Defendant Nona J. Switala is the superintendent of defendant Taycheedah Correctional Institution (TCI), the only women’s maximum security prison in Wisconsin. TCI is operated by the Wisconsin Department of Health and Social Services (DHSS), also a defendant here. Ms. Switala, an experienced prison administrator, determined that the rehabilitation of TCI’s inmates would be enhanced by employing only female correctional officers in TCI’s living units. The plaintiffs Raymond Torres, Franklin Utz, and Gerald Schmit were reassigned to other positions at TCI, with no loss in pay, because of this plan. Unhappy with this reassignment, the plaintiffs brought this action alleging sex discrimination by the defendants in violation of Title VII of the Civil Rights Act of 1964. 42 U.S.C. §§ 2000e to 2000e-17. The defendants responded that sex was a bona fide occupational qualification (BFOQ), 42 U.S.C. § 2000e-2(e)(l), for the positions formerly held by the plaintiffs. The district court determined that the defendants had not established a valid BFOQ because the defendants had not offered “objective- evidence, either from empirical studies or otherwise,” proving that the BFOQ would further inmate rehabilitation, and because prison security and inmate privacy were not materially advanced by the plan. We reverse and remand.
I
Facts
TCI has three buildings for housing inmates. Each building has three residence floors, and the inmates live in single, double, or multiple occupancy rooms. The rooms are not cells with bars, but are more akin to college dormitory rooms. Each room has a solid door with a clear glass window at eye level that is approximately four inches by six inches. The rooms have one bed per inmate, a desk, chair, light, toilet, and wash basin. In two buildings, privacy curtains have been installed around the toilets. When an inmate is behind the curtain, only her feet are visible. At the time of trial, TCI had plans to install privacy curtains in the other residence building soon.
From 6 a.m. until 9 p.m., inmates may place “privacy cards” inside the door windows so that they can use the toilet or change their clothes without being observed. TCI’s rules allow the privacy cards to be up for only ten minutes per day per inmate. In multiple inmate rooms, this means that the cards can be up permissibly for as much as thirty or forty minutes. However, the testimony offered at trial suggests that correctional officers are not able to keep careful track of the time that a privacy card has been in place, and that inmates sometimes leave their cards up for more than ten minutes per inmate. On one occasion, two inmates used their privacy card to facilitate an escape, and on another occasion an inmate beat up her roommate while the privacy card covered the window.
From 9 p.m. until 6 a.m., inmates may not place their privacy card on the inside of their window, but they may place their card outside the window all night long in order to prevent light from entering the room. Correctional officers then lift up the card for body counts and inspections. TCI’s *1525rules require that the officers conduct a body count each day at 7:30 a.m., 12:30 p.m., 5:30 p.m., 9:30 p.m., and once each hour between 10 p.m. and 6 a.m. Inmates know this schedule. TCI provides appropriate sleepwear for the inmates but they are not required to wear it. The guards are required to see the inmates’ skin or hair during nighttime body counts.
Each floor of the residence buildings has a shower room. Inmates must sign up with their floor officer before taking a shower, and they are required to wear some sort of clothing when walking to the shower room. Testimony at trial suggested that guards normally do not enter the shower rooms when occupied. The doors to the shower rooms are solid, although some contain windows that have been rendered opaque. The shower rooms have one to three shower stalls, one to three toilets, and some have one or more bathtubs. The showers and toilets have privacy curtains or privacy doors. Only one inmate may occupy the shower room at a time, except that roommates may enter together. TCI allows each inmate fifteen minutes in the shower room.
The Wisconsin Administrative Code allows prison officials to perform four types of inmate searches. Wis.Admin.Code § HSS 306.16 (1987). Correctional officers may perform pat searches at any time. During a pat search, inmates empty their pockets and the officer runs his or her hands over the inmates’ entire body. § HSS 306.16(a). The custom at TCI is that only female guards perform pat searches. The second type of allowable search is the strip search. Strip searches must be authorized by a supervisor and performed in private by an officer of the same sex, except during emergencies. § HSS 306.16(b). The third type of search is the body cavity search. Body cavity searches are only performed by medical personnel in special circumstances. § HSS 306.16(c). The fourth permissible search is a body content search, such as urinalysis or blood analysis. § HSS 306.16(d). Under this regulation, body content searches are permitted only in extreme circumstances. In addition, correctional officers are expected to search occasionally inmates’ rooms and the shower rooms when they are unoccupied.
