OPINION OF THE COURT

Petitioner appeals from an order finding it guilty of unlawful discrimination in discharging complainant, Moses Harary. Complainant originally charged that petitioner discriminated against him because of his creed and national origin. He subsequently amended the complaint to add the charge of retaliation. After a hearing, the administrative law judge found that complainant had failed to establish a prima facie case on the grounds stated. He permitted complainant to amend his complaint, however, to allege that he was discharged as the result of. his “race, color and ethnicity” because petitioner had applied an unlawful quota in determining which employees were to be discharged under its 1976 retrenchment policy. Pursuant to that amendment to the complaint, the administrative law judge found petitioner guilty of unlawful discrimination and awarded complainant substantial damages. The decision has been affirmed by the State Division of Human Rights, the Human Rights Appeal Board and the Appellate Division. We find no substantial evidence to support the division’s decision and order, however, and therefore grant the petition and dismiss the complaint.

Complainant is Jewish and American born. In 1972 he was hired as associate dean of administration and management planning by petitioner CUNY-Hostos Community College, a unit of the New York City Board of Higher *73Education.1 He received successive annual appointments to his position until 1975. It is important to this proceeding that he had no teaching duties and that he was not tenured.

In 1975 New York City suffered a severe financial crisis which caused the board of higher education to direct all 19 units under its jurisdiction to drastically reduce expenses (see, generally, Klein v Board of Higher Educ., 434 F Supp 1113). Indeed, conditions were so serious that it appeared for a time that CUNY-Hostos might be closed. A few months later, in July, the board issued a statement of policy to guide the units in retrenchment. Noting that the primary function of the board was education, it directed that economies, insofar as possible, were to be made by staff reductions in administrative positions rather than teaching positions. In August, 1975 petitioner’s president, Cándido de León, called complainant into his office with the dean of administration, told him that he was not satisfied with complainant’s understanding of how a university works (a reference apparently to antagonism between complainant and some faculty members) and told him to start looking for a new position. When complainant asked for a further explanation, de León told him the “reasons are anthropological but I do not want to go into them because you will just use them as a rationalization.” Complainant was told that he could have time to find a new job and in fact he was continued on the payroll until the end of the 1975-1976 school year. On July 28, 1976 he received notice that he would not be reappointed for the coming year.2

On July 21, 1976 President de León, as directed by the board of higher education, promulgated a retrenchment plan. The plan was subsequently revised in August, 1976, but for purposes of this appeal, both plans were the same. Under it, 15 employees, including petitioner, were discharged that fall. Eventually a total of 82 employees were *74eliminated from the college, 33 by attrition and 49 by discharge.

On July 8, 1976 complainant filed charges with the division alleging that petitioner had unlawfully discriminated against him. After hearings extending over two years, the division found that there was “no credible evidence in the record to show that complainant was terminated because of malice based on his being Jewish and American-born”; that in fact President de León had hired complainant in 1972 knowing his religion and national origin. The division also found that under the retrenchment plan some American-born Jews were discharged and others retained and that the evidence in the record did not support complainant’s contention that he was discharged so that he could be replaced by a person representing a different racial or ethnic group. It found petitioner was not guilty of retaliation because complainant had been orally notified of his discharge before he filed his complaint. The division granted complainant relief because it found that he was discharged pursuant to the retrenchment plan and that the plan was “a deliberate effort to perpetuate unlawful quotas based on ethnicity of employees.”

Section 296 (subd 1, par [a]) of the Executive Law provides that it shall be an unlawful discriminatory act to discharge an employee because of age, race, creed, color, national origin, sex, disability, or marital status. Its purpose is to avoid discriminatory preference for any group, minority or majority, in hiring and firing. Neither the statute, the division’s regulations nor any case law cited to us refers to unlawful employment quotas or a prohibition against them, but most people have a sense of the wrong. The words refer to percentages applied to redress or maintain an existing proportion of the work force. The evil of quotas is that they subordinate merit as a basis for hiring and business judgment (which includes consideration of the ability of the employee) as the basis for discharge. Instead, arbitrary percentages are used to maintain a certain ethnic or religious balance. In quota discharges, the wrong consists of singling out employees for discharge, not because of a valid business judgment, but because the employees’ identities as members of a group are necessary *75to maintain a statistical relationship among the employee groups remaining. The discharge acts as a preference for those retained and is a form of unlawful discrimination.

