dissenting in part:
I agree that the Federal Communications Commission is duty bound to investigate, by one means or another, substantial allegations that its licensees have engaged in employment discrimination,1 and then to allow renewal-challengers “meaningful participation” in evaluating the fruits of that scrutiny.2 And we will all agree that if the Commission is unwilling to seek relevant information itself3 it must provide challengers with the opportunity to gain access to the evidence bearing on their charges.4 Thus I join unreservedly in the court’s opinion in these respects.
My difficulty with today’s decision stems solely from the court’s failure to utilize the occasion as an en banc body to review thoroughly the Commission’s present philosophy toward employment discrimination by its licensees. This default is especially troubling in light of the court’s professed endeavor “to state definitively the position of this Court on the issues raised and to govern related cases in the future.”5 Although I concur to the result in Bilingual II6 I think the court errs grievously in rejecting out of hand the allegations of employment bias in Chinese for Affirmative Action,7 and in placing its imprimatur on the license renewal granted therein.
The statistical showing in Chinese, while not so compelling as that in Bilingual II, did establish a prima facie case of intentional racial discrimination when measured by normal evidentiary standards, from which neither the Commission nor the court has advanced any cause for departure. The Commission, however, gave those statistics only cursory examination — on the thesis that the minority group involved is not dominant in the broadcast community — and ultimately it disposed of the challenger’s presentation on grounds that I can regard only as an effective ábandonment of its equal-employment policies in the recent past. The court, in turn, ignores the Commission’s stated rationale and substitutes another, and, even if it were not our duty to refrain from a displacement of that sort, I could not accept the court’s unexplained conclusions. The court disdains both a hard look at the Commission’s decision, and a rigorous study and elucidation of its own reasoning.
I must, then, dissent from the court’s endorsement of the Commission’s limitation of the coverage of its anti-discrimination rules to groups dominant in the licensee’s service area. I must also demur to the court’s condonation of Commission acceptance — without even the seeds of a logical explanation — of statistical disparities that naturally give rise to an inference of purposeful racial bias. I must protest, too, the court’s undifferentiated use of the licensee’s post-term statistics and similarly misleading factors in an effort to rebut that inference.
I
The history of Commission concern over discriminatory employment practices in the *252broadcasting industry has been amply recounted elsewhere.8 It is necessary here merely to highlight the evolution of current regulatory policy. In a 1968 statement of unusual eloquence, the Commission first indicated that it could not countenance racial distinctions,9 and two years later this position was extended to gender differentiations as well.10 An allegation that discrimination was a factor in an employment decision would, the agency reasoned, “in almost all cases where a substantial showing is made, require a hearing for its resolution” because “the matter is of such a serious nature as to call into question the basic grant of operating authority.”11 Never during the ensuing decade has the Commission overtly deviated from this fundamental concept,12 and even today the court acknowledges that “intentional discrimination almost invariably would disqualify a broadcaster from a position of public trusteeship.” 13
The Commission has predicted its condemnation of disparate treatment on a number of grounds.14 In 1976, the Supreme Court held that administrative agencies are not required by general public-interest considerations “to seek to eradicate discrimination,”15 but it specifically noted that the Commission’s interest therein “can be justified as necessary to enable [it] to satisfy its obligation under the Communications Act . to ensure that its licensees’ programming fairly reflects the tastes and *253viewpoints of minority groups.”16 The Commission in turn has identified two related ways in which discrimination can frustrate that objective. First, the duty to ascertain and meet the programming needs of all segments of the community served is undercut when the licensee himself has indulged conspicuously in employment bias— both because minorities might not bother to present their preferences to him and because he might well be insensitive to them.17 Second, minority employees are more likely to originate programming that rectifies deficiencies encountered by their groups.18 The Commission assumes, quite logically, that purposeful differentiation inexorably will have some or all of those untoward effects,19 and thus does not demand a showing in each case that prohibited conduct actually affected programming.20 The Commission has also declared that its rules prohibiting purposeful discrimination are required by its responsibility to foster the continuing good character of its licensees,21 including the allied obligation to license only those who conform to federal and state law.22 Thus, while neither is it the Equal Employment Opportunity Commission nor is it charged with enforcement of Title VII for its own sake, the Commission has perceived and pronounced that it can fulfill its special mandate only by a complete ban on race and gender discrimination.
And so, the Commission’s antidiscrimination regulations are not gratuitous. The Commission has itself noted the substantiality of the argument that, all else aside, it has a constitutional duty not to sanction discrimination.23 One need not approach the problem at hand on ground so high, however, for indubitably the Commission bears a statutory responsibility to consider information highly pertinent to the public interest, and, “[f]rom the outset,” as the court notes, “the Commission has recognized that the public interest is not served by licensees who engage in intentional em*254ployment discrimination.”24 Consequently, the Commission is not free lightly to abandon its attentiveness to licensee prejudice in employment activity. It may exercise discretion in considering relevant factors, of course, but it cannot ignore any material contention fairly raised on that score.
More recently, the Commission has added a second distinct requirement to its antidiscrimination rules. Licensees must not only be racially neutral in employment decisions but also must make affirmative efforts to recruit, hire, train and promote women and members of minority groups.25 The Commission has refined its directives in this area and has developed a model equal-employment-opportunity (EEO) program to guide licensees in discharging their affirmative-action obligation.26 The Commission tells us “that its EEO rules are prospective in nature and, in most cases, [it] seek[s] to lead a licensee who lacked an adequate affirmative action plan in the past to adopt specific policies to ensure an active program in the future.”27
While prospective orientation for the affirmative-action requirement is understandable,28 we encounter difficulty here because the Commission apparently now seeks to extend the same treatment to its bar against intentional discrimination,29 and that of course is another matter entirely. I am unable to comprehend how a completely prospectively-oriented antidiscrimination policy meets the Commission’s own declarations of what the public interest demands, and at least on this point the court seems to be in accord.30
If, as the Commission found in 1968, licensees who purposely discriminate are unlikely to ascertain or satisfy the programming needs of minorities, it is hard to see how affirmative-action programs, even those to proceed at a more than deliberate pace, become a panacea for change of licensees’ — as opposed to their workforces’ — insensitivity to the programming tastes of significant segments of the broadcast audience.31 Nor is such action calculated to *255modify basic character shortcomings — although it might disguise them a bit. Two of the factors critically underlying the Commission’s concern with willful employment discrimination are thus abandoned by a policy that looks only to the future.32 And the Commission has made no attempt to explain its dramatic departure from past principles; 33 it merely asserts that this court approved the shift in NOW v. FCC.34
Our decision in NOW does not support that construction. There we determined initially that the Commission had acted reasonably in finding that allegations of past discrimination against women, in light of all the information before it, were insufficient to warrant a hearing.35 We then analyzed the affirmative-action contentions and held that the Commission did not abuse its discretion by electing to push the existing licensee toward adequate execution rather than to confer the license upon someone who could only promise better performance in overcoming unintentionally-erected barriers to equal opportunity.36 I consider the distinction drawn in NOW well founded, as I did when I joined in that decision, and the full court, as I read its opinion herein,37 today adopts it. Nothing in NOW, however, supports the theses that the Commission is free to focus only upon assuring future compliance with its rules, that it may ignore the licensee’s past activity or that, if that conduct is unacceptable and unmitigated, it may nonetheless extend the grant of public trusteeship. To the contrary, this court and, until recently, the Commission have steadfastly resisted such an abrogation of the policy set in motion just a decade ago by the Commission.38
II
It is now far beyond dispute that broadcasters must not intentionally discriminate, and that past activity of that nature will justify, if not demand, nonrenewal of a communications license.39 Still we are left with the problem of determining what showing is necessary to generate a “substantial and material [issue] of fact” statutorily requiring — absent Commission efforts otherwise resolving the dispute40 — a hearing before the Commission properly may “find that grant of the application would be consistent with” the public interest.41
*256The type of showing most commonly attempted — and that with which we deal here — is statistical in nature. The Commission initially expressed distrust of statistics42 but, before that misgiving was subjected to judicial scrutiny, the Commission purported to come around to the view that “ ‘[i]n the problem of racial discrimination, statistics often tell much, and Courts listen.’ ”43 Quite recently the Commission has declared that “a statistical prima facie case of employment discrimination may be made out by comparing the composition, by race and sex, of the company’s staff with the city’s labor force composition in general.” 44 Indeed, because a statistical presentation is perfectly capable of supporting an inference of prohibited conduct, it would be irrational to ignore it totally.45 Moreover, in the field of discrimination, it is my experience that statistics are recognized as a uniquely valuable type of evidence: “statistics are frequently the best available evidence of . discrimination”46 because they are often “the only available avenue of proof.” 47 Consequently, the pertinent question is not whether statistics will ever suffice, for they often will, but what level of statistically-indicated disparity will trigger an administrative investigation.
