Cornelia Ward brought this action against her former employer, Westland Plastics, complaining of sex discrimination in compensation, working conditions, and discharge in violation of Title VII of the Civil Rights Act of 1964, 42 U.S.C. §§ 2000e, et seq., and the equal pay provision of the Fair Labor Standards Act, 29 U.S.C. § 206 (d). After two lawyers withdrew from the case, Ward proceeded to trial, acting as her own attorney. The court awarded judgment for the defendant on the Title VII claim, following the jury’s advisory verdict. The court also gave the employer judgment on the equal pay claim on the jury verdict. Ward appeals from both judgments, charging that the court erred as to the law in several respects, that the judgments are not supported by the evidence, and that the court’s conduct during trial was prejudicial. We affirm.

I. Defense of “Business Necessity.”

Ward contends that the trial court erroneously defined the defense of “business necessity.” The term “business necessity” is usually reserved for the affirmative defense available to a Title VII defendant once it has been established that the employer’s practice or policy has disparate adverse impact on a protected group. See Griggs v. Duke Power Co., 401 U.S. 424 , 91 S.Ct. 849 , 28 L.Ed.2d 158 (1971); Blake v. City of Los Angeles, 595 F.2d 1367, 1376 (9th Cir. 1979); deLaurier v. San Diego Unified School Dist., 588 F.2d 674, 678 (9th Cir. 1978).

Although Ward complained of individual treatment rather than raising a disparate impact claim, Westland offered an analogous defense for one particular incident challenged by Ward, the company president’s request that she not participate in a business lunch because the potential buyer, for whom the lunch was scheduled, wanted to be the only woman in attendance. The court found that this admittedly gender-based treatment was necessitated by legitimate business goals which could not be achieved by a reasonable alternative. This excuse or explanation is not a “business necessity” defense as that term is generally used in Title VII cases. Nevertheless, it is a defense that an apparently discriminatory action was required to achieve a legitimate business goal. The trial court used the term “business necessity” to describe this defense. The trial court effectively communicated the correct standard to the jury, both in trial remarks and instructions, and further demonstrated with correct conclusions of law that the proper standard had been applied.

Ward relies on cases holding that customer preference does not excuse discriminatory employment policies or practices. See, e. g., Diaz v. Pan American Airways, Inc., 442 F.2d 385 (5th Cir.), cert. denied, 404 U.S. 950 , 92 S.Ct. 275 , 30 L.Ed.2d 267 (1971). But cf. Fernandez v. Wynn Oil Co., 20 EPD 130,237 (C.D.Cal. July 30,1979), appeal docketed, No. 79-3598 (9th Cir. Sept. 26, 1979) (maleness is bona fide occupational qualification for position of International Marketing Director in company doing substantial business in foreign countries in which prevailing mores were against business dealings with females). Westland did not have a policy of prohibiting female employees generally, or Ward in particular, from dealing with buyers who preferred doing business with males. What happened here was an isolated request necessitated by an eleventh hour, unexpected, and unique situation. Management responded as it did primarily to avoid potential embarrassment to itself and to the individuals involved, including Ward, not to accommodate a buyer's sexist preference. The court’s finding in this regard is not clearly erroneous.

Westland raised two types of defenses to Ward’s other allegations of discrimination. First, it refuted the very existence of the facts Ward sought to prove in making a prima facie showing of illegal discrimination. See Mosby v. Webster College, 563 F.2d 901, 903-04 (8th Cir. 1977). Second, Westland attempted to articulate legitimate, nongender related business reasons for its actions. Board of Trustees of Keene State College v. Sweeney, 439 U.S. 24 , 99 S.Ct. 295 , 58 L.Ed.2d 216 (1978); McDonnell-Douglas Corp. v. Green, 411 U.S. 792 , 93 S.Ct. 1817 , 36 L.Ed.2d 668 (1973).

It is apparent from the transcript and the trial court’s findings and conclusions that the court was using the term “business necessity” to refer to Westland’s second type of defense. That is, Westland did what it did for business reasons completely unrelated to Ward’s gender. The trial court apparently meant, and effectively told the jury, that if Westland proved it acted for legitimate business purposes it must prevail in the action unless Ward proved that West-land’s “purposes” were a pretext for illegal discrimination. This statement of the law was essentially correct. The court used the term “business necessity” in two different ways. Neither of the meanings the court applied corresponded to the usual sense of “business necessity” in Title VII actions. This perhaps confused the plaintiff, but it did not constitute error.

