Coates v. Johnson & Johnson

CUDAHY, Circuit Judge,

concurring:

This is not necessarily a case where “figures lie and liars figure.” Statistical evidence, as long as it remains rooted in knowledgeable interpretation, can provide a unique perspective on otherwise elusive problems of employment discrimination. In this case, had I been the trier of fact, I believe I would have given more weight to some of the statistical evidence than did the district court. See Mozee v. Jeffboat, Inc., 746 F.2d 365, 371-73 (7th Cir.1984). But while I agree with the majority that Judge Leighton may have been mistaken in some particulars and that the case is very close, I too think the result is probably justified. There are however two matters concerning the statistical evidence which it may be worthwhile to address in more detail. With respect to defendants’ assertion of disciplinary differences as an explanation for differences in rates of discharge, plaintiffs argued that (a) defendants could not cite discipline as an explanation unless they articulated some reason for thinking that discipline-was not applied discriminato-rily; and (b) that in any case discipline was a less plausible explanation if the correct disciplinary period were used, which plaintiffs believed to be the longer of the periods discussed in the majority opinion. The majority has said that the plaintiffs must lose on the first issue because the burden does not fall to the defendants to show that the reason adduced — discipline—is not itself biased; and that although the plaintiffs are right about the second issue, the statistics were not taken seriously enough by the trial judge for the mistake to have created reversible error. T agree that plaintiffs lose on both issues, but for reasons substantially different from those set out by the majority.

*554 On the matter of burden, I do not agree that in every case in which defendant has alleged a factor within his control — one that is itself subject to discrimination — as an explanation of discrimination, the burden shifts mechanically back to the plaintiff to prove that that factor is biased. As I understand the law, it will ordinarily be up to the defendant in such a case to at least articulate a reason for thinking that the explanatory factor in question is not biased, and a failure to raise that point will work to the plaintiffs advantage. In this case, however, I do not think that a blind application of that principle is called for.

Here the defendant pointed to disciplinary actions to explain the differing rates of discharge, and the plaintiff alleged that discipline itself was discriminatory. When discrimination in the explanatory factor is alleged, a failure to present evidence either way should ordinarily work against the defendant. James v. Stockham Valves & Fittings Co., 559 F.2d 310, 332 (5th Cir. 1977), cert. denied, 434 U.S. 1034 , 98 S.Ct. 767 , 54 L.Ed.2d 781 (1978) (merit ratings within defendant’s control cannot be used to explain discrimination where defendant has not proved that they are free of discrimination themselves); 1 Trout v. Hidalgo, 517 F.Supp. 873 , 886 n. 47 (D.D.C.1981), affd in relevant part sub nom. Trout v. Lehman, 702 F.2d 1094 (D.C.Cir.1983) (“inappropriate to include as an independent variable a factor within defendants’ control unless it has been established that they did not discriminate in exercising that control”).

Nevertheless it seems to me acceptable to hold in the case before us that the plaintiffs have failed to meet their ultimate burden. The fairness of discipline can in principle be established only by measuring the discipline imposed against the deficiency of the underlying behavior or conduct. Although I am doubtful that a convincing case one way or the other could often be built upon the quality of the underlying behavior (of which there may be little record beyond the discipline itself), I am prepared to concede that there may be instances in which such a case (measuring discipline against underlying behavior) could be made. In the case before us, however, no evidence of the underlying behavior — independent of the fact of discipline itself — exists which sufficiently confirms or refutes the hypothesis of discrimination. Whomever the burden falls on in such circumstances, whether it is the defendant who must give some reason for thinking that discipline is not discriminatory or the plaintiff who must show that it is, will very likely lose; and, although I am sure that it is wrong to routinely place that burden on the plaintiff, I do not see that justice is done here by mechanically placing it on the defendant either.

An allegation of discrimination is a hypothesis, after all, to be tested against the totality of the evidence; and when the prescribed routine for evaluating that hypothesis breaks down, common sense must take over. With due deference to the subtle and complex statistical machinery the parties have invoked, the essential facts are that blacks were discharged at about twice the *555 rate at which whites were discharged, and that in spite of the higher discharge rate, the percentage of blacks working at the plant remained about the same over the life of the plant. EEOC Brief for Appellants at 5. Evidently blacks were both hired and fired at a higher rate than whites. Those who are concerned with efficiency above all tell me that it is generally inefficient to discriminate, but it would seem to take a particularly obtuse and self-destructive employer to fire, for discriminatory reasons, blacks who did not deserve to be fired, only to replace them with other blacks. The hypothesis that discipline was unfairly administered (and discharges therefore unfairly fostered) requires, given the fact that minority representation did not decrease, the assumption that the defendant acted in an apparently irrational way; it seems to me more plausible to suppose the employer acted more or less in his own interest. If he did, and lacking other explanations, the evidence viewed in its totality would tend more or less to support the competing hypothesis that discipline was not discriminatorily applied. In these circumstances, that is, it is reasonable to suppose that the employer would have preferred keeping qualified blacks on the job over firing them without reason and hiring other blacks to replace them. This is a reasonable hypothesis, at least, lacking any direct method of determining whether discipline as a whole was fair or unfair, and one which could justify the conclusion of the trier of fact.