*257' We held in NOW that the Federal Communications Commission may employ standards different from those utilized by the Equal Employment Opportunity Commission in carrying out its mandate.48 That conclusion seems unassailable, especially in light of the Supreme Court’s decision in NAACP v. FPCl49 FCC’s main retrospective concern — as distinguished from its interest in prospective affirmative action — is with intentional discrimination;50 EEOC, on the other hand, is chartered to search out and remedy both discriminatory intent and discriminatory effect.51 And not all statistics that indicate disparate impact raise an inference of prohibited design. Similarly, an EEOC charge against an FCC licensee will not necessarily require license denial if the charge implicates only untoward effects and not improper purpose.52
The Communications Commission must allocate its limited resources to a broad range of important responsibilities, of which the fight against job-bias is but one — although certainly a vital one. Because any statistical showing leaves open the possibility, slight though it might be, that the disparity is due to chance,53 the Commission may adopt a level of statistical significance54 reasonably calculated to avoid exorbitant efforts over coincidental differentials — what statisticians call “type of errors.”55 And more than that, because the Commission is concerned with long-term effects on the use of airwaves in the public interest and not with remedying isolated incidents,56 it could perhaps ignore some differences that are significant but of small magnitude.57 Disparities may often be attributable to the temporarily-undetected character flaw of a single personnel officer or employment agency, and not to licensee-approved or -condoned employment practices.
More broadly, all that statistics can ever show is that the null hypothesis — that the composition of the licensee’s labor force is the result of random distribution and that *258race and sex had no relationship to employment opportunity — should be rejected.58 Even though a disparity is statistically significant, it does not irrebuttably prove the reason for the difference, which could be any number of things — intentional racism by employment officials,59 a disinclination among minority-group members to seek media jobs, unintentional use of culturally-biased employment tests and standards or the lingering effects of past discrimination by educational institutions.60 Thus, when the magnitude of disparity is relatively low in comparison with a convincing refutation of calculated bias, the Commission might conclude that the probability that a hearing would reveal a discriminatory intent is too insubstantial to justify an expenditure of precious resources.61 But, “absent explanation, it is ordinarily to be expected that nondiscriminatory hiring practices will in time result in a work force more or less representative of the racial and ethnic composition of the population in the community from which employees are hired.”62 The corollary is that where a long-term disparity is established and no satisfactory explanation is given, it ordinarily can be expected that intentional manipulation has been worked.
To list some of the factors that the Commission could employ in examining statistically-based license challenges is not to say that it has ever looked at any of them. And indeed it has not. So far as I am able to discern, it has merely adopted — without explanation — a test used by this court to review its own action; and more recently it has chosen — again without offering any rationale whatever — an apparently arbitrary cut-off figure. Though one might sympathize with the Commission’s disinclination to understand and interpret an array of employment statistics, that is a duty that the Commission cannot — and must not— avoid. To be sure, the Commission has great discretion in choosing among policies to best implement the directives of the Communications Act. But having settled upon a policy, it must either implement it or discard it in a manner that is both well reasoned and well explicated.63 Since the Commission has never expressly abandoned the policy announced in 1968, it is our duty *259to see that Commission practice in this field reasonably carries out the' prohibition of intentional bias.
The Commission’s decisional course has been to determine whether a particular disparity falls safely within a “zone of reasonableness,” 64 but the numerical boundaries of this zone were not set by the Commission until several months ago.65 Before that time, the Commission’s decisionmaking traded on what this court had tolerated as within a “zone of reasonableness” — for we ourselves had coined the phrase in 1972.66 Because we then upheld Commission approval, as a matter within the ambit of administrative judgment, of hiring practices resulting in minority employment at only about 29 percent of parity with minority representation in the general workforce,67 the Commission seized on that troublingly low figure — which had survived judicial scrutiny, but simply in the context of that particular lawsuit — as a substitute for judgments tailored to the individual circumstances of other employment discrimination cases.
We, of course, do not directly assess the performances of licensees; that the Commission does, and we review the Commission. Thus the zone-of-reasonableness concept which we have applied in several cases,68 and which the Commission has borrowed, started out as an appellate-court test of Commission judgmental action in examining licensees’ conduct in choosing their employees, and not as an administrative yardstick for measurement of what the Commission must determine — the consistency of that conduct with the public interest.69 All we held in the first of that line of cases, Stone v. FCC,70 was that in 1972 the Commission had not slipped into the zone of arbitrariness by holding that minority employment at 29 percent of parity did not give rise to a substantial question of intentional discrimination in a situation involving voluntary and active recruitment and *260placement by the licensee.71 That certainly was not to say that the Commission reasonably could pivot its decisions — in all of the myriad situations coming before it — at the very edge of the naked statistics in that one case. As Stone itself illustrates, a particular ratio of minority employment to minority representation in the total workforce may portend different inferences when viewed against varying backgrounds and circumstances. And while the Commission must, of course, reach the same result in identical cases, the same bare statistics do not necessarily germinate identical or even similar situations, and the Commission too often has not visibly given attention to such nonstatistical factors as evidence connoting or negativing intent, or circumstances suggesting a need for renewal despite past employment improprieties or for nonrenewal in light thereof. In so doing, the Commission does not exercise discretion but simply concludes that statistical showings within a broad range cannot sufficiently indicate intentional discrimination. Decisions of that kind are statistically arbitrary since a figure as low as 29 percent of parity would usually give rise to a substantial inference of purposeful bias absent countervailing circumstances.72
We can — and indeed we must — indulge the Commission great latitude in deciding whether to designate a renewal application for investigation or hearing. But it goes without saying that the Commission may not yield to licensees, the freedom to engage in willful misconduct. Thus the Commission must develop — obediently to its responsibilities as guardian of the public interest in broadcasting — its own framework for analyzing statistics, and as a court we must appraise its effort to determine whether it comports with the Commission’s statutory duties.
The Commission has informed us that it will at least make further inquiry when licensee employment of minorities is less than 50 percent of parity with their percentages in the area workforce,73 or when employment in upper-level positions74 is *261less than 25 percent of parity.75 Because this proposal is not at issue here, I have no need to discuss it, but I must note that it arrived unadorned with any explanation. This is the more discomforting because figures showing 50 percent of parity — not to mention those indicating only 25 percent of parity — would, both as a matter of common sense and of statistical science, often be significant and relatively large in magnitude.76 I can only assume that when the Commission first makes a decision under its newly-revealed template it will fully elucidate it and its application. It may not simply invoke talismanically the fact that it is not the Equal Employment Opportunity Commission to reject out of hand a statistical showing that in analogous areas of the law would indicate “substantial under-representation” 77 and erect a prima facie case of intentional discrimination. Because it is not EEOC, the Commission may justifiably be guided by consideration other than those central to EEOC’s responsibilities. But if the question is what demonstrates intentional discrimination — without regard to the factors that go into deciding what to do about it — the Commission cannot reasonably decide that statistics that have been accepted as proving disparate treatment do not so much as raise a substantial issue of such conduct in its eyes.
Ill
My central concern in this opinion is the Commission’s handling of the statistical attack by Chinese for Affirmative Action on Columbia Broadcasting System’s employment of Asian-Americans at KCBS, its San Francisco AM station. An initial but easily surmountable objection to CAA’s challenge is the Commission’s apparent78 position that it need not examine statistics reflect*262ing job-treatment of Asian-Americans — a minority group not “clearly the predominate minority in a licensee’s SMSA.” 79 To begin with, this unquestionably was a major — and unexplained — departure from past Commission policy. In 1974, for instance, the Commission decreed that “[t]he nondiscrimination provision applies to all persons, whether or not the individual is a member of a conventionally defined minority group,”80 and many other manifestations of this view might be cited.81
Despite the court’s approval today of this abrupt shift,821 see no reason whatever for countenancing purposeful discrimination merely because it is aimed at only one small group.83 The Commission’s antidiscrimination rules were promulgated in large part to assure that programming needs of minority groups in the licensees’ audiences are met.84 In other programming-related areas such as ascertainment, the Commission has not limited the licensee’s responsibility to one of only serving “dominant” groups.85 Even if a justification for a divergent view on job-bias were possible, the Commission has not yet expressed it, and an explanation — by the Commission, not the *263court — is necessary.86 Since the Commission has articulated no rationale for ignoring the complaints of Asian-Americans merely because of the size of their group, I must proceed to the question whether the Commission properly can reject on the merits the statistical showing they proffered.
I am satisfied that CAA’s figures, unless somehow successfully rebutted,87 generated an inference of calculated employment discrimination. With 6.9%88 of the Standard Metropolitan Statistical Area89 labor force consisting of Asian-Americans, the expected number within the licensee’s 249 employees during the license term was 17.90 As a result of a preliminary Commission inquiry, we know that KCBS employed a total of ten, and perhaps only nine, Asian-Americans for various lengths of time during the license term under review.91 Although this *264performance exceeds the 50-pereent-of-pari-ty range now proposed by the Commission, the unlikelihood that so great a disparity occurred solely by chance is only 3.67 percent — or about one in 27 — which is within the region normally considered significant.92 This requires rejection of the null hypothesis that the licensee’s hands are unquestionably clean. It does not conclusively prove intentional discrimination, of course, but it certainly does summon the Commission to ascertain whether, in light of all relevant information — on hand or procurable through further inquiry — the chances of an eventual finding of willful discrimination merit a hearing.
It was because Asian-Americans are not a dominant minority group in KCBS’s service area that the Commission felt that an investigation into intentional discrimination was *265unwarranted.93 The court shuns the ground administratively assigned for rejecting CAA’s contentions and instead detects in the Commission’s opinion a finding that CAA’s statistical showing was examined but found wanting.94 That opinion is hardly a model of precise analysis, but it is possible to discern four factors that the Commission could have deemed incompatible with an inference of purposeful discrimination against Asian-Americans had it thought itself bound to inquire into that subject. These are KCBS’s in-term performance with respect to women and minorities overall, its affirmative action plan, its post-term performance with respect to Asian-Americans and the fact that it had employed some Asian-Americans during the term.95
Even assuming that the Commission actually did consider these factors to the end the court supposes, it never ventured any rationale for an implicit conclusion that they served to so dispel the inference of past purposeful discrimination naturally flowing from the statistical disparity demonstrated by CAA. And it is by now a basic tenet of administrative law that agency explication of the basis of its decision not only aids judicial review but also improves agency decisionmaking;96 a corollary is that judicial review of agency action is not to be undertaken in the dark.97 I would *266thus remand Chinese, as well as Bilingual II, so that the Commission can tell us just how it arrived at its decision. Because, however, the court does not do that, I think it necessary to protest the fallacies into which the Commission- — with the court’s approval — seems to be slipping. In doing so, I will treat the Commission’s decision as if it had expressly relied upon the four factors mentioned by the court as counteractants to the inference of discrimination against Asian-Americans.