Ms. Switala became superintendent of TCI in 1978. She previously had worked for three years as treatment director at TCI and for eight years as a probation and parole agent with the Wisconsin Division of Corrections. Ms. Switala, her superiors at the DHSS, and her personnel at TCI, soon began discussions regarding TCI’s staffing needs. It ultimately was decided, principally by Ms. Switala, that certain positions at TCI should be staffed only by female correctional officers. It is clear from Ms. Switala’s testimony at trial that the principal reason for this decision was her concern for inmate rehabilitation and security. TCI’s administrators then advised all correctional officers in 1980 that a BFOQ program would be implemented gradually in the next two years.
TCI has three different ranks for its correctional officers. The lowest position is a correctional officer 1 (CO-1), followed by correctional officer 2 (CO-2) and correctional officer 3 (CO-3). The CO-3 in charge of a living unit is a “sergeant.” The positions to be affected by the BFOQ plan were nineteen of the twenty-seven correctional officer positions in the living units, including all of the CO-3 posts in the living units. As a result of the plan, only three CO-3 positions at TCI would be open to men. The three plaintiffs, Mr. Torres, Mr. Utz, and Mr. Schmit, were all CO-3’s prior to implementation of the BFOQ plan. Because of the limited number of available CO-3 positions, the three plaintiffs were required to accept CO-2 positions, although this demotion resulted in no loss of pay. The plaintiffs presently work under female CO-3’s who have less seniority and experience.
II
The District Court Opinion
The district court first determined that the BFOQ is a narrow exception to Title VII’s prohibition of discrimination in employment. The court noted that “adminis*1526trative convenience is insufficient to justify a BFOQ exception,” and that discrimination based on sex is permissible only “ ‘when the essence of the business operation would be undermined.’ ” Torres v. Wisconsin Dep’t of Health and Social Servs., 639 F.Supp. 271, 277 (E.D.Wis.1986) (quoting Diaz v. Pan Am. World Airways, 442 F.2d 385, 388 (5th Cir.), cert. denied, 404 U.S. 950, 92 S.Ct. 275, 30 L.Ed.2d 267 (1971)). The burden of establishing a BFOQ is a “heavy one,” according to the court, and is “justified only in rare, appropriate circumstances.” Id. at 278.
The court then analyzed the BFOQ under each of the defendants’ proposed justifications: security, rehabilitation, and inmate privacy. Regarding security, the court found that male correctional officers had filled all of the BFOQ positions from 1975 until 1982 “without any contention that TCI suffered from a lack of security.” Id. Moreover, the court noted that TCI continued to allow inmates to use privacy cards even after the BFOQ program had been installed, thus rebutting the argument that the presence of male guards reduced observation of the inmates. Although the court acknowledged that staffing problems at TCI might have been reduced by the BFOQ, since all guards then could perform all tasks, the court found that this justification merely constituted administrative convenience and not a genuine justification for sex discrimination. Thus, since the defendants had not presented evidence of any decline in inmate escapes or violence as a result of the BFOQ, the court found “that defendants have failed to justify the Plan based on security reasons.” Id. at 279.
The court next analyzed the BFOQ under the defendants’ rehabilitation justification. The court acknowledged that the defendants had “presented various witnesses familiar with the field of corrections who testified in support of this theory.” Id. at 280. However, the court said that the plaintiffs also had presented witnesses who testified that male guards can enhance rehabilitation by “normaliz[ing]” the prison environment. Id. at 280 & n. 9. Thus, the court determined that the defendants had offered only a “theory of rehabilitation as a justification for the BFOQ Plan. They offered no objective evidence, either from empirical studies or otherwise, displaying the validity of their theory.” Id. at 280 (emphasis in original). Having concluded that the rehabilitation theory was unproven, the court refused to justify the BFOQ based on the theory.
