Plaintiffs-appellants, a class of skilled and unskilled blacks in the Toledo, Ohio metropolitan area seeking employment as carpenters, appeal the District Court’s dismissal of their Title VII employment discrimination claim, 42 U.S.C. § 2000e et seq. Defendants-appellees are four local carpentry unions,1 the union international,2 the collective bargaining agent for the locals,3 and a joint contractor-union apprenticeship council.4 Originally appellants’ complaint issued against a group of local contractors; prior to trial the court below certified a defendant class which included all local contractors. However, that certification was dissolved and the defendant-contractors were dismissed after appellants signed a court-approved consent decree with one representative group of contractors.
The case proceeded to trial against the joint apprenticeship council (TCJAC) and *166the carpentry union, both the Toledo locals, and the international office. At the conclusion of the trial, the District Court issued Findings of Fact and Conclusions of Law pursuant to Fed.R.Civ.P. 52(a). Although the court held that the complaint should be dismissed for failure to join indispensable parties under Fed.R.Civ.P. 19, it went on, despite this procedural defect, to decide the substantive issues before it.
We find error in the court’s resolution of the Rule 19 question and remand the entire case. Furthermore, in order to facilitate the resolution of this lengthy litigation, we will undertake to correct certain legal errors apparent in the court’s substantive analysis of appellants’ claim. First, however, we will address the procedural question.
I.
The District Court’s dismissal under Rule 19 was error. That rule is not to be applied in a rigid manner but should instead be governed by the practicalities of the individual case. Provident Tradesmen’s Bank & Trust Co. v. Patterson, 390 U.S. 102, 116 n.12, 88 S.Ct. 733, 741 n.12, 19 L.Ed.2d 936 (1968). In this particular case, the question of proper parties was muddled by the distinction between the contractors, on the one hand, who controlled the hiring process, and TCJAC, the locals and the international, on the other, who set the entrance requirements to the union and its training programs.
Ideally, all four parties would be before the court. Yet Rule 19 calls for a pragmatic approach; simply because some forms of relief might not be available due to the absence of certain parties, the entire suit should not be dismissed if meaningful relief can still be accorded. Sandobal v. Armour & Co., 429 F.2d 249 (8th Cir. 1970). Here, that relief could have focused on the union membership criteria which were in the control of the defendants who remained after the contractors were dismissed. Indeed, failing to proceed perpetuates the dilemma posed by the division of power among all interested parties with respect to controlling the level of black employment in the carpentry trade.
Throughout this litigation, the union and the contractors have pointed to one another as the source of the low black employment in the trade. The unions have stated that they would welcome more black members but have been powerless to obtain them because the contractors have failed to hire blacks. Conversely, contractors have pointed to the union’s entrance requirements and decried them as constraints on their efforts to find qualified blacks. This litigation offered the opportunity to examine one half of that dilemma — the entrance requirements. If those requirements were discriminatory, then they might have been reformed within the context of this lawsuit as it stood before the District Court. The interests of the absent contractors would not have been impaired so long as any modified entrance requirements did not deny them an adequate supply of willing carpenter apprentices. Other courts under similar circumstances have proceeded with Title VII litigation against the employee union in the absence of the employer. See, e.g., Kaplan v. Inter Alliance of Theatrical, etc., 525 F.2d 1354 (5th Cir. 1975). We direct the court below to do so in this ease.
II.
In the interest of judicial economy, we turn now to correct certain legal errors which appear in the remainder of the District Court’s order. '
A. Class Certification
The order contains contradictory resolutions of the class certification question. At one point the plaintiff class is approved as certified, yet at another, the appellants are found to have failed the numerosity requirement of Fed.R.Civ.P. 23. We believe the court viewed the adequacy of the representation question too narrowly. Although the Supreme Court has condemned “across the board representation,” see East Texas Motor Freight v. Rodriguez, 431 U.S. 395, 97 S.Ct. 1891, 52 L.Ed.2d 453 (1977), the *167facts of this case do not pose that problem. Here, only the entrance requirements to the trade were challenged. This case is not analogous to a broad attack on an employer’s hiring, firing, and seniority procedures. That instance requires separate representatives because the interests of employees are pitted against those of non-employee applicants. In contrast, this lawsuit presents only the claims of the latter group. Inasmuch as applicants’ interests are the same, it is not necessary to divide them into subgroups according to the individual entrance criteria. To do so unreasonably burdens the plaintiffs. The class certification, as approved by the District Court at page sixteen in its order should stand.