It is important to recall at the outset that the question is not whether a challenger has proven intentional discrimination but whether it has made a showing substantial enough to require further investigation on the Commission’s part.98 And as we have cautioned before, discrimination, even when consciously engaged in, is a very subtle process, and its attackers have “limited resources and no procedural tools” for rooting it out.99 Primarily for those reasons, no one could seriously argue that indicia of discrimination, if in fact it has occurred, will always be readily available, or that absent direct evidence the propriety of employment practices must be presumed.100 But the Commission often does seem to operate on a presumption to that effect,101 as is exemplified by its oft-expressed distrust of challengers’ statistics and its generally unquestioning acceptance of the protestations and statistical offerings of its licensees. A classic example of that is its treatment of the four factors that purportedly contradict the inference of deliberate bias in this case.
The Commission’s misuse of statistical evidence is perhaps best shown by its conclusion that KCBS’s statistics with regard to women and minorities overall somehow undercut the inference of intentional discrimination against Asian-Americans.102 I need not consider whether evidence tending to show that a licensee does not discriminate against some minority groups is at all probative of whether he discriminated against another,103 for these statistics — while seemingly better than those of many other licensees — still disclosed significant disparities, especially with respect to the more important positions at KCBS.104 The figures are *267at best no indication of anything,105 and I am unable to comprehend the Commission’s half-stated position that statistics can be trusted to indicate an absence of intentional discrimination but not to indicate its presence.106
Similarly puzzling is the Commission’s reliance on the fact that KCBS has an affirmative action plan. Four years ago we held that, when a licensee has voluntarily developed and energetically pursued a recruitment and placement program, the Commission may be justified in leaving uninvestigated a statistical disparity that otherwise would demand further inquiry.107 Since the license term then under study, however, the Commission has imposed upon all licensees the duty of formulating quite extensive affirmative action plans,108 and here the Commission simply noted that KCBS had such a program — which apparently reflects no more than compliance with the Commission’s rules.109 A broadcaster’s decision to abide by the requirement that he devise a plan is hardly conclusive proof that he has not intentionally discriminated in the past or is not planning to do so in the future; even hardcore racists are likely to comply superficially or grudgingly with Commission edicts once no alternative is in sight. The existence of an affirmative action program evidences good intentions only when on paper it is more than the rules demand or its implementation indicates a sincere desire to employ more minority-group workers, and not just to creep within the terms of the rules.
Only slightly less perplexing is the Commission’s nearly complete dependence on KCBS’s post-term statistics,110 which the court accepts.111 We have held that post-term statistics may be viewed, as obviously they must, in .applying the acid test to the adequacy of a licensee’s affirmative action program — whether it has actually resulted in the hiring and promotion of members of *268the protected class.112 But we have refused to consider post-term statistics in examining a contention of term-time intentional discrimination.113 This position is bottomed on the well-established requirement that a licensee must run on his in-term record;114 it is soundly based, too, on the rationales underlying the Commission’s concern with willful discrimination, which as noted earlier do not support a policy of total prospectivity.115 I need not discuss whether post-term statistics can more limitedly be referred to — for example, to get a statistically-valid sample of a licensee’s employment practices — for here both the Commission and the court put controlling emphasis on the post-term figures.116
Lastly, we come to the circumstances that some Asian-Americans were employed by KCBS during the license term. The value of this evidence is seriously diminished as an initial matter by the stark fact that most did not remain for long.117 On a more general level, it should suffice to say that tokenism and other attempts to hide discriminatory designs are no more commendable than overt prejudice. Racism and sexism assume many forms, and a complete bar to employment is but one of them. Statistieally-significant underrepresentation can be the result of purposeful discrimination as much as can absolute exclusion.118 For instance, race or sex may simply be one factor in an employment decision — as for an employer who hires only exceptional black applicants119 — but the prohibited animus is still there.120 Thus, while a total absence of minorities is damning in most situations, the employment of but a few is not a plus factor.121 Their presence is taken into account in the initial statistical computation and should not be counted twice.122
IV
To sum up, it seems to me that the Commission’s performance in these two cases reflects its broad misapprehension of the value — indeed, the logical compulsion — of statistical evidence and of the methods by which statistical showings can be rebutted. That malaise has fatally infected the result in Chinese. It could be that the Commission ultimately should renew KCBS’s license, but it has not yet satisfactorily ex*269plained why that should be so. The Commission cannot brush statistical challenges aside with flip rejoinders about the nature of its responsibilities. It is not the Equal Employment Opportunity Commission, and the Communications Act is not the Civil Rights Act, but evidence is evidence, and no less than other tribunals must the Commission give it due respect.
In both cases, the challengers’ statistics indicated intentional discrimination, which the Commission admits is normally a compelling ground for denial of a license renewal. The Commission refused to consider the showing in Chinese, and even had it done so the fact remains that it has not elucidated what countervailing factors could and how they would overcome the evidentiary inference. And if they do outweigh the inference, it should be for the Commission, not the court, to explain how that is done. Accordingly, I concur in the court’s disposition of Bilingual II, but respectfully dissent from the court’s failure to reach the same outcome in Chinese.
. Majority Opinion (Maj. Op.) text at notes 34-38.
. Id., 193 U.S.App.D.C. at 249, 595 F.2d at 634.
. I concur in the court’s refusal to require predesignation discovery as a general matter. Maj. Op. text at notes 51-52; see FCC v. Schreiber, 381 U.S. 279, 290, 85 S.Ct. 1459, 1467-1468, 14 L.Ed.2d 383, 391-392 (1965) (“[t]o permit federal district courts to establish .administrative procedures de novo would, of course, render nugatory Congress’ effort to insure that administrative procedures be designed by those most familiar with the regulatory problems involved”).
. Maj. Op., 193 U.S.App.D.C. at 249, 595 F.2d at 634.
. Id., 193 U.S.App.D.C. at 239, 595 F.2d at 624.
. No. 75-1855.
. No. 75-2181.
. E. g., NOW v. FCC, 181 U.S.App.D.C. 65, 78, 555 F.2d 1002, 1015 (1977); Nondiscrimination in the Employment Policies and Practices of Broadcast Licensees (Nondiscrimination— 1976), 60 F.C.C.2d 226, 227-228 (1976); United States Commission on Civil Rights, Window Dressing on the Set: Women and Minorities in Television 131-132 (1977).
. Petition for Rulemaking to Require Licensees to Show Nondiscrimination in Their Employment Practices (Nondiscrimination — 1968), 13 F.C.C.2d 766 (1968).
. See Petition for Rulemaking to Require Broadcast Licensees to Show Nondiscrimination in Their Employment Practices (Nondiscrimination — 1970), 23 F.C.C.2d 430, 431 (1970).
. Nondiscrimination — 1968, supra note 9, 13 F.C.C.2d at 771. As the Commission elucidated, “a petition or complaint raising substantial issues of fact concerning discrimination in employment practices calls for full exploration by the Commission before the grant of the broadcast application. . . . ” Id. (emphasis supplied).
. See, e. g., Nondiscrimination in the Employment Policies and Practices of Broadcast Licensees (Nondiscrimination — 1975), 54 F.C.C.2d 354, 361 (1975) (“[w]here it appears that the broadcaster has followed discriminatory employment practices, we will not hesitate to order a hearing to resolve any substantial and material questions of fact relating to the broadcaster’s employment policies and practices”).
. Maj. Op., 193 U.S.App.D.C. at 244, 595 F.2d at 629.
. By “disparate treatment” I mean not only consciously plotted acts that result from racial animus, but also conduct arising from the thoughtlessness of stereotypes and irrational generalizations. Both conscious purpose and subconscious purpose — neglect—can fall within this category, for behind today’s failure to think of minority interests is yesterday’s deliberate decision to discriminate. As Justice Stevens has explained:
Habit, rather than analysis, makes it seem acceptable and natural to distinguish between male and female, alien and citizen, legitimate and illegitimate; for too much of our history there was the same inertia in distinguishing between black and white.
Mathews v. Lucas, 427 U.S. 495, 520, 96 S.Ct. 2755, 2769, 49 L.Ed.2d 651, 669 (1976) (Stevens, J., with Brennan & Marshall, JJ., dissenting). It makes little difference whether a licensee purposely calculated that an applicant was unable to handle the responsibilities of a position because of his race or whether the licensee’s socialization was so imbued with the stereotype that it simply never occurred to him that the applicant could do the job. Improper purpose and neglect have the very same impact on the policies that the Commission’s job-bias rules seek to further. As has been said in another context, “the arbitrary quality of thoughtlessness can be as disastrous and unfair to private rights and the public interest as the perversity of a willful scheme.” Hobson v. Hansen, 269 F.Supp. 401, 497 (D.D.C.1967), aff'd as modified sub nom. Smuck v. Hansen, 132 U.S.App.D.C. 372, 408 F.2d 175 (en banc 1969). So although I will use such words as “intentional” and “purposeful,” I am not ruling our inconsiderateness bom of prejudice.