The court concluded its analysis by reviewing the BFOQ under the justification of inmate privacy. The court ruled that inmates do have a constitutional right to be free from unwarranted privacy intrusions, but that this right necessarily was limited. Id. at 280-81. The court then determined that any intrusions of privacy occasioned by male guards were not unwarranted nor unreasonable because of other precautions taken by the prison administration. For example, the court noted that toilet facilities have privacy curtains, that correctional officers do not observe inmates while they are showering, and that guards do not see inmates unclothed except during an emergency or when the inmate voluntarily allows it to happen. Therefore, the court found that privacy concerns did not justify the defendants’ BFOQ plan.
Because the defendants had overtly discriminated against the plaintiffs on account of their sex, and since, in its view, there was no valid BFOQ, the court found that the defendants had violated Title VII. However, the court did not find that the plaintiffs were entitled to monetary damages. The court determined that the plaintiffs’ reassignment had not caused any reduction in salary or overtime hours, and that therefore the plaintiffs had not established any monetary damages. For that reason, the plaintiffs’ only remedy was an order abolishing the BFOQ plan and reinstating the plaintiffs in their prior assignments.
Ill
Analysis
A. Title VII and the BFOQ
. We begin by reiterating several basic propositions, long established in the precedent, that must guide our decision today. “Title VII prohibits employment discrimination, which is ‘one of the most deplorable forms of discrimination known to our society, for it deals not with just an individual’s *1527sharing in the “outer benefits” of being an American citizen, but rather the ability to provide decently for one’s family in a job or profession for which he qualifies or chooses.’ ” Hardin v. Stynchcomb, 691 F.2d 1364, 1369 (11th Cir.1982) (quoting Culpepper v. Reynolds Metals Co., 421 F.2d 888, 891 (5th Cir.1970)); see also Rowe v. General Motors Corp., 457 F.2d 348, 354 (5th Cir.1972). The statute forbids both overt discrimination and practices that are “ ‘fair in form, but discriminatory in operation.’ ” Pullman-Standard v. Swint, 456 U.S. 273, 276, 102 S.Ct. 1781, 1784, 72 L.Ed.2d 66 (1982) (quoting Griggs v. Duke Power Co., 401 U.S. 424, 431, 91 S.Ct. 849, 853, 28 L.Ed.2d 158 (1971)).
This case involves overt sex discrimination. It presents one issue — whether the district court correctly determined that sex was not a BFOQ for correctional officer positions at TCI. Section 703(e) of Title VII only permits classifications based on sex “where ... sex ... is a bona fide occupational qualification reasonably necessary to the normal operation of that particular business or enterprise.” 42 U.S.C. § 2000e-2(e)(l) (emphasis supplied). It is universally recognized that this exception to Title VII was “meant to be an extremely narrow exception to the general prohibition of discrimination....” Dothard v. Rawlinson, 433 U.S. 321, 334, 97 S.Ct. 2720, 2729, 55 L.Ed.2d 786 (1977); see also Pime v. Loyola Univ., 803 F.2d 351, 356 (7th Cir.1986) (Posner, J., concurring). As the Supreme Court recognized in Dothard, this narrow exception “has been variously formulated.” 433 U.S. at 333, 97 S.Ct. at 2729. “ ‘[Discrimination based on sex is valid only when the essence of the business operation would be undermined by not hiring members of one sex exclusively.’ ” Id. (quoting Diaz v. Pan Am. World Airways, 442 F.2d 385, 388 (5th Cir.), cert. denied, 404 U.S. 950, 92 S.Ct. 275, 30 L.Ed.2d 267 (1971) (emphasis in original)).1 As an alternative formulation, noted the Supreme Court, it has been held that “an employer could rely on the BFOQ exception only by proving ‘that he had reasonable cause to believe, that is, a factual basis for believing, that all or substantially all women would be unable to perform safely and efficiently the duties of the job involved.’ ” Id. (quoting Weeks v. Southern Bell Tel. & Tel. Co., 408 F.2d 228, 235 (5th Cir.1969)); see also Phillips v. Martin Marietta Corp., 400 U.S. 542, 544, 91 S.Ct. 496, 498, 27 L.Ed.2d 613 (1971).