• Complainant’s claim on this appeal is that he was discharged to maintain the racial balance of the groups employed by petitioner. In examining the evidence to support his claim, our review is limited to a consideration of whether the division’s determination in complainant’s favor is supported by substantial evidence on the record. We may not weigh the evidence or reject its choice where the evidence is conflicting and room for a choice exists. When a rational basis for the conclusion approved by the division is found, the judicial function is exhausted (see 300 Gramatan Ave. Assoc. v State Div. of Human Rights, 45 NY2d 176, 181; State Div. of Human Rights v Columbia Univ. in City of N. Y., 39 NY2d 612, 616; Matter of Pace Coll. v Commission on Human Rights of City of N. Y., 38 NY2d 28, 35).

The division, in finding that petitioner implemented an illegal quota has relied on a statistical racial analysis included in President de Leon’s retrenchment plan, his remark to complainant in 1975 that his discharge was for “anthropological reasons”, and his hearing testimony that he was concerned that his plan should not penalize any ethnic group.

The retrenchment plan of 1976 was an ad hoc document devised to meet an extraordinary financial crisis. It called for an extensive reorganization of the college’s facilities and staff intended to achieve annual savings exceeding $1,000,000 in petitioner’s budget. The personnel and budget committee of the college, which included the chairpersons of all departments and voting representatives of the student government, had reviewed it and it was discussed with various deans and program directors and with the business manager of the college. As a part of the plan, six departments were eliminated. Some of the employees of these departments were discharged and some were transferred to other departments or other community colleges. Reductions were made in administrative and service areas as much as possible with preference given to retaining the teaching staff. Of the seven individuals holding positions *76as deans (some of whom had tenure), two were retained, four were reassigned to teaching duties and complainant was discharged. All those retained were required to teach classes. President de Leon’s explanation of the decision to discharge complainant rather than others was that complainant had no tenure and he was the dean the college could lose with least damage to its programs because his educational background did not qualify him to teach any courses offered by the college.3 Moreover, because he had been given notice of his discharge orally a year earlier, it was only natural that he should be included in the retrenchment plans in 1976. The division apparently accepted this explanation because it found no discrimination directed at complainant personally.

The written explanation of the plan recapitulated the jobs and functions changed and showed the estimated savings expected from each. At the end of it, the president noted he had discussed the changes with the college’s affirmative action office and he set forth the effects of the retrenchment on the racial mix of employees both before and after retrenchment and also the racial mix of those discharged.

“The consequences of this ‘Retrenchment Plan’ were discussed with the Affirmative Action Office.

“No significant difference will take place in the ethnic distribution of the faculty as a result of this plan and all of the other changes due to resignations, non-reappointments, etc. The total numbers affected by retrenchment and all other changes are as follows:

*77‘Ethnic Group Number %

White 27 33

Black 24 29

Hispanic 26 32

Asian/Pacific Islander 5 6

Total 82 100%

“The ethnic distribution resulting from all of these changes compared to this spring’s ethnic distribution is as follows:

March, 1976 For September, 1976

‘Ethnic Group Number % Number %

White 101 29.4 74 28.5

Black 95 28.0 71 27.3

Hispanic 135 39.4 109 41.9

Asian/Pacific Islander 11 3.2 6 2.3

Total 342 100.0 260 100.0

“The percentage of women on the entire faculty and staff will go down by 4.4%. This should not be surprising given the retrenchment of the Nursing Department, most of whose instructors are always women.