. NAACP v. FPC, 425 U.S. 662, 670, 96 S.Ct. 1806, 1812, 48 L.Ed.2d 284, 292 (1976).
. Id. at 670 n.7, 96 S.Ct. at 1812 n.7, 48 L.Ed.2d at 292 n.7.
. Nondiscrimination — 1968, supra note 9, 13 F.C.C.2d at 770; accord, Nondiscrimination— 1976, supra note 8, 23 F.C.C.2d at 229; see NOW v. FCC, supra note 8, 181 U.S.App.D.C. at 80, 555 F.2d at 1017, quoting NAACP v. FPC, supra note 15, 425 U.S. at 670-671 n.7, 96 S.Ct. at 1812 n.7, 48 L.Ed.2d at 292 n.7.
. See Black Broadcasting Coalition v. FCC, 181 U.S.App.D.C. 182, 187 n.21, 556 F.2d 59, 64 n.21 (1977), citing NAACP v. FPC, supra note 15, 425 U.S. at 670-671 n.7, 96 S.Ct. at 1812 n.7, 48 L.Ed.2d at 292 n.7. See also TV 9, Inc. v. FCC, 161 U.S.App.D.C. 349, 358, 495 F.2d 929, 938 (1973), cert. denied, 419 U.S. 986, 95 S.Ct. 245, 42 L.Ed.2d 194 (1974) (Commission must give weight to minority ownership and participation in comparative-application proceeding as a portent of increased diversity of program content); Garrett v. FCC, 168 U.S.App.D.C. 266, 273, 513 F.2d 1056, 1063 (1975) (“[t]he entire thrust of TV 9 is that black ownership and participation together are themselves likely to bring about programming that is responsive to the needs of the black citizenry”).
. NAACP v. FPC, 172 U.S.App.D.C. 32, 42, 520 F.2d 432, 442 (1975), aff'd, NAACP v. FPC, supra note 15 (“[t]he [Commission] has reached the conclusion that a broadcaster’s programming will inevitably fail fairly to reflect the tastes and viewpoints of minorities if those minorities are systematically excluded from the broadcaster’s employ” (emphasis supplied)).
. Both the Commission and this court have recoiled from direct examination of licensees’ programming decisions. See, e. g., NOW v. FCC, supra note 8, 181 U.S.App.D.C. at 73, 555 F.2d at 1010.
. NOW v. FCC, supra note 8, 181 U.S.App.D.C. at 80, 555 F.2d at 1017, citing Nondiscrimination — 1968, supra note 9, 13 F.C.C.2d at 769.
. Nondiscrimination — 1968, supra note 9, 13 F.C.C.2d at 769; see FCC v. American Broadcasting Co., 347 U.S. 284, 289 n.7, 74 S.Ct. 593, 597 n.7, 98 L.Ed. 699, 705 n.7 (1954).
. Petition for Rulemaking to Require Broadcast Licensees to Show Nondiscrimination in Their Employment Practices (Nondiscrimination — 1969), 18 F.C.C.2d 240, 241 (1969); see . Comment, EEOC Regulatory Intervention: An Undeveloped Means of Enforcing Title VII, 62 Geo.L.J. 1753, 1763-1764 (1974) (“[t]o avoid placing their imprimatur on discrimination, regulatory agencies must ascertain that regulatees do not engage in unlawful practices before awarding them benefits”). Compare NAACP v. FPC, supra note 15.
. Maj.Op. text at note 21, citing Nondiscrimination — 1969, supra note 23, 18 F.C.C.2d at 241-242 and Nondiscrimination — 1968, supra note 9, 13 F.C.C.2d at 769; cf. Black Broadcasting Coalition v. FCC, supra note 18, 181 U.S.App.D.C. at 187 n.21, 556 F.2d at 64 n.21.
. Maj.Op., 193 U.S.App.D.C. at 243, 595 F.2d at 628; see 47 C.F.R. §§ 73.125(b), 73.-599(b), 73.680(b), 73.793(b) (1976); Nondiscrimination — 1976, supra note 8, 60 F.C.C.2d at 230; Inquiry Into the Employment Policies and Practices of Certain Broadcast Stations Located in Florida, 44 F.C.C.2d 735, 735-736 (1974).
. Nondiscrimination — 1976, supra note 8.
. FCC Responses to Questions Posed by Court in D.C. Cir. Nos. 75-1855 & 75-2181, at 7 (Sept. 16, 1977); see National Broadcasting Co. (1972 WRC Renewal), 58 F.C.C.2d 419, 422 (1976).
. The affirmative-action requirement, unlike the prohibition on intentional discrimination, is concerned with future results, not the present character of the licensee — it is an effort to lead broadcasters to better performance. See Maj.Op. text at notes 24-26. Character defects, on the other hand, are more subtle and permanent than mere numbers, and it is obviously more difficult to determine whether a correction in character has actually been made. For that very reason, the Commission has adamantly refused to give a “second bite at the apple” to licensees whose character infirmities surfaced in acts of deceit. See Grenco, Inc., 39 F.C.C.2d 732, 737 (1973). Thus the Commission will resort to prospective orientation — for example, in the form of a short-term renewal— only if the misrepresentation was seemingly inadvertent. See id.; Bluegrass Broadcasting Co., 43 F.C.C.2d 990, 993-994 (1973).
. Supplemental Memorandum of Appellee FCC in No. 75-2181, at 6 n.4; United States Commission on Civil Rights, supra note 8, at 135; see CBS, Inc., (1974 KCBS Renewal), 56 F.C.C.2d 296, 302 (1975) (referring to the “licensee’s current performance in hiring and promotion”).
The question whether the Commission could properly determine never to impose the sanction of renewal denial for unintentional disparities and failures to comply with the affirmative action rules is not before us today. See note 12 supra. The statistical presentation in this case is sufficient, in the absence of reasoned explication to the contrary by the Commission, to give rise to an inference of purposeful bias. See note 92 infra.
. See Maj.Op., 193 U.S.App.D.C. at 243-244, 595 F.2d at 628-629.
. As has been posited, “ownership is the most significant factor influencing the content of media.” Case Note, 43 U.Cin.L.Rev. 669, 677 (1974). See also note 18 supra.
. We are not called upon by this case to determine whether the underlying objectives are forsaken by the prospective orientation reflected in a Commission decision to take a chance on a licensee who had admitted his past misdeeds and apparently is prepared to make a sincere effort to reform. Compare Office of Communication of United Church of Christ v. FCC (UCC I), 123 U.S.App.D.C. 328, 342 & n.28, 359 F.2d 994, 1008 & n.28 (1966) with note 28 supra.
. E. g., Continental Air Lines, Inc. v. CAB, 179 U.S.App.D.C. 334, 344, 551 F.2d 1293, 1303 (1977) (“[r]easoned decisionmaking requires an agency to explain changes of policy from past decisions”).
. Supra note 8.
. 181 U.S.App.D.C. at 81-82, 555 F.2d at 1018-1019. As a panel, we were bound in NOW by the “zone of reasonableness” concept, the contours of which had been established by our earlier decisions. See notes 66-68 infra and accompanying text. We are now sitting en banc and thus are free to constrict the zone to parallel an evolving understanding of statistical proof vis-a-vis the concerns that eventuated in the Commission’s antidiscrimination endeavors. See notes 76-77, 92 infra and accompanying text.
. See 181 U.S.App.D.C. at 82-83, 555 F.2d at 1019-1020.
. See Maj.Op., 193 U.S.App.D.C. at 243-244, 595 F.2d at 628-629.
. See notes 11-13 supra and accompanying text.
. See notes 11 & 13 supra and accompanying text.
. The court indicates that challengers have no right to Commission assistance in checking the veracity of facts represented by licensees. Maj.Op., 193 U.S.App.D.C. at 249-250, 595 F.2d at 634-635. I do not take this as absolving the Commission of the responsibility of making such inquiry itself when the circumstances necessitate. Nor do 1 take it as indicating that a challenger’s presentation can be outweighed solely by a licensee’s sworn, conclusory denial of intentional discrimination.
. Communications Act of 1934, tit. III, § 309, 48 Stat. 1085, as amended, 47 U.S.C. § 309(d) (1970). See generally Stone v. FCC, 151 U.S.App.D.C. 145, 150-151, 466 F.2d 316, 321-323 (1972). The Commission’s function under *256§ 309(d) is not the same as that of a court faced with a motion either for dismissal or summary judgment. The Commission is neither prohibited from looking to countervailing evidence nor compelled to draw every reasonable inference in favor of the challenger. Thus the Commission is not required to hold a hearing merely on the basis of an unsupported allegation of intentional discrimination, see Folkways Broadcasting Co. v. FCC, 126 U.S.App.D.C. 123, 127, 375 F.2d 299, 303 (1967), or on some slight evidence giving rise to no more than a weak inference in that regard. See Columbia Broadcasting Coalition v. FCC, 164 U.S.App.D.C. 213, 218, 505 F.2d 320, 325 (1974). The allegation must not only be material and genuinely • in dispute, but must create a substantial issue. On the other hand, however, allegations not rising to the level of those necessitating a hearing might still be sufficient to preclude the Commission from making the essential finding that renewal is in the public interest without first engaging in an inquiry less demanding than a hearing. And, indeed, even if no “substantial and material question of fact is presented” by any challenger, the Commission is required to hold a hearing if “for any reason” it cannot “make the requisite finding that the public interest would be served.” Citizens Comm. to Save WEFM v. FCC, 165 U.S.App.D.C. 185, 198, 506 F.2d 246, 259 (en banc 1974).