It is also well established that a BFOQ may not be based on “stereotyped characterizations of the sexes.” Dothard, 433 U.S. at 333, 97 S.Ct. at 2729. “Myths and purely habitual assumptions about a woman’s [or a man’s] inability to perform certain kinds of work are no longer acceptable reasons for refusing to employ qualified individuals, or for paying them less.” City of Los Angeles Dep’t of Water & Power v. Manhart, 435 U.S. 702, 707, 98 S.Ct. 1370, 1375, 55 L.Ed.2d 657 (1978). On the other hand, there are real as well as fictional differences between men and women. Id. For instance, the Supreme Court has never hesitated to recognize sex-based differences, particularly in cases involving physiology, marriage, childbirth, or sexuality. See Michael M. v. Superior Court of Sonoma County, 450 U.S. 464, 469, 101 S.Ct. 1200, 1204, 67 L.Ed.2d 437 (1981) (“[T]his Court has consistently upheld statutes where the gender classification is not invidious, but rather realistically reflects the fact that the sexes are not similarly situated in certain circumstances.”); id. at 481, 101 S.Ct. at 1210-11 (Blackmun, J., concurring) (“The Constitution surely does not require a State to pretend that demonstrable differences between men' and women do not really exist.”). See generally Parham v. Hughes, 441 U.S. 347, 354, 99 S.Ct. 1742, 1747, 60 L.Ed.2d 269 (1979) (opinion of Stewart, J.) (“In cases where men and women are not similarly situated, however, and a statutory classification is realistically based upon the differences in their situations, this Court has upheld its validity.”); *1528Schlesinger v. Ballard, 419 U.S. 498, 508, 95 S.Ct. 572, 42 L.Ed.2d 610 (1975) (“[T]he different treatment of men and women ... reflects, not archaic and overbroad generalizations, but, instead, the demonstrable fact that male and female line officers in the Navy are not similarly situated with respect to opportunities for professional service.” (emphasis in original)). This same principle has been recognized in the Title VII area. See, e.g., Backus v. Baptist Medical Center, 510 F.Supp. 1191, 1195 (E.D.Ark.1981), vacated because of mootness, 671 F.2d 1100 (8th Cir.1982) (recognizing the need to have female registered nurses care for obstetrical patients); see also I A. Larson & L. Larson, Employment Discrimination — Sex § 14.30 (1987) (“[GJiving respect to deep-seated feelings of personal privacy involving one’s own genital area is quite a different matter from catering to the desire of some male airline passengers to have ... an attractive stewardess.”).
Nevertheless, while recognizing that sex-based differences may justify a limited number of distinctions between men and women, we must discipline our inquiry to ensure that our tolerance for such distinctions is not widened artificially by — as the district court aptly put it — our “own culturally induced proclivities.” Torres, 639 F.Supp. at 278. Nor, of course, can we tolerate the same preconceptions or predilections on the part of employers. Rather, we must ask whether, given the reasonable objectives of the employer, the very womanhood or very manhood of the employee undermines his or her capacity to perform a job satisfactorily. Dothard, 433 U.S. at 336, 97 S.Ct. at 2730.
One of the most important means by which an appellate court disciplines its inquiry is by keeping in mind the appropriate standard of review. The district court’s “[fjindings of fact shall not be set aside unless clearly erroneous, and due regard shall be given to the opportunity of the trial court to judge of the credibility of the witnesses.” Fed.R.Civ.P. 52(a). In Anderson v. City of Bessemer City, the Court said that a finding is clearly erroneous when “ ‘the reviewing court on the entire evidence is left with the definite and firm conviction that a mistake has been committed.’ ” 470 U.S. 564, 573, 105 S.Ct. 1504, 1511, 84 L.Ed.2d 518 (1985) (quoting United States v. United States Gypsum Co., 333 U.S. 364, 395, 68 S.Ct. 525, 542, 92 L.Ed. 746 (1948)). “Where there are two permissible views of the evidence, the fact-finder’s choice between them cannot be clearly erroneous.” Id. at 574, 105 S.Ct. at 1511. Of course, this deferential standard of review presumes that the district court has applied correct legal standards—decisions of law are reviewed de novo.
B. The “Business” of the Employer
With the foregoing propositions reaffirmed, we turn to the case before us. In our view, the decision of the district court that the defendants’ BFOQ plan cannot be justified by concerns for prison security or for the basic privacy rights of the inmates is correct in law and fact. Thus, we turn to the only remaining question: whether the district court properly rejected defendants’ contention that the BFOQ was necessary to further their goal of inmate rehabilitation.