I reiterate that I am far from believing, as the majority apparently does, that when a defendant has alleged a factor within his control as an explanation for apparent discrimination, it is always the plaintiffs burden to show that the factor is itself biased. In the ordinary course of things, the defendant should not be allowed to appeal to a factor within his control as an explanation unless he can show that it is not biased, and a complete lack of evidence either way should ordinarily work to the detriment of the defendant and not the plaintiff. I am not even prepared to say that in the case of discipline, where proof of bias or the lack of it may be forever out of reach, the plaintiff must always bear that burden. What I do say is that in such a circumstance as discipline presents we are sometimes called on to forsake the routine and rely on other ways of testing the hypothesis of discrimination, and that in this case it would take something more than the evidence available to convince me that the trial judge should be reversed. 2 The trial judge felt that the plaintiff did not carry his ultimate burden, and I am not prepared to disagree with the majority that his determination was not clearly erroneous. We are in disagreement only in that I think there ought not to be an automatic shifting of burdens such as the majority contemplates.

I also note the second matter relating to the appropriate time frame for disciplinary statistics, which is pressed by the plaintiffs as a second answer to the defendants’ reliance on discipline as an explanation. In this connection, the majority finds that the district judge’s acceptance of the defendants’ measure of a relevant disciplinary period to be clearly erroneous, and then excuses the matter because the district judge did not find the statistical evidence to be particularly helpful in any event. If a judge ignores the evidence that he should have taken into account, it seems odd to say that it does not matter because he did not find it that helpful anyway. The better procedure might be to send the case back to see if the district judge finds it helpful when he does take it into account. The other side of the coin is that, although I would probably have given more weight to the statistical evidence, I am not sure that the district judge’s choice of the relevant disciplinary period was in fact clearly erroneous. On this point, defendants’ position *556 seems defensible: the number of disciplinary measures an employee has incurred in the immediate past seems clearly more relevant than the number incurred over the whole period of his employment. An employee who, in spite of a prior clean record, has been disciplined ten times in the last year is in trouble. An employee who has been disciplined ten times in ten years would not, on the face of it, seem to be in the same sort of trouble. The majority has relied on anecdotal evidence to support plaintiff’s position that the longer record is the appropriate measure — and the evidence does lend some support to that position. Nevertheless, when all the evidence is taken together, it would be difficult for me to say that the trial judge was not within his discretion in declining to rely heavily on this evidence. If I felt, as the majority does, that the judge was clearly wrong on this matter, I would be in favor of remanding. Since I do not feel that his decision was clearly erroneous, I am in accord with the majority’s affirmance.

Thus, although this is a close and troubling case, I concur in the result reached by the majority.

. The majority argues that in Stockham the court had before it evidence provided by the plaintiff that showed a disparity in the explanatory factor, and concludes that that may explain the court’s decision to require the defendant to rebut the inference of discrimination in the explanatory factor. That explanation of Stock-ham seems to miss the point that in providing an explanation of discrimination the defendant must always rely on a factor in which there is a disparity between the group alleging discrimination and others. The factor adduced could not explain a disparity unless it applied unequally to the two groups itself. A difference can only be explained by a difference: unequal pay can be explained by differing numbers at a certain job level; differing rates of discharge can be explained by differing rates of discipline. Thus the defendant himself, in providing the explanation, will claim a disparity in the explanatory factor. Th¿ reasoning in Stockham cannot have rested on the fact that the plaintiff produced evidence of such a difference, and the proposition that I draw from the case is that a defendant cannot rely on possibly biased factors, in general. If the plaintiff alleges discrimination in the explanatory factor, then the defendant must produce some non-discriminatory motive for the disparity that, in the nature of the thing, must exist.

. I emphasize that I am not suggesting that a compensating hiring rate will always (or even frequently) defeat the allegation of discrimination in discharge (or discipline); what I am suggesting is that the higher hiring rate raises a question. I have no doubt that an answer to that question could be given, in other cases, on other facts. It has not been given here.