. Nondiscrimination — 1968, supra note 9, 13 F.C.C.2d at 771-772.
. Nondiscrimination — 1969, supra note 23, 18 F.C.C.2d at 243, quoting Alabama v. United States, 304 F.2d 583, 586 (5th Cir.), aff'd, 371 U.S. 37, 83 S.Ct. 145, 9 L.Ed.2d 112 (1962) (footnote omitted); accord, Nondiscrimination —1970, supra note 19, 23 F.C.C.2d at 432 (“[statistics may not give definitive answers, but they clearly can raise valid questions”); Inquiry Into the Employment Policies and Practices of Certain Broadcast Stations Located in Florida, supra note 25, 44 F.C.C.2d at 736 (statistics showing “highly disproportionate representation . . may constitute evidence of discriminatory practices”); Nondiscrimination —1976, supra note 8, 60 F.C.C.2d at 229; see id. at 236 (“[o]ur past experience leads us to believe that . . statistics of available minorities and women represents [sic] an acceptable means of measuring a licensee’s EEO performance”). Proper statistics involving other material issues can meet the requirements of Section 309(d) and thus necessitate a hearing. See WLVA, Inc. v. FCC, 148 U.S.App.D.C. 262, 273-274, 459 F.2d 1286, 1297-1299 (1972).
. Teleprompter of Worcester, Inc., 42 Rad. Reg.2d (P&F) 287, 302 (1978).
. E. g., Teamsters Int’l v. United States, 431 U.S. 324, 339, 97 S.Ct. 1843, 1856, 52 L.Ed.2d 396, 418 (1977), citing Mayor of Philadelphia v. Educational Equality League, 415 U.S. 605, 620, 94 S.Ct. 1323, 1333, 39 L.Ed.2d 630, 644 (1974) (“[statistics are equally competent in proving employment discrimination”); Kinsey v. First Regional Sec., Inc., 181 U.S.App.D.C. 207, 216, 557 F.2d 830, 839 (1977) (“statistical disparity between the proportion of blacks in the employer’s work force and the proportion of blacks in the relevant labor market constitutes a prima facie case of discrimination”).
. Copus, The Numbers Game Is the Only Game in Town, 20 How.L.J. 374, 380-381 (1977).
. United States v. Ironworkers Local 86, 443 F.2d 544, 551 (9th Cir.), cert. denied, 404 U.S. 984, 92 S.Ct. 447, 30 L.Ed.2d 367 (1971); accord, Brown v. Gaston County Dyeing Mach. Co., 457 F.2d 1377, 1382 (4th Cir.), cert. denied, 409 U.S. 982, 93 S.Ct. 319, 34 L.Ed.2d 246 (1972), quoting United States v. Dillon Supply Co., 429 F.2d 800, 804 (4th Cir. 1970) (“proof of overt racial discrimination in employment is seldom direct”; “[Recognizing this, we have found ‘error in limiting Title VII to present specific acts of racial discrimination’ ”).
. NOW v. FCC, supra note 8, 181 U.S.App.D.C. at 80, 555 F.2d at 1017. It is noteworthy, however, that Congress did not intend the creation of EEOC to “relieve any Government agency . . of its . . . primary re-, sponsibility to assure nondiscrimination in employment as required by the Constitution and statutes . . .” Civil Rights Act of 1964, 42 U.S.C. § 2000e-16(e) (Supp. V 1975). The two agencies have recently agreed to a formal coordination of their responsibilities. See 46 U.S.L.W. 2500 (Mar. 28, 1978).
. See notes 15-16 supra and accompanying text.
. See notes 28-32 supra and accompanying text.
. See Teamsters Int’l v. United States, supra note 45, 431 U.S. at 335-336 n.15, 97 S.Ct. at 1854-1855 n.15, 52 L.Ed.2d at 415 n.15; Griggs v. Duke Power Co., 401 U.S. 424, 432, 91 S.Ct. 849, 854, 28 L.Ed.2d 158, 165 (1971) (“Congress directed the thrust of [Title VII] to the consequences of employment practices, not simply the motivation” (emphasis in original)).
. See NOW v. FCC, supra note 8, 181 U.S.App.D.C. at 81, 555 F.2d at 1018. Such statistics would certainly indict the licensee’s commitment to meeting its affirmative action duty, however, and certain failures of this sort might well justify nonrenewal. See notes 25-26 supra.
. Note, Beyond the Prima Facie Case in Employment Discrimination Law: Statistical Proof and Rebuttal, 89 Harv.L.Rev. 387, .394 (1975).
. Social scientists commonly accept 1 in 20 as the maximum chance of erroneously rejecting random distribution, and a statistical showing less likely to mislead is considered to be significant. See Note, supra note 53, 89 Harv.L.Rev. at 394-395 n.30, 400-401 n.58, citing Albermarie Paper Co. v. Moody, 422 U.S. 405, 433-434, 95 S.Ct. 2362, 2379, 45 L.Ed.2d 280, 305 (1975). A 1 in 40 chance of error has been suggested as the standard for “overwhelming” statistical showings. Bogen & Falcon, The Use of Racial Statistics in Fair Housing Cases, 35 Md.L.Rev. 59, 77 (1974).
. P. Hoel, Introduction to Mathematical Statistics 108 (4th ed. 1971). See generally, H. Blalock, Social Statistics 159-165 (2d ed. 1972).
. NOW v. FCC, supra note 8, 181 U.S.App.D.C. at 80, 555 F.2d at 1017; see Focus Cable of Oakland, Inc., 65 F.C.C.2d 35, 43 (1977); Nondiscrimination — 1969, supra note 23, 18 F.C.C.2d at 241.
. See H. Blalock, supra note 55, at 163 (statistical significance is not the same as practical significance because in isolation it tells nothing about the importance or magnitude of the differences).
. Finkelstein, The Application of Statistical Decision Theory to the Jury Discrimination Cases, 80 Harv.L.Rev. 338, 359 (1966).
. This explanation is especially likely when the challenger has proffered specific instances of discriminatory conduct, but such direct evidence of intent is not essential if the statistics alone are strong. Montlack, Using Statistical Evidence to Enforce the Laws Against Discrimination, 22 Clev.St.L.Rev. 259, 260 (1973); see Sweeney v. Board of Trustees of Keene State College, 569 F.2d 169, 175 (1st Cir. 1978) (statistics alone can fulfill burden of proof in individual disparate-treatment case).
. See Note, Employment Discrimination: Statistics and Preferences Under Title VII, 59 Va.L. Rev. 463, 467 (1973). No innocuous explanations were proffered by the licensees at bar. Methods for determining the weight and employer gave various factors in making employment decisions are discussed in Note, supra note 53.
. On the other hand, unlike most other employers, broadcasters operate under a requirement that they make affirmative efforts to increase their employment of women and minority individuals. See note 25 supra. Thus a disparity in a broadcaster’s employment profile is less likely to be the result of unintentional practices than would be the case for such differentials in the figures for other employers. See also Maj.Op. at note 49.
. Teamsters Int'l v. United States, supra note 45, 431 U.S. at 340 n.20, 97 S.Ct. at 1856 n.20, 52 L.Ed.2d at 418 n.20; accord, Nondiscrimination — 1975, supra note 12, 54 F.C.C.2d at 362; Note, supra note 53, 89 Harv.L.Rev. at 393 (“courts are willing to proceed as though discrimination explains the observed conditions even in the absence of direct evidence”).
. See Atchison, T. & S. F. Ry. v. Wichita Bd. of Trade, 412 U.S. 800, 808, 93 S.Ct. 2367, 2375, 37 L.Ed.2d 350, 362-363 (1973); UAW v. NLRB, 148 U.S.App.D.C. 305, 317, 459 F.2d 1329, 1341 (1972); Distrigas of Mass. Corp. v. FPC, 517 F.2d 761, 766 (1st Cir. 1975). Fairly read, the Commission’s brief discussion in this case
is so perplexing as to sow doubt whether this is a process of reasoned policy making, with a change in direction put in effect for a navigational objective, or the confusion of an agency that is rudderless and adrift.
Public Serv. Comm’n v. FPC, 167 U.S.App.D.C. 100, 115, 511 F.2d 338, 353 (1975).
. Nondiscrimination — 1976, supra note 8, 60 F.C.C.2d at 228-229 (noting that the zone of reasonableness contracts over time, but without telling us what its parameters now are).
. See notes 73-75 infra and accompanying text.
. Stone v. FCC, supra note 41, 151 U.S.App.D.C. at 161, 466 F.2d at 332.
. Id. at 161 & n.6, 466 F.2d at 332 & n.6. The use of general workforce statistics is legally acceptable because most broadcasting jobs involve skills “that many persons possess or can fairly readily acquire.” Hazelwood School Dist. v. United States, 433 U.S. 299, 308 n.13, 97 S.Ct. 2736, 3742 n.13, 53 L.Ed.2d 768, 777-778 n.13 (1977); see Note, The FCC’s Role in Providing Equal Employment Opportunity for Minority Groups, 53 B.U.L.Rev. 657, 658 (1973).