* * *

“The proportion of women in each ethnic group, however, has remained stable.”

The division has relied heavily on this statistical analysis to establish its finding that petitioner discharged complainant to satisfy an unlawful quota. It seems apparent, however, that the analysis is more a report of the results after formulating the plan of retrenchment than a design for retrenchment, and this is confirmed by the only reference to employment by sex. The chart showed women, also a target for affirmative action, would be affected disproportionately, indicating that the analysis was a review of the effect of the proposal, not a guide for the decisions previously made. Moreover the analysis contained no reference to religious preferences.

*78But more fundamentally, when the statistics are examined they do not establish a prima facie case of discriminatory discharge based upon a quota. In so stating, we recognize that statistics have a place in discrimination cases (see Matter of New York City Bd. of Educ. v Batista, 54 NY2d 379, 383; Matter of Pace Coll. v Commission on Human Rights of City of N. Y., 38 NY2d 28, 36, supra; State Div. of Human Rights v Kilian Mfg. Corp., 35 NY2d 201, 210; and see New York Tr. Auth. v Beazer, 440 US 568, 587; McDonnell Douglas Corp. v Green, 411 US 792, 805). When used with other objectively established evidence, they may permit an inference of unintentional discrimination. Thus, when the results of acknowledged past hiring practices are compared to statistics of the available labor pool and show a consistent underrepresentation of a particular minority group in the labor pool, an inference of discrimination is permissible (see, e.g., State Div. of Human Rights v Kilian Mfg. Corp., supra). A statistical predicate alone, however, will not support a challenge (Matter of Pace Coll. v Commission on Human Rights of City of N. Y., supra, at p 36). There must be proof also of employment practices which warrant the conclusion that discrimination exists (see State Div. of Human Rights v Kilian Mfg. Corp., supra). Moreover, statistics, even when supported by other established objective evidence, will not support an inference of unlawful employment practices unless they rest on a sufficient data base and are adjusted to reflect various factors which may influence the reliability of the numbers, e.g., the history of employment practices, the available labor pool, the rate of turnover, the effect of tenure, and the nature of the employment. In this case, there is the additional factor that the courts traditionally are reluctant to oversee academic judgments. Indeed, we have acknowledged that subjective factors may be considered in making faculty employment decisions (see Matter of Pace Coll. v Commission on Human Rights of City of N. Y., supra, at p 38).

Applying these considerations, we hold that the division erred in accepting these statistics at face value and inferring discrimination based upon them. Importantly, the statistics were based upon a single program undertaken to *79meet an unusual emergency. The record fails to disclose that it was symptomatic of past practices or that it required the discharge of whites other than complainant; the division found that others of similar ethnic background and religion had been retained at the college. Moreover, the figures and positions affected were not final. They were subject to change after review and appeal by affected employees and what changes, if any, were made after these appeals does not appear. Finally, the division failed to take into consideration petitioner’s historical affirmative action hiring practices and its obvious success in maintaining a faculty of diverse backgrounds and cultures. Given this prior history and the scope of retrenchment, some statistical correlation between those retained and those discharged was inevitable under accepted rules of probabilities and militates against an inference based upon the report that complainant’s discharge was motivated by the desire to maintain a quota. In short, while in some cases statistical evidence when corroborated by other objective evidence may support a reasonable inference of discrimination, the statistical evidence here does not.

The division also relied upon the president’s statement in 1975 that complainant’s discharge was for “anthropological reasons”. That statement was ambiguous at best and the division did not find it to be evidence of discrimination when it dismissed complainant’s original charges.