On all but the luncheon episode, the findings do not clearly indicate at what stage Ward lost her Title VII case, whether by failing to make the prima facie showing or by failing to prove that defendant’s proffered reasons were pretextual. This is neither improper nor unusual. Mosby v. Webster College, 563 F.2d 901 (9th Cir. 1977). However, the ultimate finding remains that Ward failed to prove sex discrimination, not that sex discrimination was somehow justified by “business necessity.” Again, ample evidence supports this result.

II. “Subjective" Evaluation.

The parties argue over whether Ward’s supervisor’s evaluation of her ability and performance, which led to her discharge, was subjective. According to Ward, it was and therefore was also suspect; that is, Ward maintains that subjectivity creates a strong inference of impermissible discrimination sufficient to require the defendant to rule out the possibility of sex bias.

Westland insists that the absence of written criteria does not imply subjectivity of evaluation. The employer points to the relative informality, closeness, and smallness of its management team as indicative of ongoing, daily opportunities to observe and evaluate Ward’s performance objectively.

Even assuming subjectivity was involved here, it has never been held that subjective evaluation by an employer is per se prohibited by Title VII, or alone shifts to the defendant the burden of proving absence of intentional sex bias, as Ward would have this court hold. (See Bolton v. Murray Envelope Corp., 493 F.2d 191 (5th Cir. 1974), for an example of how a subjective evaluation claim is treated differently in individual treatment versus class impact cases.)

Of the seven managers during Ward’s tenure, three were family member owners. While there was undisputed evidence of sex bias on the part of one of them, Jim Forsman, Ward was evaluated and discharged by Gus Forsman, for whom there was no such evidence. In addition, Gus testified that Jim’s sexist remark concerning Ward had no influence on the firing decision and in fact had annoyed Gus. On these facts, the court was not compelled to find for Ward or to make an independent assessment of her abilities, as Ward now suggests the court should have done.

III. Statistical Evidence.

Ward contends that the judge erroneously criticized both the relevance and value of her statistical evidence. Again, we disagree.

While the Supreme Court has recognized the existence of a role for statistics in disparate treatment cases, for example, in a plaintiff’s attempt to establish pretext, McDonnell-Douglas Corp. v. Green, 411 U.S. 792, 805 , 93 S.Ct. 1817, 1825 , 36 L.Ed.2d 668 (1973), it has cautioned against giving controlling influence to statistics. Id. at n. 19. Regardless of how devastating and reliable the statistics may look, the issue remains in these cases whether a particular isolated historical event was discriminatory. An individual’s discharge may be justified despite overall statistics suggesting discriminatory policies. See, e. g., Bolton v. Murray Envelope Corp., 493 F.2d 191 (5th Cir. 1974); Peters v. Jefferson Chemical Co., 516 F.2d 447 (5th Cir. 1975). Thus, some cautionary remarks are not inappropriate in these cases.

Caution may have been even more necessary here because Ward’s statistical proof in the management sector was based on an unreliably small sample. See Morita v. Southern California Permanente Medical Group, 541 F.2d 217, 219-20 (9th Cir. 1976), cert. denied, 429 U.S. 1050 , 97 S.Ct. 761 , 50 L.Ed.2d 765 (1977); White v. City of San Diego, 605 F.2d 455, 459-61 (9th Cir. 1979).

IV. Expert Testimony.

Ward assigns as reversible error the district court’s refusal to permit her expert in discrimination and affirmative action to testify that in his opinion Westland discriminated against Ward on account of her sex. Evidently concerned that the testimony would invade the province of the jury, the court rejected it as “contrary to all rules of evidence” and “not permissible.”

Ward’s attack on the legal correctness of the court’s reasoning is not without merit. See Fed.R.Evid. 702, 704; United States v. Johnson, 319 U.S. 503, 519-20 , 63 S.Ct. 1233, 1241 , 87 L.Ed. 1546 (1943). However, district courts enjoy broad discretion in admitting or rejecting evidence, including the testimony of experts. Cf. Kline v. Ford Motor Co., 523 F.2d 1067, 1070 (9th Cir. 1975) (district court may properly refuse to admit expert opinion on the ultimate issue based on its assessment of the borderline value of the evidence to the jury).