. See Black Broadcasting Coalition v. FCC, supra note 18; NOW v. FCC, supra note 8; Bilingual Biculturai Coalition of Mass Media, Inc. v. FCC (Bilingual I), 160 U.S.App.D.C. 390, 492 F.2d 656 (1974).
. One might, with the benefit of hindsight, argue that the court erred in creating so broad a zone of reasonableness, and that the mistake should now be corrected by the court en banc. After all, the zone in which we acquiesced in Stone comes close to labelling as reasonable statistics that “do more than speak for themselves — they cry out ‘discrimination’ with unmistakable clarity.” Muniz v. Beto, 434 F.2d 697, 702-703 (5th Cir. 1970) (figures showing 15 to 20% of parity). And the full court’s adoption of the “contracting zone” concept, see Maj.Op. at notes 16 & 33, might be taken as an implicit admission of earlier error, for statistics that give rise to an inference of intentional discrimination in 1978 had the same capability in 1972. Nonetheless, in Stone we dealt not only with a licensee that had made voluntary efforts toward equal opportunity, but also with one of the first challenged applications of a Commission policy formulated only four years earlier. See Bilingual I, supra note 68, 160 U.S.App.D.C. at 393, 492 F.2d at 659 (“Stone represented an initial effort, not a final codification!)] [w]hile we did not endorse the statistical challenge raised in Stone, we did not signal satisfaction with the status quo on employment discrimination”). It is not extraordinary for courts to allow something of a grace period before enforcing full compliance with newly-enunciated policy — even if everyone should have known how to conduct himself before the law itself required conformance. Compare Brown v. Board of Educ. (Brown II), 349 U.S. 294, 301, 75 S.Ct. 753, 757, 99 L.Ed. 1083, 1106 (1955) with Alexander v. Holmes County Bd. of Educ., 396 U.S. 19. 20, 90 S.Ct. 29, 24 L.Ed.2d 19, 21 (1969).
.Supra note 41.
. Stone v. FCC, supra note 41, 151 U.S.App.D.C. at 159, 466 F.2d at 330.
. Whether such a disparity is actually significant would depend on the number of station employees and the number of minority-group employees expectable in the absence of discrimination. See note 92 infra; Ochoa v. Monsanto Co., 473 F.2d 318, 320 (5th Cir. 1973) (“the smallness of the numbers demonstrates that the Court was not compelled to allow such statistical showing to set in train the usual presumptions”).
. FCC responses, supra note 27, at 1. A guideline beginning at 50% of parity and increasing 5% per year to a maximum of 90% of parity was proposed in Note, supra note 67. Although we have no call to discuss particular figures, such a stepped criterion seems a reasonable effort to combine the requirements of the antidiscrimination and affirmative-action rules. The Commission on Civil Rights argues for a test of 80% of parity. United States Commission on Civil Rights, supra note 8, at 151.
Licensees with fewer than ten employees are not rigidly subject to the Commission’s 50% guideline because the Commission believes that the sample size is too small to give rise to any statistically reliable finding. FCC Responses, supra note 27, at 1-2. This would often but not always be true. See note 72 supra.
.The Commission currently requires licensees to report on employment practices in their “upper-four” job categories. Licensees as a whole have reported that fully 77.2% of all broadcast employees occupied such positions in 1975, United States Commission on Civil Rights, supra note 8, at 143, and we are in the dark as to why this may be so. Without elucidation, I can accept this startling statistic only as reflecting a distorted view of what jobs are important, making a mockery of any figures purporting to show that a licensee has an appreciable number of women or minority-group members in critical positions. Put another way, the most mundane jobs must often be classified as upper-level.
For instance, in this very case CBS asserts that women are proportionally represented in important positions because in 1975 half of all females occupied upper-four positions. This claim ignores the telling fact that of only 14 employees not in the highest categories all but one were women. This is hardly an “exemplary” performance. Contra, Brief for Intervenor CBS, at 12. Because statistics showing overall proportional employment may easily mask a design to keep minorities and women out of truly influential jobs and because those positions are the more likely to have an impact on programming, the Commission needs reliable information on upper-level employment. See NOW v. FCC, supra note 8, 181 U.S.App.D.C. *261at 82, 555 F.2d at 1019. It has at long last decided to consider changes in its method of categorization. Petitions for Rule Making to Amend FCC Form 395 and Instruction, 66 F.C.C.2d 955 (1977).
. The Commission has not explained why the parity guideline should differ for upper-level positions, and none is obvious to me. In fact, because the primary underlying concern is with programming, it would seem that the more important jobs should receive closest scrutiny by the Commission.
. For instance, the Supreme Court recently held that a showing that the percentage of Mexican-Americans selected for grand juries over a ten-year period in a Texas county was only 50% of parity not only gave rise to an inference of intentional discrimination but made out a prima facie case in that regard. Castaneda v. Partida, 430 U.S. 482, 97 S.Ct. 1272, 50 L.Ed.2d 498 (1977); see Turner v. Fouche, 396 U.S. 346, 90 S.Ct. 532, 24 L.Ed.2d 567 (1970) (violation found although 62% of parity); Whitus v. Georgia, 385 U.S. 545, 87 S.Ct. 643, 17 L.Ed.2d 599 (1967) (violation found on 33% of parity). See also Washington v. Davis, 426 U.S. 229, 241, 96 S.Ct. 2040, 2048, 48 L.Ed.2d 597, 608 (1976) (showing of intentional discrimination is a necessary element in establishing that Equal Protection Clause was breached in constituting jury).
Particularly characteristic of the Commission’s reasoning is its counsel’s statement in oral argument that the agency uses the “zone of reasonableness” only to find out whether the figures promote an inference that something is wrong with a licensee’s affirmative-action plan, and not to ascertain whether there is an indication of purposeful discrimination. See, e. g., Swanson Broadcasting, Inc., 42 Rad.Reg.2d (P&F) 1002 (Renewal Branch 1978) (although licensee’s performance fails under 50/25 standard, renewal will be granted without hearing but with imposition of reporting conditions). The Commission will infer intentional misconduct only when the statistics unveil a situation so far outside the zone that it may be classified as egregious. Because statistics that do not seem sufficiently outrageous to the Commission often are probative of intentional discrimination, see, e. g., notes 88-92 infra and accompanying text, the Commission’s decision to assume in all such cases that ineffective affirmative action is the real culprit is totally arbitrary. The Commission has advanced this theory to justify prospective orientation in Bilingual II, but the court pays it the deference it truly deserves — absolutely none. See Maj.Op. at note 49 (“the Commission could but speculate as to whether the disparities owed to intentional discrimination, unintentional discrimination, or chance”).
. Castaneda v. Partida, supra note 76, 430 U.S. at 495, 97 S.Ct. at 1280, 51 L.Ed.2d at 511 (figures showing 50% of parity indicate “substantial under-representation”).
. The Commission’s opinion does not clearly indicate that it reached this conclusion, see CBS, Inc., supra note 29, 56 F.C.C.2d at 302, but the Commission’s brief on appeal endeavors to remove the uncertainty. See note 79 infra and accompanying text. See also note 94 infra.
. Brief for Appellee FCC in No. 75-2181, at 12. Even a brief perusal of the Commission’s opinion shows that it did not make the kind of zone-of-reasonableness comparison between the percentage of KCBS’s employees who were Asian-American and that for the general labor force that it made for the overall minority population. The Commission has subsequently indicated that it may not look even at figures for a single dominant group, but only to those for all minorities in toto. Rahall Broadcasting of Indiana, Inc., 66 F.C.C.2d 295, 296 (1977).
. Inquiry Into the Employment Policies and Practices of Certain Broadcast Stations Located in Florida, supra, note 25, 44 F.C.C.2d at 735-736. Compare Mitchell v. United States, 313 U.S. 80, 94, 97, 61 S.Ct. 873, 877, 878, 85 L.Ed. 1201, 1210, 1212 (1941).
. Compare National Broadcasting Co., supra note 58, 58 F.C.C.2d at 425 n.ll (“our own scrutiny of the station’s EEO performance, we believe, must extend in some measure to a 11 protected groups” (emphasis supplied)); Inquiry Into the Employment Policies and Practices of Certain Broadcast Stations Located in Florida, supra note 28, 44 F.C.C.2d at 735-736. (“[n]o person may be denied employment or related benefits on the grounds of his or her race, color, religion, national original or sex” (emphasis in original)); Nondiscrimination— 1970, supra note 10, 23 F.C.C.2d at 432 n.2 (concern is with all minorities, not just the largest group); Nondiscrimination — 1968, supra note 9, 13 F.C.C.2d at 770 (“[a] refusal to hire Negroes or persons of any race or religion clearly raises a question of whether the licensee is making a good faith effort to serve his entire public” (emphasis supplied)).
When the Commission first became concerned with discrimination against women, it stated that Owing to its limited resources it would continue to focus on blacks, Indians, Mexican-Americans and Asian-Americans — but not women. Nondiscrimination — 1970, supra note 10, 23 F.C.C.2d at 431. See also Nondiscrimination — 1969, supra note 23, 18 F.C.C.2d at 244 (Asian-Americans listed as one of the “significant minority groups”). Now, however, without explanation, the Commission separately considers statistics for women but seemingly not for Asian-Americans.
. Maj.Op. at note 7. The court says that the Commission is not permitting discrimination aimed at small groups but is merely rejecting statistical evidence for those groups. Because there is no justification for rejecting all statistics for all small groups in ail situations, see note 86 infra, and because direct evidence of discrimination is seldom if ever obtainable, see text accompanying notes 46-47 supra, refusing to accept their statistical presentations is an effective abandonment of concern for small groups. When the Commission is checking up on a licensee in the absence of a challenger’s contentions related to a specific group, a look to the overall statistics might well be a reasonable step, but here we have specific allegations related to a single group.
