designate which agent is to receive a merit increase out of its funds. The United States argues this statement is not supported by the evidence, and while the fact that a county designating which agent is to receive an increase is not generally accurate, the fact that counties do give increases to their own agents is amply supported by the record. We have, however, come across one incident in which a county raised its County Chairman’s pay in order to keep him from moving elsewhere, and another where a county increased the pay of an agent who was acting county chairman pending selection of a permanent replacement.
. For example, County Chairmen salaries in 1971 ranged from $11,601 (Dare County) to $16,588 (Wake County), and similar disparities persist.
. For merit pay increase, see part D, infra.
. In arguing that the district court erred in finding no discrimination on the merits of the private plaintiffs’ salaries claim, the dissent fails to take account of certain matters undisputed or not appealed by the plaintiffs. The dissent fails to mention that there has been no discrimination in hiring or in promotions, and no discrimination in entrance salaries since 1965. Because no dispute exists with respect to hiring, promotion, and post-1965 entrance salaries, the only discrimination that could exist would be in the quartile ranking system. As we demonstrate, however, the record does not support a finding of discriminatory effect in the assignment of quartile rankings. The dissent, nonetheless, on this subject as elsewhere addresses itself to certain weaknesses in the district court’s opinion rather than the correctness of that court’s judgment. Since we review judgments, not reasons, Eltra Corp. v. Ringer, 579 F.2d 294, 298 (4th Cir.1978), even erroneous reasoning does not require reversal so long as the judgment is correct.
. An inspection of the data bases for the regression analyses shows numerous pre-Act hires, both pre-1965 and pre-1972.
. The formula used to determine the amount of variation between the actual and expected occurrence percentages in terms of standard deviations is as follows:
n(p-a) v" np(l-p)
where a is the proportion of blacks actually in the first three quartiles, p is the proportion of all agents that would be expected to be in the first three quartiles, and n is the total sample size. See Moultrie v. Martin, 690 F.2d 1078, 1083, n. 8 (4th Cir.1982).
. The student’s t adjustment is done to accommodate the small sample size found in this district (nine blacks). The student’s t distribution teaches that when the sample size is less than approximately thirty, the number of standard deviations must be increased in order to achieve the same significance level. See R. Winkler & W. Hays, Statistics: Probability, Inference, and Decision, 366 (2d ed. 1970).
. Indeed, the comparisons made in this statistical analysis constitute a more advantageous analysis for the plaintiffs than might have otherwise been conducted. The present analysis defines the expected occurrence of blacks in the first three quartiles as either 75% or the actual occurrence ratio of an entire group of a district’s agents. A more mathematically correct and appropriate analysis would define the expected occurrence ratio of blacks in the first three quartiles as the proportion of blacks in a district’s population of agents. This ratio would then be compared to the actual proportion of blacks in the first three quartiles. The resulting standard deviations would then provide a more accurate reflection of any difference between the representation of blacks in a district’s population of agents and the representation of blacks in the first three quartiles.
This type of comparison should be considered more precise in mathematical terms. Barnes, Statistics as Proof (1983), p. 228. More importantly, it permits the standard deviation analysis to focus on the essential issue of whether blacks are being represented in the first three quartiles in proportions similar to their representation in the population of agents. It is this comparison between the representation of a minority in a suspect group and its representation in a relevant larger population which was the principal focus in Hazelwood, and has been the preferred comparison in employment discrimination and like litigation. Castaneda v. Partida, 430 U.S. 482, 494-497, 97 S.Ct. 1272, 1280-1281, 51 L.Ed.2d 498 (1977) (racial discrimination in grand jury selection); Teamsters v. United States, 431 U.S. 324, 337-338, 97 S.Ct. 1843, 1855-1856, 52 L.Ed.2d 396 (1977) (racial discrimination in hiring).
In this instance, a comparison of the Hazel-wood standard with the standard deviation analysis used in the body of the opinion establishes that our use there of 75% or actual occurrence ratios as the expected ratio was done to present the plaintiffs’ position in the best possible light. It is not a precedent to be followed: the Hazel-wood standard rather is the rule. The application of a standard deviation analysis to the representation of black agents in the first three quartiles as compared to their representation in a district’s agent population fails to produce any standard deviations greater than one. Under the Hazelwood comparison, these results made the evidence even less convincing that quartile assignments were a function of racial discrimination. The table below summarizes the data:
District Expected Actual Std. Dev.
NC .358(33/92) .338(24/71) .36
NE .232(13/56) .222(10/45) .16
NW .12(9/75) .081(5/62) .95
SE .264(23/87) .229(16/70) .68
sw .134(11/82) .106(7/66) .67
. McDonnell-Douglas v. Greene, 411 U.S. 792, 793, 93 S.Ct. 1817, 1820, 36 L.Ed.2d 668 (1973).
. GX69 (A1736); Tr. 95 (Sasser); GX172.1 at 54-9, 95 (Sasser Dep.)
. GX69 (A1736); A. 792, 915-6; Tr. 3666, 4005-7 (Blalock).