The validity of a BFOQ can only be ascertained when it is assessed in relationship to the business of the employer. Our first step, therefore, must be to come to an understanding of the employer’s business —its mission and the methodologies necessary to fulfill that mission. In accomplishing this task, we cannot deal in generalities. See Trans World Airlines v. Thurston, 469 U.S. 111, 122, 105 S.Ct. 613, 622, 83 L.Ed.2d 523 (1985). Rather, we must focus on the “particular business” of the employer in which the protected employee worked. Id. Oftentimes, this task requires that a court recognize factors that make a particular operation of an employer unique or at least substantially different from other operations in the same general business or profession. See Pime v. Loyola Univ., 803 F.2d 351, 353-54 (7th Cir.1986) (upholding the maintenance of a Jesuit “presence” as “important to the successful operation of the University,” when there was evidence that it was “significant *1529to the educational tradition and character of the institution that students be assured a degree of contact with teachers who have received the training and accepted the obligations which are essential to membership in the Society of Jesus”).
Here, of course, the broadest description of the “business” of the defendants is to say that they are in the business of governance at the state level. This general description, standing alone, gives them no special license with respect to Title VII. In Dothard v. Rawlinson, 433 U.S. 321, 332 n. 14, 97 S.Ct. 2720, 2728, 53 L.Ed.2d 786 (1977), the Supreme Court said that “Congress expressly indicated the intent that the same Title VII principles be applied to governmental and private employers alike.” A more precise definition of the “business” of the defendants is to recognize that they are in the business of administering a penal institution. Few tasks are more challenging. As the Supreme Court has explained:
Prison administrators are responsible for maintaining internal order and discipline, for securing their institutions against unauthorized access or escape, and for rehabilitating, to the extent that human nature and inadequate resources allow, the inmates placed in their custody. The Herculean obstacles to effective discharge of these duties are too apparent to warrant explication. Suffice it to say that the problems of prisons in America are complex and intractable....
Procunier v. Martinez, 416 U.S. 396, 404-05, 94 S.Ct. 1800, 1807, 40 L.Ed.2d 224 (1974). Those who assume the responsibility of administering any prison must grapple with the “perplexing sociological problems of how best to achieve the goals of the penal function in the criminal justice system: to punish justly, to deter future crime, and to return imprisoned persons to society with an improved chance of being useful, law-abiding citizens.” Rhodes v. Chapman, 452 U.S. 337, 352, 101 S.Ct. 2392, 2402, 69 L.Ed.2d 59 (1981); see also Bell v. Wolfish, 441 U.S. 520, 547-48, 99 S.Ct. 1861, 2402, 60 L.Ed.2d 447 (1979). See generally Jones v. North Carolina Prisoners’ Labor Union, 433 U.S. 119, 127-29, 97 S.Ct. 2532, 2538-40, 53 L.Ed.2d 629 (1977). In fulfilling these responsibilities, prison administrators always have been expected to innovate and experiment. See Turner v. Safley, 482 U.S. 78, 107 S.Ct. 2254, 2262, 96 L.Ed.2d 64 (1987) (prison administrators must be allowed “to adopt innovative solutions to the intractable problems of prison administration”). Unless prison administrators try new approaches, the “intractable problems” will remain and the lot of the incarcerated individual will not improve. Indeed, it probably will deteriorate.
This general description of the task of prison administrators is still too general to permit us to assess accurately the claims of the parties. In Dothard, the Supreme Court focused not on maximum security institutions as a generic class but rather on the environment then existing in Alabama’s penitentiaries — “a peculiarly inhospitable one for human beings of whatever sex.” 433 U.S. at 334, 97 S.Ct. at 2729. In conformity to the mandate of the Supreme Court in Dothard, we must therefore refine further our focus. The defendants here are charged with the administration of a distinct type of penal institution — a women’s maximum security facility. As the district court and one of the plaintiffs’ own witnesses quite candidly acknowledged,2 the same historical and empirical evidence that might guide the administrator of a similar institution for males simply is not available with respect to this environment. Therefore, the administrators of TCI were obliged, to a greater degree than their counterparts in male institutions, to innovate in achieving one of the tasks mandated by the Wisconsin legislature — rehabilita*1530tion. The defendants’ “business” explicitly included — by legislative mandate — the task of rehabilitation. See Wis.Admin.Code §§ HSS 303.01(3)(c), HSS 306.01 (1987). One of the plaintiffs’ expert witnesses testified that this goal was not considered, in modern penological theory, a high priority or, by some, even a realistic goal. Tr. 6 at 160 (testimony of Dr. David Kalinich). Wisconsin has made a contrary decision and rehabilitation, no matter how elusive, must be attempted. Wisconsin’s goal is clearly a reasonable course, see Rhodes, 452 U.S. at 352, 101 S.Ct. at 2402, and we must accept it as part of the “business” of the defendants.