. Historically, the smallest groups have often ■ been those most heavily victimized by racial prejudice.
. See notes 16-20 supra and accompanying text.
. See Primer on Ascertainment of Community Problems by Broadcast Applicants, 27 F.C.C.2d 650, 683-684 (1971) (licensee must determine “what groups constitute the community” and ascertain problems of those that are “significant,” as measured by size, influence and other factors). Ironically, the Commission in this very case assumed that KCBS had a duty to ascertain and attempt to meet the programming needs of not only Asian-Americans but also the even smaller group of Chinese Americans. CBS, Inc., supra note 29, 56 F.C.C.2d at 296-301; see Joint Appendix (J.App.) 132.
. On the Commission’s behalf, the court posits the rationale that “such small percentages generally will provide no meaningful basis for comparison with the workforces of most radio stations, which often employ only 25 to 75 workers.” Maj.Op. at note 7. Yet the sample size is actually much larger, for over the three-year license term many employees come and go. Moreover, we do not even know where the “25 to 75” figure came from. Even if the statistics for such stations would often be unreliable, that would be no reason to reject those statistical presentations, like the one in this case, that undeniably are meaningful. And even if one accepts the court’s rationale, it does not rescue the Commission in this case, since KCBS employed more than 75 workers.
As a general matter, the value of a statistical showing depends not merely upon the size of the sample but also upon the extent of the expected observation and the degree of disparity — elements unknowable until the Commission looks at the particular group’s statistics. In short, as the application of the logic of statistics that it purports to be, the court’s conclusion that statistics for small groups are always unreliable is simply wrong.
Curiously, the court somehow later concludes that Asian-Americans constitute “a significant minority group.” Maj.Op., 193 U.SApp.D.C. at 246, 595 F.2d at 631. I am not sure how a significant group differs from a dominant one and why statistical evidence as to this significant group should not have been accepted even under the theory the court supplies for the Commission.
. See note 60 supra and accompanying test.
. CBS questions this statistic, arguing that the correct figure is 6.2%. Brief for Intervenor CBS in No. 75-2181, at 4 n.6. Because the Commission did not reject CAA’s figure, it must be accepted for present purposes.
. The use of Standard Metropolitan Statistical Area figures is proper in cases in which the labor force in that region reasonably approximates that from which the licensee draws its employees. See Stone v. FCC, supra note 41, 151 U.S.App.D.C. at 161, 466 F.2d at 332. In this case, 6.9% should be considered a “strong” figure since the percentage of Asian-Americans in the city itself is much higher.
. Due to the inadequacies of the Commission’s reporting form, see note 91 infra, we do not know how many employees worked for KCBS during the license term. We do know that about 83 workers were employed on the average at any one time, that the average Asian-American left after nine months and that the turnover rate for Asian-Americans was only “relatively high.” See Maj.Op. text at note 43. Even assuming that the tenure for all employees averaged a third again longer than that for Asian-Americans, we have a total sample of 249 employees, 6.9% of which is roughly 17.
. CBS, Inc., supra note 29, 56 F.C.C.2d at 302, ¶ 19. The minority-employment reports initially filed by KCBS with the Commission disclosed the employment of only one Asian-American during the license term and that for only a short time. After CAA challenged this as being totally disproportional, KCBS responded that nine others had actually been employed but for various reasons had not appeared on the earlier report forms. While CAA did not contest the accuracy of this claim, the Commission requested of KCBS and received further explanatory and supplementary information. See Maj.Op. text at notes 6-11. One of the nine was a Filipino, and the parties disputed whether he should be classified as an Asian-American. See Opposition of CAA to Motion to Dismiss and Reply to Opposition of CBS, Inc., at 20 n.l (Dec. 20, 1974), J.App. 174 n.l. But see CAA’s Petition to Deny at 7 (Nov. 1, 1974), J.App. 15. The Commission apparently assumed that he should not, concluding that only “eight individuals were hired who do not appear” on the reporting forms. CBS, Inc., supra note 29, 56 F.C.C.2d at 302. The court now implicitly overturns this, finding that nine Asian-American employees originally were unreported. Maj.Op., 193 U.S.App.D.C. at 246, 595 F.2d at 631; see id. at notes 9 10 and accompanying text. Nonetheless, inclusion of this employee in the Asian-American category does not materially alter the statistical outcome. See note 92 • infra and accompanying text.
Although I am unable to accede to the court’s conclusion that there were really ten Asian-*264American employees and not just nine, I see nothing unreasonable in the Commission’s evident feeling of general satisfaction with KCBS’s explanation of why the data had not appeared earlier. Nonetheless, the fact that the Commission deemed further information necessary indicates a substantial defect in the reporting system. As KCBS itself put it, “Form 395 is, at best, an imperfect measure of a station’s good faith efforts to provide equal employment opportunities. Because the Form 395 provides for a listing of employees as of a single payroll period each year, it cannot reflect employees who are not with the station during' the designated payroll period.” Motion to Dismiss and Opposition of CBS, Inc., at 21 (Nov. 27, 1974), J.App. 149. The report would be more useful if it reflected employees who come and go in the interim between current reporting periods and if it also called for applicant flow data — information on the characteristics of every person who applied during the preceding year. See B. Schlei & P. Grossman, Employment Discrimination Law 1165 (1976), quoting Hester v. Southern Ry., 497 F.2d 1374, 1379 (5th Cir. 1974) (“[t]he most direct route to proof of racial discrimination in hiring is proof of disparity between the percentage of blacks among those applying for a particular position and the percentage of blacks among those hired”).
. In the absence of Commission adoption and justification of some other standard of statistical significance, we are left to rely upon what social statisticians traditionally have accepted. See note 54 supra. The figure in text was arrived at by using the chi square test. See P. Hoel, supra note 55, at 298. See also F. Mos-teller, R. Rourke, G. Thomas, Probability with Statistical Applications 494 (2d ed. 1970) (binomial distribution chart). This result is close to the upper boundary of normally acceptable type I error, see note 54, supra, but I am unwilling to ignore it until the Commission provides a reasoned explanation for why it is in the public interest to do so. And if the Commission should decide to exclude the Filipino, the result would become even more compelling.
Just looking at the 1974 figures alone, as the court challenges me to do, the evidence still indicates discrimination. Accepting KCBS’s corrections as true, still only one Asian-American was employed in May of 1974 when the 1974 report was submitted. This is barely more than 1% of KCBS’s employees and should be compared to an expected (in the absence of discrimination) figure of nearly six.
The computations could also be made on the basis of person/months, using an expected figure of 206 and an observed frequency of no more than 94. These figures are arrived at •without too much difficulty from the information available to the Commission. The license term in question ran from Decembér 1, 1971, through November 30, 1974. The eight unreported Asian-Americans were employed during that time for a total of approximately 34 person/months. See CBS, Inc., supra note 29, 56 F.C.C.2d at 302; Appendix to Letter from Eleanor S. Applewaite, General Attorney, CBS, Inc., to Richard J. Shiben, Chief, FCC Renewal and Transfer Division (Oct. 3, 1975), J.App. 208. The one reported Asian-American could not have worked more than 18 months during the license term. Compare Memorandum of CBS in Response to a Question Raised in the Department of Justice Amicus Brief, appendix at 6-7 with id., appendix at 11-12. The Filipino employee may have worked as many as 42 months.
Using the test employed by the Supreme Court in Castaneda v. Partida, supra note 76, 430 U.S. at 496-497 n.17, 97 S.Ct. at 1281 n.17, 51 L.Ed.2d at 512 n.17, and in Hazelwood School Dist. v. United States, supra note 67, we arrive at a standard deviation of about 13. The gross disparity is thus approximately 9 standard deviations — and nearly 12 if we exclude the Filipino, see note 91 supra — and “the hypothesis that the [employment practices were] random would be suspect to a social scientist.” Castaneda v. Partida, supra note 76, 430 U.S. at 496-497 n.17, 97 S.Ct. at 1281 n.17, 51 L.Ed.2d at 512 n.17. Using a person/month as the sample measurement factors in the length of the employees’ stay, the shortness of which might be due to intentional discrimination. The choice of months as the unit of time is somewhat arbitrary, however, and, if we are concerned primarily with hiring practices, person/month figures are misleading because each unit is not tied to an independent hiring decision.
. See notes 78-79 supra and accompanying text.
. The court is perhaps confused by the fact that the Commission did look into the employment of Asian-Americans, but the reason it did so was not to explain intentional discrimination but to determine'whether the licensee’s affirmative action plan was working. See CBS, Inc., supra note 29, 56 F.C.C.2d at 302, quoted in note 95 infra. See also note 28 supra (actual employment results are the best test of effectiveness of affirmative action program). Admittedly, I am at a loss to understand why the Commission itself would not apply its antidiscrimination rule to small minority groups to the same extent that it imposes its affirmative action requirement. Our failure to insist that the Commission explain itself allows it to engage in reasoning so murky that the court interprets it in two contradictory ways. Compare Maj.Op. at note 7 (Commission refused to examine statistically-based contentions of discrimination against Asian-Americans) with id. at note 42 (“the Commission examined statistics on the number of Asian-American . . employees during the term”). See also notes 78-79 supra and accompanying text. Although the court believes that “clear guidelines . . . are essential,” Maj.Op. at note 7, it apparently is not sure exactly what the Commission did do in this case.