. GX69 (A1736); Tr. 4922 (Black).
. GX69 (A1736).
. GX69 (A1736); Tr. 4927 (Black).
. GX69 (A1736).
. GX69 (A1736).
. GX69 (A1736).
. GX69 (A1736).
. GX69 (A1736); Tr. 3847, 4375 (Blalock), 4687-88 (Black), 5562 (Dew).
. GX69 (A1736); Tr. 92 (Sasser).
. GX69 (A1736); Tr. 3833, 4375 (Blalock), 4715-22, 4877 (Black), 5706 (Patterson), 6488 (Hunter).
. GX69 (A1736); A. 795, Tr. 3669 (Blalock); Tr. 5548, 5598 (Dew); GX169.1 at 49-50 (Patterson Dep.).
. GX69 (A1736); Tr. 6399-6402 (Yancey).
. GX69 (A1736); Tr. 3847 (Blalock), 4682-87 (Black). But see GX166.1 at 115-16 (Liner Dep. —claims Belfield sole other applicant).
. GX69 (A1736); Tr. 4922-23 (Blalock).
. DX67-46 (Montgomery County Folder); Tr. 4923 (Blalock).
. DX67-97 (Surry County Folder); Tr. 4925 (Blalock).
. DX67-96 (Stokes County Folder); Tr. 4926 (Blalock).
. GX69 (A1736).
. GX69 (A1736).
. GX69 (A1736).
. GX69 (A1736); Tr. 1935 (Barber), 3859-60, 4375 (Blalock), 4677-82, 4849 (Black).
. GX69 (A1737).
. GX69 (A1737); Tr. 6393 (Liner).
. GX69 (A1737).
. GX69 (A1737).
. GX69 (A1737).
. GX69 (A1737); A. 918 (Blalock); Tr. 1893-95 (Cooper), 3858, 4010, 4375 (Blalock), 4703-5 (Black).
. GX69 (A1737).
. GX69 (A1737).
. GX69 (A1737).
. GX69 (A1737); GX172.1 at 92 (Sasser Dep.)
. GX69 (A1737).
. GX69 (A1737); A. 796, Tr. 3670 (Blalock); Tr. 4827-8 (Black), 5598 (Dew); GX172.1 at 54-59, 94 (Sasser Dep.).
. GX69 (A1737); A. 793-4, Tr. 3667-8 (Blal-ock);. Tr. 4944 (Black), 5965 (Meldau).
. GX69 (A1737); Tr. 4929 (Black).
. GX69 (A1737).
. GX69 (A1737); A. 745, Tr. 3617-18 (Blalock).
. GX69 (A1737); A. 1004, Tr. 3821, 4375 (Blal-ock); Tr. 2530 (James).
. GX69 (A1737).
. GX69 (A1737).
. GX69 (A1738).
. GX69 (A1738); A. 794, Tr. 3668 (Blalock); Tr. 4928 (Black), 5962 (Meldau); GX167.1 at 13-15 (Meldau Dep.).
. GX69 (A1738); Tr. 1613, 1617 (Murfree); A. 792-4, 1004, Tr. 3666-68, 4375 (Blalock); A. 1069, Tr. 4659 (Black); A. 1356-58, Tr. 5829-31, 5876 (Dobbins).
. GX69 (A1738); A. 793, Tr. 3667 (Blalock); Tr. 5964 (Meldau); GX167.1 at 13-15 (Meldau Dep.).
. GX69 (A1738). But see Tr. 5863 (Dobbins claims Jernigan sole applicant).
. GX69 (A1738).
. GX69 (A1738); A. 1005, Tr. 4376 (Blalock); Tr. 1397-98, 1420 (McNeil).
. RX69 (A1738).
. GX69 (A1738).
. GX69 (A1738); Tr. 4669-76, 4841 (Black), 5605 (Dew).
. GX69 (A1738).
. GX69 (A1738).
. GX69 (A1738).
. GX69 (A1738).
. GX69 (A1738); A. 1005, Tr. 3819-24, 4376 (Blalock); Tr. 94 (Palmer), 2529 (James), 4691-94 (Black), 5880 (Dobbins), 6421 (Lynn).
. GX69 (A1738).
. GX69 (A1738); A. 797, Tr. 3671 (Blalock); Tr. 5547 (Boone).
. GX69 (A1738); A. 797, Tr. 3671 (Blalock); Tr. 5547-8 (Boone), 5597, 5602, 5691 (Dew).
. GX69 (A1739).
. GX69 (A1739); A. 1005, Tr. 4376 (Blalock).
. GX69 (A1739).
. GX69 (A1739).
. GX69 (A1739); Tr. 4924 (Black).
. GX69 (A1739); A. 1005, Tr. 4376 (Blalock).
. GX69 (A1739); A. 1005, Tr. 3861-2, 4376 (Blalock); Tr. 2779, 2792-97 (Edwards).
. GX69 (A1739); A. 1005, Tr. 4376 (Blalock).
. GX69 (A1739).