C. Rehabilitation: The TCI Approach
Ms. Switala, the superintendent of TCI, made a professional judgment that giving women prisoners a living environment free from the presence of males in a position of authority was necessary to foster the goal of rehabilitation. This decision was based on Ms. Switala’s professional expertise and on her interviews and daily contact with female prisoners. Tr. 10 at 110-11. She also based her decision on the fact that a high percentage of female inmates has been physically and sexually abused by males. Indeed, she noted that sixty percent of TCI’s inmates have been so abused. Tr. 9 at 39.
1.
There can be no question that the proposed BFOQ is directly related to the “essence” of the “business” — the rehabilitation of females incarcerated in a maximum security institution. See Dothard, 433 U.S. at 333, 97 S.Ct. at 2728-29 (quoting Diaz v. Pan Am. World Airways, 442 F.2d 385, 388 (5th Cir.), cert. denied, 404 U.S. 950, 92 S.Ct. 275, 30 L.Ed.2d 267 (1971)). As we already have noted, the Wisconsin legislature has mandated that rehabilitation is one of the objectives of that state’s prison system. See Wis.Admin.Code §§ HSS 303.-01(3)(c), HSS 306.01 (1987). Therefore, we stress that this is not a case where the interest of the plaintiffs in their continued employment simply conflicts with the basic privacy rights of the inmates. Here, the interest of the plaintiffs conflicts with the task of the administrators to rehabilitate the inmates for whom they are responsible. Forts v. Ward, 621 F.2d 1210 (2d Cir.1980), involved a female penitentiary. However, in that case, the suit was brought by prisoners against prison administrators. The prisoners sought to enjoin the administrators from assigning male guards to duties in the housing units of the prison. Forts therefore involved a straightforward conflict between the prisoner’s right to privacy and the guards’ rights, secured by Title VII, to employment security. Rehabilitation was not an issue in the case; the administrators made no claim that the presence of male personnel interfered with the performance of their duty. By contrast, this case involves just such a claim. Consequently, we are not dealing here with a mere matter of “consumer preference.” See Fernandez v. Wynn Oil Co., 653 F.2d 1273 (9th Cir.1981); Diaz v. Pan Am. World Airways, 442 F.2d 385, 389 (5th Cir.), cert. denied, 404 U.S. 950, 92 S.Ct. 275, 30 L.Ed.2d 267 (1971). See generally Schlei & Grossman, Employment Discrimination Law 341 (2d ed. 1983). The issue here is not what the prisoners want or believe is beneficial or appropriate. Rather, the issue is whether the state, charged with the responsibility of rehabilitation — a responsibility not only to the inmate but to the public — may pursue that goal in the manner at issue here.
2.
The more difficult question is whether the proposed BFOQ was “reasonably necessary” to furthering the objective of rehabilitation. The defendants can establish that the BFOQ was reasonably necessary only if they “had reasonable cause to believe, that is, a factual basis for believing, that all or substantially all [men] would be unable to perform safely and efficiently the duties of the job involved.” Dothard, 433 U.S. at 333, 97 S.Ct. at 2729 (quoting Weeks v. Southern Bell Tel. & Tel. Co., 408 F.2d 228, 235 (5th Cir.1969)).