B. The High School Diploma Requirement
As part of their prima facie case, plaintiffs in this disparate impact ease were required to come forward with evidence indicating that the entrance criteria imposed by defendants disproportionately excluded members of their class. See International Brotherhood of Teamsters v. United States, 431 U.S. 324, 335-39, 97 S.Ct. 1843, 1854-56, 52 L.Ed.2d 396 (1977). Consequently plaintiffs submitted detailed statistical evidence to demonstrate that defendant’s high school diploma requirement more often excluded blacks than whites. However, in its analysis, the District Court chose to examine the percentages of blacks and whites in the Toledo area who had only a high school diploma. Those persons possessing higher educational degrees were thereby excluded. This survey revealed no disparate impact. We feel the District Court’s focus was flawed. In this case, and any other in which only general population figures are available, the better analysis examines the exclusionary effects of particular entrance requirements. Thus the relevant statistic compares blacks and whites who do not possess the needed credential. Authority for this form of analysis is plentiful. See Griggs v. Duke Power Co., 401 U.S. 424, 91 S.Ct. 849, 28 L.Ed.2d 158 (1971); Donnell v. General Motors Corp., 576 F.2d 1292 (8th Cir. 1978); E. E. O. C. v. Local 14, Inter. Union of Op. Engineers, 553 F.2d 251 (2d Cir. 1977); Pettaway v. American Cast Iron Pipe Co., 494 F.2d 211 (5th Cir. 1974); Johnson v. Goodyear, 491 F.2d 1364 (5th Cir. 1974); United States v. Georgia Power Co., 474 F.2d 906 (5th Cir. 1973). On remand the District Court should consider the exclusionary effect of requiring a high school diploma for entrance into the carpenter apprenticeship program.
C. Entrance Examination
With respect to the entrance tests, we feel the District Court failed to review critically the probative evidence before it.5 Even though both sides presented a wealth of data on this subject, the court, rather than examining it for evidence of discriminatory impact, simply concluded that the tests were not discriminatory and were in-fact job related. The latter conclusion, the job relatedness of the tests, was erroneously entered, for the court never determined whether plaintiffs had established their prima facie case. Absent this initial determination the defense of business necessity is not material.
Furthermore, the court erred when it dismissed all of appellants’ examination evidence on the ground that their experts’ sample size was too small. This is clearly contrary to the trend in cases employing this type of statistical analysis.6 We seri*168ously doubt that none of appellants’ evidence was probative. On remand, those statistics which are valid should be recognized and considered in light of all circumstances. That is, statistics should not be considered in a vacuum, see Teamsters, supra, at 340, 97 S.Ct. at 1856; instead the entrance criteria should be evaluated in terms of the credence given them by those who impose them. The record reveals that many new union members were not required to submit to all or some of the entrance criteria. We are speaking here of those new members who were organized into the union from existing job sites as opposed to those who entered via the apprenticeship program. The District Court failed to make specific findings on this question. In our view, however, the discretionary application of the entrance requirements itself concedes that they are not strictly related to job performance. This fact then bolsters the plaintiff’s prima facie case and can only render proving the defense of business necessity more difficult.
Furthermore, the court used a “bottom line” analysis in reaching its conclusions concerning the entrance criteria. We feel that this is the wrong way to approach this type of case. The focus in a disparate impact action must be upon discriminatory barriers to fair employment. Bottom line figures cannot insulate unfair intermediate hurdles to equal opportunity. The analysis we favor is that described in Teal v. State of Connecticut, 645 F.2d 133 (2d Cir. 1981).
Finally we note that the statistical data is fragmented and that both sides have attempted to prove their case by using both pre- and post-complaint figures. Therefore an exception to the general rule excluding post-complaint evidence is warranted, and hence the court may properly consider all valid statistics. Yet this approach must be undertaken with an eye for post-complaint data generated under the pressure of the legal action. See Rice v. Gates Rubber Co., 521 F.2d 782 (6th Cir. 1975).
We also find that the District Court improperly applied the principal of res judicata to the facts of this case. Although certain claims of the appellants were heard by the Ohio Civil Rights Commission, that body’s conclusions are not binding on those litigants in federal court. Sinicropi v. Nassau County, 601 F.2d 60 (2d Cir. 1979); Mitchell v. National Broadcasting Co., 553 F.2d 265 (2d Cir. 1977). Cf. Cummins v. Parker Seal Co., 516 F.2d 544 (6th Cir. 1975) (federal remedy not barred by an adverse decision of the state agency).
The order of the District Court is reversed and remanded for proceedings consistent with this opinion.