. GX69 (A1739); Tr. 5548, 5597 (Dew).
. GX69 (A1739).
. GX69 (A1739).
. GX69 (A1739); A. 1005, Tr. 4376 (Blalock); Tr. 1010, 1014-15 (Belfield), 5639 (Dew), 5866 (Dobbins).
. GX69 (A1739).
. GX69 (A1740).
. GX69 (A1740); A. 1005, Tr. 4376 (Blalock).
. GX69 (A1740).
. GX69 (A1740); Tr. 1355 (Hyatt), 1821-22 (McDaniel), 4654-58, 4816 (Black). • But see GX152.1 at 55-58 (Andrews Dep.) (claims W.N. Payton, a black, was also recommended; our table accepts the district court’s finding that only McDaniel was recommended).
. GX69 (A1740); Tr. 5550 (Dew).
. GX69 (A1740).
. GX69 (A1740).
. GX69 (A1740); GX150.1 at 32-5 (Dew Dep.).
. GX69 (A1740).
. GX69 (A1740).
. GX69 (A1740); A. 1006, Tr. 4377 (Blalock); Tr. 1324-25 (Hyatt).
. GX69 (A1740); A. 795, Tr. 3669 (Blalock).
. GX69 (A1740); A. 1006, Tr. 4377 (Blalock); Tr. 983-85 (Grimes), 1310, 1316-17 (Hyatt); GX148.1 at 46-8, 85, addendum item 23 (Jones Dep.).
. GX69 (A1740); GX 162.1 at 12-13 (Haas Dep.).
. GX69 (A1740).
. GX69 (A1741).
. GX69 (A1741).
. GX69 (A1741); Tr. 2248-49, 2252, 2264-66 (Godfrey); GX151.1 at 30-32 (Dobbins Dep.).
. GX69 (A1741).
. GX69 (A1741); Tr. 1736-37 (Lloyd).
. GX69 (A1741); Tr. 828-9 (Strowd).
. GX69 (A1741).
. GX69 (A1741).
. GX69 (A1741).
. GX69 (A1741).
. GX69 (A1741). P. E. Bazemore was informed he would be recommended to the county but withdrew due to salary considerations before any action was taken. Tr. 3825-26 (Blal-ock).
. GX69 (A1741). See Also Tr. 827 (Strowd) (Thompson sent to Chowan county unwillingly — more like a transfer than open recruitment).
. GX69 (A1741).
. GX69 (A1741); GX150.1 at 32-35 (Dew Dep.).
. GX69 (A1741).
. GX69 (A1741).
. GX69 (A1741).
. GX69 (A1741); GX150.1 at 32-35 (Dew Dep.).
. GX69 (A1741); GX150.1 at 32-35 (Dew Dep.).
. GX69 (A1742); A. 795, Tr. 3669 (Blalock); Tr. 1309-10, 1318 (Hyatt), 1417 (McNeil), 6342 (Woodard); GX152.1 at 55-58 (Andrews Dep.).
. GX69 (A1742).
. GX69 (A1742); Tr. 5837-38 (Dobbins); GX151.1 at 33-34 (Dobbins Dep.).
. GX69 (A1742); Tr. 1297 (Hyatt), 1828, 1819-20, 1838 (McDaniel).
. GX69 (A1742).
. GX69 (A1742); Tr. 1309-10, 1322 (Hyatt), 1887-90 (Cooper), 3855-56 (Blalock), 6342 (Woodard); GX152.1 at 55-58 (Andrews Dep.).
. GX69 (A1742).
. GX69 (A1742).
. The dissent takes issue with the failure of the Extension Service to affirmatively integrate the entirely voluntary membership of local 4-H and Extension Homemaker Clubs in racially mixed communities. The balance of all activities are completely integrated. Absent proof of alleged racial discrimination, the mere existence of all white and all black 4-H and Extension Homemaker Clubs in some racially mixed communities violates neither Title VI nor the equal protection clause of the fourteenth amendment. The record is totally devoid of any proof of discrimination with respect to services provided by any 4-H or Extension Homemaker Club. Moreover, the record shows no proof of any discrimination with respect to membership in any Extension Homemaker Club. While the record indicates one incident in which a voluntary worker told a black child he was not welcome in a white 4-H club, the Extension Service at once removed the voluntary worker and admitted the black child to membership in the 4-H club. Especially since the plaintiffs failed to prove their allegations of discrimination, the district court correctly denied the plaintiffs’ claim with respect to the alleged affirmative duty to require integrated membership in local Extension Homemaker and 4-H Clubs in racially mixed communities.
Even assuming that the record contained evidence of discrimination, it would yet be doubtful if plaintiffs are entitled to the relief sought because no evidence in the record exists to show that any plaintiff was denied membership in, or services of, any Extension Homemaker Club or 4-H Club. Consequently, if any party has a complaint it would seem to be the United States, which, significantly, chose not to appeal this issue. We do not decide that issue in this case, however, because the total absence of proof is fatal to plaintiffs' claim.