*1531The district court determined that the defendants had not met this hurdle because “[t]hey offered no objective evidence, either from empirical studies or otherwise, displaying the validity of their theory.” Torres, 639 F.Supp. at 280. We respectfully differ with our colleague in the district court as to the legal appropriateness of this standard. Certainly, there is no general requirement that the necessity of a BFOQ be established by this type of evidence. In Dothard, the Supreme Court determined that Alabama prison officials were justified in removing female guards from the state’s “peculiarly inhospitable” penitentiaries. 433 U.S. at 334, 97 S.Ct. at 2729. The Court said that “[t]here is a basis in fact for expecting that sex offenders who have criminally assaulted women in the past would be moved to do so again if access to women were established within the prison.” Id. at 335, 97 S.Ct. at 2730. The Court also said that there would “be a real risk that other inmates, deprived of a normal heterosexual environment, would assault women guards because they were women.” Id. (footnote omitted). These appraisals were not based on objective, empirical evidence, but instead on a commonsense understanding of penal conditions, and, implicitly, on a limited degree of judicial deference to prison administrators.3 As noted earlier, in Pime v. Loyola University, 803 F.2d 351 (7th Cir.1986), this court held that a Jesuit university could discriminate permissibly in favor of Jesuit instructors even though it had “not been shown that Jesuit training is a superior academic qualification, applying objective criteria, to teach the particular courses.” Id. at 354. Similarly, in Chambers v. Omaha Girls Club, 834 F.2d 697 (8th Cir.1987), the Eighth Circuit held that a BFOQ need not always be supported by objective evidence. The Omaha Girls Club had discharged an unmarried pregnant staff member. The Club expected its staff members “to act as role models for the girls, with the intent that the girls will seek to emulate their behavior.” Id. at 699. The court concluded that the Club’s “role model rule” was justified by business necessity even though the plaintiff contended that “the role model rule is based only on speculation by the Club and has not been validated by any studies showing that it prevents pregnancy among the Club’s members.” Id. at 702. The court said:
Although validation studies can be helpful in evaluating such questions, they are not required to maintain a successful business necessity defense. Indeed, we are uncertain whether the role model rule by its nature is suited to validation by an empirical study. Consequently, the court’s conclusion in Hawkins [v. Anheuser-Busch, Inc., 697 F.2d 810 (8th Cir.1983)] is apt in this case: “We cannot say ... that validation studies are always required and we are not willing to hold under the facts of this case that such evidence was required here.”
Id. (footnote and citations omitted). Certainly, it is hardly a “[m]yth or purely habitual assumption,” Los Angeles Dep’t of Water & Power v. Manhart, 435 U.S. 702, 707, 98 S.Ct. 1370, 1375, 55 L.Ed.2d 657 (1978), that the presence of unrelated males in living spaces where intimate bodily functions take place is a cause of stress to females. See, e.g., EEOC v. Mercy Health Center, 29 FEP 159, 163 (W.D.Okla.1982) [available on WESTLAW, 1982 WL 3108] (employment of male nurses in labor and delivery area would cause medically undesired tension); Backus v. Baptist Medical Center, 510 F.Supp. 1191, 1194 (E.D.Ark.1981) (same); Norwood v. Dale Maintenance System, 590 F.Supp. 1410, 1422-23 (N.D.Ill.1984) (stress when attendant in office washroom is of opposite sex).
There is, it would seem, a substantial question as to whether the sort of problem *1532that the defendants’ BFOQ plan was designed to resolve ever “is suited to validation by an empirical study.” Chambers, 834 F.2d at 702. We need not resolve this matter definitively here.4 In this case, there was general agreement among the parties and the district court that such material simply did not exist. One of the plaintiffs’ own witnesses, Dr. David Kali-nich, testified that there is little scholarship in the area of rehabilitation of the female felon. Tr. 6 at 138, 142. The plaintiffs emphasized this point by failing to introduce any witness with substantial experience or who had done scholarly work in the field of rehabilitation within female institutions. The plaintiffs’ case was, to a very great extent, based on experts who had experience in male, not female, institutions. Indeed, the district judge, in accepting the credentials of Dr. Kalinich despite his lack of background in female institutions, specifically noted that comparable information about female institutions — and comparable experts — simply were not available. Tr. 6 at 143.