. CBS, Inc., supra note 29, 56 F.C.C.2d at 302. In response to the contention of conscious discrimination against Asian-Americans, the Commission reasoned:
At present, according to the 1975 form 395 report, KCBS employs one male and two female Orientals, all of whom are professionals. Overall, of KCBS’ 84 full-time employees, 29.8 percent are female and 23.8 percent belong to racial or ethnic minority groups. The 1975 report also shows within the upper four job categories eight male and six female employees from minority groups. These figures represent 13.6 and 24.0 percent respectively of the overall full-time employment (58 male and 25 female) at KCBS; there are six other minority employees presently holding full-time office and clerical positions.
. The combination of licensee’s current performance in hiring and promotion, as reflected on the 1975 annual employment report, and its explanation for the low numbers of Asian-Americans appearing on reports for 1971-74, suffice to show that KCBS’ EEO results — numbers of protected-group employees, viewed in the light of an affirmative action program — are within a zone of reasonableness, and that its past record, while characterized by high job turnover, reflects a willingness to hire minority individuals. Without more, we cannot say this failed to comply with our EEO rules and policies.
Id. (emphasis supplied and footnotes omitted).
. Greater Boston Television Corp. v. FCC, 143 U.S.App.D.C. 383, 393, 444 F.2d 841, 851 (1970), cert. denied, 403 U.S. 923, 91 S.Ct. 2233, 29 L.Ed.2d 701 (1971) (courts must insist “that the agency articulate with reasonable clarity its reasons for decisions, and identify the significance of the crucial facts, a course that tends to assure that the agency’s policies effectuate general standards, applied without unreasonable discrimination”); see Environmental Defense Fund v. Ruckelshaus, 142 U.S.App.D.C. 74, 88, 439 F.2d 584, 589 (1971); K. Davis, Administrative Law of the Seventies § 6.13, at 223-225 (1976).
. Central Power & Light Co. v. FERC, 188 U.S.App.D.C. 56, 58, 575 F.2d 937, 939 (1978) (“[o]ur judicial function embraces ruling on what we fairly discern as the basis of the Commission’s action, but it does not include speculating on the Commission’s intentions”). That *266philosophy has been one of the guiding lights for our court:
The inadequate explanation afforded this particular administrative action gives us no opportunity for intelligent review. Instead, this court is left with a dilemma: we could uncritically approve the agency action, which would amount to a failure to perform properly the duty of a reviewing court. On the other hand, we could search out this matter on our own, which would very probably result in the substitution of a judicial judgment for an administrative judgment. This dilemma can be avoided only if agencies provide an adequate explanation of the basis for the action which they take.
Humboldt Express, Inc. v. ICC, 186 U.S.App.D.C. 141, 144, 567 F.2d 1134, 1137 (1977).
. A hearing is unnecessary if the Commission can reach an informed decision after some less burdensome means of inquiry. See note 41 supra and accompanying text. The choice of method — be it predesignation discovery or something else — is up to the Commission, see note 3 supra, but if the method employed is inadequate the failure to accord a hearing must be reversed. Bilingual I, supra note 68, 160 U.S.App.D.C. at 393, 492 F.2d at 659.
. Id. at 393, 492 F.2d at 659. See also notes 45-47 supra and accompanying text.
. Cf. Office of Communication of United Church of Christ v. FCC (UCC II), 138 U.S.App.D.C. 112, 116, 425 F.2d 543, 547 (1969).
. E. g., FCC Responses, supra note 27, at 3 (“the Commission does . . . recognize that there has — in the past — been job discrimination” (emphasis supplied)).
. See CBS, Inc., supra note 29, 56 F.C.C.2d at 302 quoted in note 95 supra.
. Compare Town & Country Radio, Inc., 41 Rad.Reg.2d (P&F) 151, 161, aff’d on reconsideration, 41 Rad.Reg.2d (P&F) 1177, 1180 (Rev.Bd. 1977) (presence of numerous women employees is irrelevant to question of discrimination against blacks).
. The Commission relied upon post-term statistics in looking at the overall minority-employment picture. See notes 110-116 infra and accompanying text. These statistics were appreciably better than the in-term figures, but still they only approached parity. In 1975, 23.8% of KCBS’s full-time employees were from majority groups, compared to 25.8% of the area workforce. Similarly, the percentage of women in the area workforce was 38.4% and at the station was 29.8%, or approximately 78% of parity. Minority-group employees *267filled only 20% of the upper-four positions; women totalled a little over 17% of such employees, or less than 45% of parity. The Commission concluded simply that the overall figures were within the zone of reasonableness. CBS, Inc., supra note 29, 56 F.C.C.2d at 302, quoted in note 95 supra.
. See Hazelwood School Dist. v. United States, supra note 67, 433 U.S. at 318, 97 S.Ct. at 2747, 53 L.Ed.2d at 783-784 (Stevens, J„ dissenting).
. Note, FCC Failure to Eradicate Employment Discrimination by Broadcast Licensees, 21 St. Louis U.L.J. 150, 156 (1977) (despite its past statements, the Commission’s practice indicates that it does not believe statistics can show intentional discrimination); see, e. g., Nondiscrimination — 1976, supra note 8, 60 F.C.C.2d at 230 n.6. The court seems to sympathize with this unexplained dislike of statistical showings that are anything less than overwhelming. See Maj.Op. text at notes 32-33.
. Bilingual I, supra note 68, 160 U.S.App.D.C, at 392-393, 492 F.2d at 658-659; see Columbia Broadcasting Coalition v. FCC, supra note 41, 164 U.S.App.D.C. at 222, 505 F.2d at 329. See also Stone v. FCC, supra note 41, 151 U.S.App.D.C. at 159, 466 F.2d at 330; Nondiscrimination — 1976, supra note 8, 60 F.C.C.2d at 229; Inquiry Into the Employment Practices of Certain Broadcast Stations Located in Florida, supra note 25, 44 F.C.C.2d 736-737.
. See note 26 supra.
. CBS, Inc., supra note 29, 56 F.C.C.2d at 302 n.13. The court says that the plan was effective, Maj.Op. at note 42 and accompanying text, but the Commission’s opinion indicates no more than that a plan existed on paper.
. See CBS, Inc., supra note 29, 56 F.C.C.2d at 302, quoted in note 95 supra.
. See Maj.Op. at note 41 and accompanying text. Compare id. at note 49 (paradoxically rejecting reliance on post-term statistics in Bilingual II). In explanation, the court does mention in-term statistics for overall minority hiring, see Maj.Op. at note 42, but the Commission’s opinion relies solely upon the post-term, 1975 figures. The court is simply taking over the agency’s function when it attempts to find an acceptable basis for renewal in the in-term statistics not dealt with by the Commission. Compare, e. g., Gulf States Utils. Co. v. FPC, 411 U.S. 747, 764, 93 S.Ct. 1870, 1880, 36 L.Ed.2d 635, 647 (1973), quoting FCC v. Chenery Corp., 318 U.S. 80, 88, 63 S.Ct. 454, 459, 87 L.Ed. 626, 633 (1943) (reviewing court may not advance “an alternative, unstated ground to support an agency’s decision if that ground is one that ‘the agency alone is authorized to make’ ”).
. NOW v. FCC, supra note 8, 181 U.S.App.D.C. at 82, 555 F.2d at 1019.
. Black Broadcasting Coalition v. FCC, supra note 18, 181 U.S.App.D.C. at 184 & n.3, 556 F.2d at 61 & n.3.
. Alianza Federal de Mercedes v. FCC, 176 U.S.App.D.C. 253, 255 & n.6, 539 F.2d 732, 735 & n.6 (1976); see Nondiscrimination — 1976, supra note 8, 60 F.C.C.2d at 246; cf. UCC I, supra note 32, 123 U.S.App.D.C. at 341, 359 F.2d at 65.
. See notes 29-32 supra and accompanying text.
. Because employment practices are relatively enduring in nature, it may be that the threat of temporary renewal-time changes to gain Commission approval is somewhat less obnoxious than it is with regard to programming. See NOW v. FCC, supra note 8, 181 U.S.App.D.C. at 83, 555 F.2d at 1020. On the other hand, it does seem that nearly all Asian-Americans employed at KCBS for one reason or another left much sooner than did most employees. Furthermore, the post-term data in this case still revealed a substantial disparity, see note 104 supra and accompanying text, so that, even if post-term evidence of parity could be given some weight, this is not the case in which to do so.
. CBS, Inc., supra note 29, 56 F.C.C.2d at 302, quoted in note 95 supra. I agree with the court that a sophisticated “revolving door” challenge to the renewal of KCBS’s license is precluded by the exhaustion doctrine. See Maj.Op. text at notes 47-48.
. Castaneda v. Partida, supra note 76, 430 U.S. at 493, 97 S.Ct. at 1279, 51 L.Ed.2d at 509-510.
. Note, supra note 53, 89 Harv.L.Rev. at 394.
. Cf. Barnes v. Costle, 183 U.S.App.D.C. 90, 100-101, 561 F.2d 983, 993-994 (1977).
. Cf Mitchell v. Rose, 570 F.2d 129, 133 (6th Cir. 1978) (discussing the “common misconception” that the presence of some blacks on grand juries removes the possibility that unconstitutional intentional discrimination has occurred).
. Cf Black Broadcasting Coalition v. FCC, supra note 18, 181 U.S.App.D.C. at 187 n.21, 556 F.2d at 64 n.21.