Finally, the division relied upon statements made by President de León during his testimony at the hearing in which he expressed his knowledge of the ethnic makeup of the college and his concern that retrenchment not unduly penalize any group. Neither of these proves discrimination. The board of higher education had an established policy of affirmative action employment and its 1975 statement of policy directed that in making reductions in staff due regard should be given it.4 Manifestly, petitioner and its officers were aware of this policy and of the antidiscrimination laws. The college maintained an affirmative action office which, among other duties, compiled ethnic and *80gender data about employees and reported it annually to the United States Department of Labor to qualify petitioner for Federal assistance (see 41 CFR part 60-2 et seq.). Thus, the president’s testimony that petitioner wished to avoid penalizing any minority by this retrenchment, his statement in the plan that he had discussed the discharges with the college’s affirmative action office and his inclusion of a statistical analysis which demonstrated that as a result of the plan there would be little change in the racial balance of the university’s employees, far from being proof per se of unlawful discrimination, was entirely consistent with his sensitivity to discrimination as the law and university policy required him to be. It should not have been parlayed into a new and different claim of discrimination by the administrative law judge after the evidentiary hearing was closed. It is noteworthy that this was not the only discrimination claim filed against petitioner by disappointed employees of various racial backgrounds after the 1976 retrenchment and it is not difficult to imagine the contentions that would have been raised if a disproportionate number in one or another racial group had been discharged. Petitioner was entitled to consider that aspect of the problem also.

Complainant, and Judge Fuchsberg in dissent, have relied upon evidence produced in support of the original charges that there was prejudice or bias against complainant personally to support this charge. The division rejected that evidence. On the sustained charge of discrimination based upon quota discharges, the evidence in the record established little more than the results of the 1976 retrenchment plan and the fact that petitioner was satisfied with those results. That evidence does not warrant a conclusion of unlawful discrimination.

Accordingly, the order of the Appellate Division should be reversed and the petition should be granted.

(dissenting). In a proceeding brought pursuant to section 298 of the Executive Law (the Human Rights Law) and CPLR article 78, the underlying issue is whether there was substantial evidence to support a State Human Rights Appeal Board order affirming an order of the State Division of Human Rights, which had adjudged that the complainant, Moses Harary, had been discriminatorily discharged from his employment by reason of his employer CUNY-Hostos Community College’s resort to a formula, or quota, under which “race, color and ethnicity” impermissibly were the criteria for identifying the employees who, in the implementation of a budgetary retrenchment plan, were to be singled out for termination. By unanimously confirming “in all respects”, the Appellate Division upheld the board’s determination.

On this further appeal, Hostos and its president, Cándido de León, in effect repeat their earlier arguments that the division’s determination is grounded on no more than *82“conjecture and surmise” and therefore “does not meet the substantial evidence test”.1 Because the ensuing review of the record taken as a whole discloses the contrary, this position should be rejected.

But first I note the two guidelines by which it must be measured:

The first of these is the overriding fact that the Human Rights Law, which governs the conduct implicated in this case, was enacted, among other things, to assure that, consistent with fundamental tenets of our society, every individual receive an equal opportunity “to enjoy a * * * productive life” and “to participate fully in the economic * * * life of the state”. As relevant here, to implement these ideals, the statute flatly declares it an unlawful practice “[f]or an employer * * * because of * * * race, creed, color [or] national origin * * * of any individual * * * to discharge * * * or to discriminate against such individual”. (Executive Law, § 290, subd 3; § 296, subd 1, par [a].) And section 300 of this remedial legislation, recognizing that discrimination often travels under false colors, compels us to see that it is “construed liberally”.