Ward cites only one case in which the trial court’s denial of expert testimony, based on incompetency grounds, was held reversible error. Moran v. Ford Motor Co., 476 F.2d 289 (8th Cir. 1973). But in Moran , the issue on which the expert would have testified was the causal link between the plaintiff’s automobile accident and the defendant’s wheel suspension system. Without deciding whether the addition of that testimony would certainly have averted the directed verdict for the defendant, the court reversed because of the obvious prejudice the plaintiff may have suffered from its omission.

The prejudice Ward suffered from the omission of her expert’s opinion is not obvious. The question whether gender was the basis of differential treatment is not so technical as to require the aid of an expert to enlighten the jury or court. Nor did Ward suffer a directed verdict in this case. The omission of the opinion evidence does not amount to abuse of discretion.

In addition, one particular hypothetical Ward wanted to pose would have tracked the luncheon incident, in which even the defendant admits gender was a factor. The judge was correct in ruling the expert incompetent to voice an opinion on whether that or any other conduct constituted illegal sex discrimination.

V. District Court’s Conduct During Trial.

Ward claims that the trial judge injected himself excessively into the trial by his remarks and questions and that his overall conduct during trial indicated early bias against her case, prejudiced the jury, and denied her a fair trial.

From our perusal of the record in this case, it appears that Ward has some cause to complain. The judge exhibited impatience with her inept examination of witnesses, sometimes taking oyer the job himself. He demeaned the quality and relevance of much of Ward’s evidence, suggested his own nondiscriminatory reasons for some of the employer’s behavior, and even disparaged Congressional wisdom in enacting Title VIL, While the trial judge enjoys broad discretion in controlling the conduct of trial, United States v. Panza, 612 F.2d 432 (1979), and while plaintiffs in employment discrimination cases are not entitled to a trial judge with a liberal or generous attitude, Frausto v. Legal Aid Society of San Diego, Inc., 563 F.2d 1324, 1327 (9th Cir. 1977), this court has recognized that they are entitled to one who is “detached, fair and impartial.” Id.

After careful review of all of the evidence, we have concluded that the district court’s conduct, while borderline, did not ultimately prejudice plaintiff’s case. Ward bore the final burden of persuasion on the issue of illegal discrimination. Her case rested chiefly on her own uncorroborated and heavily impeached testimony. ' Several defense witnesses, none of whom were still employed at Westland, corroborated the employer’s version of the events, offering credible legitimate explanations for the company’s actions vis-a-vis the plaintiff.

Maheu v. Hughes Tool Co., 569 F.2d 459, 471-72 (9th Cir. 1978), provides a useful contrast. The central issue in this defamation action was the plaintiff’s credibility. In spite of voluminous conflicting evidence on that issue, including “a mass of very persuasive evidence” of the plaintiff’s dishonesty, and although the plaintiff’s own testimony was “thoroughly and convincingly impeached” — some even described as “ludicrous” — the trial judge gave the jury a personal glowing character reference for the plaintiff. The harm was compounded by the timing of the judge’s remarks just prior to the jury’s retirement and by his order that these comments be transcribed and included in the written instructions for the jury. This court found reversible error, unmitigated by the fact that the trial court cautioned the jury that it was free to disregard his comments on the evidence.

In the instant case, the judge seldom spoke directly to the ultimate issues, frequently cautioned the jury not to take his questions and remarks as evidence, and properly instructed the jury at the end of the case. The most important contrast with Maheu , however, is the lack of persuasive evidence in the instant case on the ultimate issues.

While we regret that the trial judge’s conduct fell short of exemplary, the defendant should not be made to bear the added expense of a new trial when there is no reason to believe the plaintiff could prove a case in a perfect trial.

Affirmed.

. Ward relies heavily on discriminatory impact cases, such as Albemarle Paper Co. v. Moody, 422 U.S. 405 , 95 S.Ct. 2362 , 45 L.Ed.2d 280 (1974), in which plaintiffs had proved that members of the protected class were disproportionately harmed by employer decisions (hiring, firing, layoff, promotion). Evidence that those decisions were based on vague or variably applied criteria or totally subjective evaluations then supplied the missing link between the statistics and the issue to be proved, illegal discrimination.