We believe, therefore, that the defendants were required to meet an unrealistic, and therefore unfair, burden when they were required to produce “objective evidence, either from empirical studies or otherwise, displaying the validity of their theory.” Torres, 639 F.Supp. at 280. Given the nature of their “business” — administering a prison for female felons — the defendants, of necessity, had to innovate. Therefore, their efforts ought to be evaluated on the basis of the totality of the circumstances as contained in the entire record. In the Title YII context, the decision of penal administrators need not be given as much deference as accorded their decisions in constitutional cases. Whitley v. Albers, 475 U.S. 312, 321-22, 106 S.Ct. 1078, 1085-86, 89 L.Ed.2d 251 (1986); Rhodes v. Chapman, 452 U.S. 337, 351-52, 101 S.Ct. 2392, 2401-02, 69 L.Ed.2d 59 (1981); Bell v. Wolfish, 441 U.S. 520, 547, 99 S.Ct. 1861, 1878, 60 L.Ed.2d 447 (1979). However, their judgments still are entitled to substantial weight when they are the product of a reasoned decision-making process, based on available information and experience. The fact that the program is considered a reasonable approach by other professional penologists also is a factor to be given significant consideration. In an area where the questions are so many and the answers so few, the range of reasonable options must necessarily be more extensive. Certainly, the court ought not require unanimity of opinion and ought not to substitute completely its own judgment for that of the administration.5
Conclusion
Because the district court resolved the issue before it on the ground that there was a lack of “objective evidence, either from empirical studies or otherwise,” 639 F.Supp. at 280, it had no occasion to evaluate the entire record, according the decision of the administrator the appropriate weight. Accordingly, we believe that this matter ought to be remanded to the district court for further consideration. After the parties have had an opportunity to submit additional evidence, the court may redetermine the issue in accordance with the standard set forth in this opinion.
Finally, we emphasize that it would be a mistake to read our decision today as a signal that we are willing to allow employers to elude Title VII’s requirements simply by arguing that they were “innovating.” Rare is the employment situation in which an employer could argue that gender-based distinctions are a “reasonably *1533necessary” approach to innovation in one’s business. We hold only that, given the very special responsibilities of these defendants and the obvious lack of guideposts for them to follow, it was error to require that they adopt only a course that was subject to objective validation. Accordingly, the judgment of the district court is reversed and the case is remanded for proceedings consistent with this opinion. Circuit Rule 36 shall not apply.
REVERSED AND REMANDED.
. Under this formulation, the Fifth Circuit in Diaz rejected the contention that sex was a BFOQ for airline flight attendants because the airline’s proposed justification — to provide a “pleasing environment" for passengers — was merely tangential to the airline’s primary objective of providing safe transportation.
. Dr. David Kalinich, a professor of criminal justice at Michigan State University, testified that he was not aware of any research regarding male correctional officers in female prisons. Tr. 6 at 138, 142. When the defendants challenged Dr. Kalinich’s qualifications as an expert because he lacked experience in dealing with female institutions, the district court responded that there was not enough research on the subject of female institutions "for us to be too chary about applying generalized expertise in this area.” Id. at 143.
. As in Dothard, the Court in Western Air Lines, Inc. v. Criswell, 472 U.S. 400, 105 S.Ct. 2743, 86 L.Ed.2d 321 (1985), suggested that an employer need not always validate a BFOQ with objective, empirical studies. The Court said that the defendant airline did not need "to establish the risk of an airline accident 'to a certainty....’” Id. at 419, 105 S.Ct. at 2754 (quoting Usery v. Tamiami Trail Tours, Inc., 531 F.2d 224, 238 (5th Cir.1976)). So long as the airline’s "argument has a credible basis in the record," the jury could permissibly "defer in a close case to the airline’s judgment.” Id. 472 U.S. at 420, 105 S.Ct. at 2754 (emphasis supplied).
. We are aware that the district court noted that Ms. Switala testified that the recidivism rate of 8 to 12 percent had not diminished during the BFOQ program. We do not believe that this factor taken alone is of any great relevance. First of all, the plaintiffs made no showing that such data, representing a relatively short time frame, is significant. Second, rehabilitation involves more than avoiding recidivism. It involves enhancing the individual's potential for personal achievement. This factor is not easily subject to empirical study.
. The district judge toured the prison during trial. Whatever the benefit of such an inspection with respect to matters of privacy and security, such an exercise is not relevant in determining the legitimacy of the defendants' rehabilitation program.