The second guideline, also noted by the majority, is the limited range of judicial review of an administrative determination, especially one arrived at, as here, by an agency which of necessity has been called upon to develop and apply a highly sophisticated, sensitive and specialized ex*83pertise in treating with the societal problem committed to its care (Matter of New York Roadrunners Club v State Div. of Human Rights, 55 NY2d 122, 126; cf. Matter of Burger King v State Tax Comm., 51 NY2d 614, 621; Kurcsics v Merchants Mut. Ins. Co., 49 NY2d 451, 459). This means that the “inference-making function, as it is exercised at the evidentiary or fact-finding level, is exclusively that of the administrative agency” (State Div. of Human Rights v Wagner, 39 NY2d 865, 866). Or, as stated on another occasion, “neither the weight which might be accorded nor the choice which might be made by a court are germane upon an analysis for the presence of substantial evidence” (300 Gramatan Ave. Assoc. v State Div. of Human Rights, 45 NY2d 176, 180). In like vein, realistically dubbing this principle the “doctrine of administrative finality”, the United States Supreme Court put it that, “[s]o long as there, is warrant in the record for the judgment of the expert body it must stand”, the judicial function being “ ‘exhausted when there is found to be a rational basis for the conclusions approved by the administrative body’ ” (Rochester Tel. Corp. v United States, 307 US 125, 139-140, 145-146; see, also, Davis, Administrative Law, 1982 Supp, § 29.00-1, p 520).

Mindful of these precepts, we are informed by the record here that Harary, who is white, Jewish and a native of the United States, was first appointed, with the rank of assistant professor, to his nontenured post as Hostos’ associate dean of administration and management planning as of September, 1972. Prior to his ultimate termination, effective June 17,1976, though Harary was assigned to what is acknowledged to have been the arduous and toe-stepping responsibility of reorganizing what had been the college’s morass-like management system, his immediate superior, Dean of Administration Theodore Foxworthy, was to testify that he never received a complaint of any substance concerning Harary’s performance. Indeed, it is undisputed that Foxworthy, who at the public hearing described Harary as highly competent and thoroughly knowledgeable in his field, on at least two occasions had backed this appraisal by recommending him for promotion. And, corrobo*84ratively, de León, curiously in view of the inconsistent position he took at the hearing, not only had unreservedly recommended Harary’s reappointment for each of the three successive years in which he had the opportunity to do so but, only months before initiating the challenged discharge, had put Harary’s name in for a full professorship.

It was against this background that, in August, 1975, when his consecutive appointments in this new college had made him the second most senior member of his department, Harary was summoned to the office of de León. There, in the presence of Foxworthy, de León announced for the first time that he was not satisfied with Harary’s performance and suggested that he thereafter “devote full time looking for a job”. As Harary and Foxworthy described it, when the former then asked why de León was taking this action, the latter responded by saying the reasons were “anthropological”2 and, when pressed for further explanation, would not give any details.

True, de León later was to deny that he had used the term “anthropological”, claiming instead that he had told Harary only that he was “not satisfied with his failure to communicate after repeated discussions with his lack of understanding of how our university works”. That there were, therefore, two versions on this point, of course but emphasizes that it was for the agency, and for it alone, to decide, as a simple question of fact, whether to accept the findings of the law judge who, after observing and hearing the witnesses, concluded that the account rendered by de León was the less credible, especially because, as the Human Rights Division’s decision observed, “if complainant were incompetent, he could have been discharged earlier”.

*85Other parts of the record also gave legitimate reinforcement to the division’s express finding that Harary was not terminated for incompetence. A significant factor, for instance, was the strange coincidence that, as the division in effect determined, de Leon’s resort to a pretextual complaint followed almost immediately on the heels of a so-called statement of policy then newly adopted by the board of higher education, then the ultimate governing body of Hostos and the other municipal colleges of the City of New York. The statement, to meet a dire financial emergency, required each of the colleges to institute a drastic retrenchment plan effecting savings in personnel and programs. From this sequence of events, it was inferable that de León had been thinking of the criteria by which he would have preferred, if left to his own devices, to eliminate personnel.

The board thereafter was to issue guidelines which, as the division’s opinion points out, “provided that personnel in the main, be discontinued in inverse order of length of full-time continuous service; also, non-tenured personnel were to be discontinued first, then certificated personnel, and then tenured personnel. In the case of non-tenured personnel, seniority could be disregarded for ‘special educational reasons’. Since complainant did not hold a teaching position, this latter exception could not be applied to him. Thus, under the guidelines laid down by the Board of Higher Education, seniority was to govern”.3

It is noteworthy too, or so the division could see it, that, when the guidelines came down, though Harary then had not yet been formally severed and at worst was in limbo,4 de León did not restore him to the status to which he was *86entitled as second senior employee in his category. Moreover, although the board’s announced policy, as the majority today acknowledges, was to release administrators rather than teachers, it was also its policy, when possible, to reassign qualified administrators to teaching positions. Despite the fact that Harary could have been viewed as qualified to teach at Hostos, his file indicating that he not only had extensive knowledge of accounting and engineering but had taught economics, public speaking and operational planning, he was the only retrenched dean not offered a teaching position. In short, even when seniority or ability to teach each became a governing consideration, de León excluded Harary from its benefits. And all this though the record contains direct evidence that most of Harary’s administrative functions were not eliminated but turned over to a newly engaged employee who, like de León, was of Hispanic background.

All of this and more the division could weigh in light of the nature of the retrenchment plans de León himself then promulgated. Impossible to ignore, for instance, is the reality that the 82 persons, including Harary by name, whose release they projected, consisted of individuals whose ethnic distribution for all practical purposes were well-nigh perfect matches of the pre-existent college staff population, divided by the de León plans into what essentially were three groups classified as “White”, “Black” and “Hispanic”. To be precise, 101 “White” persons comprised 29.4% of staff before the retrenchment while the 74 who remained after the plans had been put in place came to an almost identical 28.5%. Likewise, 95 “Black” persons comprised 28% of staff before the retrenchment, while the 71 who remained when the plans had taken their toll amounted to 27.3%. In presenting the same picture, the preretrenchment number of “Hispanic” employees was 135, or 39.4%, and the postretrenchment number was 109, or 41.9%.

*87The revealing nature of these statistics need not have escaped the division’s notice (State Div. of Human Rights v Kilian Mfg. Corp., 35 NY2d 201, 210 [statistics in such cases “are valuable and often demonstrate more than the testimony of witnesses, ‘and they should be given proper effect’”]). Granted that “some statistical correlation between those retained and those discharged was inevitable” (majority opn, at p 79), the mathematical probability against the nearly perfect correlation of a combination of three different groups is so great that it all but conclusively rebuts the possibility that it was the product of chance.

But the plans’ explicitness left nothing to the imagination. They expressly advised that their purpose was to make sure that “no significant difference will take place in the ethnic distribution of the faculty”.5 Thus, although, as the majority points out, the fiscal emergency which prompted the plans presented the possibility that, willynilly, aberrant conditions might result, the de León plan was meticulous in guaranteeing that, whatever and however the pre-existing ethnic distribution had come about, it was to be treated as an immutable quota. The more is this so since, as I am glad to see we all agree, discrimination in employment, whether it be achieved by a hiring quota under which its targets, if ever reached, are the “last in”, or whether it be achieved by a firing quota under which targets are the “first out”, especially under a statute intended to be liberally construed to guarantee equal opportunity to enjoy a “productive life”, is equally invidious.

Needless to say, the figures cannot be ignored. So the majority, adopting Hostos’ and de Leon’s rationale, accepts them as “more a report of the results after formulating the plan of retrenchment than a design for retrenchment” (emphasis in original), language which, reasonably read, fairly concedes that it would be rational for the division to have interpreted the matter either way. And, as to the support the majority sees in de Leon’s testimony that plans were discussed with Hostos’ affirmative action office and *88others, suffice it that the plans, once found to be discriminatory, would be no less so because they were the handiwork of many rather than that of one or few, or because they bore the imprimatur of an official college agency.

This is not to assume that the preretrenchment ethnic distribution of the Hostos faculty was in fact the product of discrimination. An academic institution, to foster intellectual creativity and achievement, is to be accorded wide latitude in determining, among other things, who will teach its students (Matter of Pace Coll. v Commission on Human Rights of City of N. Y., 38 NY2d 28, 38). So, when an institution believes the promotion of ethnic pluralism will be beneficial educationally, ethnic considerations, if not the sole determinants for selection, may, along with other facts reasonably related to the employment, such as training, experience, education, pedagogical talent or the like, be taken into account (University of Cal. Regents v Bakke, 438 US 265, 317 [opn of Powell, J.]; Matter of Fullilove v Beame, 48 NY2d 376, 379-380 [dissenting opn];6 to the same effect, see Education Law, § 6201, subd 3 [“equal access and opportunity”]).

Here, the preretrenchment mix was not challenged and so no evidence was produced by either side to prove whether it was or was not the product of discrimination. For all we know, then, the pre-existing racial balance conceivably may have come about either through chance or through a salutary and constructive affirmative action effort which, adhering to the dictates of Bakke, stopped short of disregarding the basic principle of merit selection, or through a combination of both. But, however unchallengeable the pre-existing ethnic distribution, to automatically and arbitrarily engage or discharge persons merely to *89maintain the same ratio runs afoul of the spirit and letter of the law.

The division’s determination that the prescribed de León formula for the severance of Hostos’ employees was discriminatory having, therefore, well met the requirements of the substantial evidence rule, I now turn to the proof of causal connection between the quota and Harary’s termination.

For this purpose, it is immediately noteworthy that the division, having decided to give the benefit of the doubt to de León on the charge that he had acted on considerations of creed and ethnicity as such in removing Harary, carefully distinguished such a violation from one “constitut[ing] * * * a deliberate effort to perpetuate unlawful quotas based on the ethnicity of employees”.

Available to be weighed in the balance on the latter theory were a number of what the administrative agency could treat as telling factors. One of these is that, although Harary was second in seniority among the deans, he was not one of the two spared retrenchment. In this connection, while de Leon’s hypersensitive attitude towards Harary’s creed was not enough to convince the division that this was the per se reason for his discharge, its circumstantial significance need not have been ignored in the context of the preferment of at least one of the retained deans in disregard of Harary’s seniority. In addition, and of prime significance, could be the fact that Harary, as mentioned earlier, was the only retrenched dean not to be offered a teaching position notwithstanding the board of higher education’s policy recommendations to the contrary.

In sum, review of the record discloses more than sufficient evidence to compel us to sustain the division’s conclusion that Harary was discharged in furtherance of Hostos’ and de Leon’s enforcement of the impermissible criteria at the heart of the retrenchment plan.

Accordingly, I vote to affirm the order of the Appellate Division.7

*90Chief Judge Cooke and Judges Jasen, Jones and Meyer concur with Judge Simons; Judges Wachtler and Fuchs-berg dissent and vote to affirm in separate dissenting opinions.

Order reversed, with costs, and petition granted.

(dissenting). I believe the essence of the division’s holding in this case is that the city university carried affirmative action too far when it employed the practice in firing employees to maintain ethnic balance. There is substantial evidence to support the finding that the city university was actively committed to an affirma*81tive action program which thus far had been properly limited to hiring. The statistics show that after the retrenchment the ethnic balance remained the same. This, of course, could be the natural result of a more or less random across the board reduction in which case the practice would be legitimate. But concededly conflicting inferences could be drawn and it is the division’s province to determine which is the proper one. Indeed the president’s own statement, that in formulating the retrenchment plan one of his goals or “concerns” was that “no group be penalized any more than the other and that the affirmative efforts of the university not be reversed by the retrenchment” could be considered a type of concession.

We have repeatedly noted that the scope of judicial review in these cases is extremely narrow (see, e.g., Matter of Imperial Diner v State Human Rights Appeal Bd., 52 NY2d 72). The question is not whether we find the proof of discrimination convincing, but whether the division could do so. Because the evidence before the division in this case would at least support such an inference the court should not disturb the division’s determination.