This Title VII matter comes before us on appeal from the district court's finding after a bench trial that plaintiff Dennis Soria, a Catholic of Italian background, was not discharged from his employment as a cement truck driver with defendant Ozinga Bros., Inc., a company largely owned and managed by individuals who were of traceable Dutch ancestry or members of the Christian Reformed Church, or both, due to his differing religion or national origin. On appeal, plaintiff chiefly argues that the district court erred in its refusal to consider as decisive certain statistical evidence offered by him linking company disciplinary patterns and religion and national origin. Plaintiff also contends that the district court erred in refusing to consider as relevant statistical evidence concerning the ethnic composition of defendant's work force, and claims that the court erred in certain findings of fact. Because we find no significant error in the district court's consideration of the evidence and no error at all in its ultimate findings, we affirm.

Defendant Ozinga Bros., Inc. (the company) is a small family-owned and -operated business which prepares and delivers building materials such as ready-mix concrete. It was founded by a Dutch immigrant whose son and three grandsons occupy the crucial management positions. Of the other managers, one is of Italian origin and two others are affiliated with the Catholic church. Although roughly half the company's employees are of traceable Dutch ancestry or Christian Reformed Church members (CRC), there was no allegation that plaintiff was discriminated against in regard to hiring or promotion. Instead, the sole issue at trial was whether plaintiff would not have been fired but for his non-Dutch, non-CRC background, or whether, as the company claimed, he was fired solely for persistently uncooperative and hostile behavior toward management and carelessness and irresponsibility in performance of his duties throughout his five years of employment, culminating in two serious truck accidents in the several days prior to his final discharge.

The evidence adduced at trial established a lengthy history of job-related problems between plaintiff and the company. The company's chief truck dispatcher and supervisor testified that during the 1977 season, plaintiff was the only cement truck driver with whom he experienced significant difficulty in the assignment of deliveries.1 Indeed, at least one contractor-customer had requested that the company not send plaintiff to its job-site because of plaintiff's uncooperative attitude. There was also testimony that on several occasions, plaintiff displayed a careless attitude toward his delivery obligations. For example, the company's dispatcher testified that even during busy periods, plaintiff several times stated in mid-afternoon that he would not accept any more deliveries for the remainder of the day; when queried as to the reason for his early departure, plaintiff was either silent or responded, at least on one occasion, "Well, I'll be sick after this [delivery]." In addition, the dispatcher testified that, as supervisor, his attempts to communicate with the plaintiff were unavailing, as plaintiff repeatedly ignored him or walked away. Further attempts to address this communications problem were rebuffed with the same behavior.

The supervisor's testimony was buttressed by that of a manager of Italian origin who noted that, while Soria may have had a "roughly average" record in regard to vehicle maintenance, the plaintiff manifested a "[v]ery unconcerned" and "lackadaisical" attitude toward specific problems noted to him. This picture was confirmed by one of plaintiff's witnesses, a fellow driver of Irish Catholic origin, who noted that plaintiff had a bad attitude at times and argued with the company's dispatcher and customers.

During plaintiff's final ten days of employment, he was involved in two serious accidents. Plaintiff's conduct in connection with these accidents apparently crystallized the company's long-standing dissatisfaction with plaintiff and served as a catalyst for his final discharge. In the first of these accidents, plaintiff, failing to watch his right hand mirror, drove his truck into a large ditch. Plaintiff and an assisting driver attempted to remove the truck from the ditch, but in so doing caused it to tip over, causing great damage. A member of company management arrived at the scene and concluded that the accident was due to plaintiff's carelessness. After consultation with other management members, plaintiff was laid off for the balance of that week. More important than the tip-over accident itself (other drivers employed by the company had been involved in similar accidents), the company testified, was the fact that the plaintiff belligerently denied responsibility for the accident and offered what management considered to be inappropriate excuses (e.g. that the accident was caused by the distraction of another car and the failure of the contractor to cooperate).2 When a manager admonished plaintiff that he was responsible for control over his vehicle despite the claimed distractions, plaintiff "proceeded to tell [him] that he didn't like what [he] was saying and that if [he] didn't back off, that he was going to straighten [him] out." Plaintiff offered no testimony to rebut this recollection.3 Even though such hostility was unprecedented in the manager's experience, plaintiff was allowed to return to work the following week.

The second accident occurred ten days later, when plaintiff ran his truck into a viaduct, causing damage to the cement-mixing barrel. Plaintiff again denied responsibility for the accident, claiming that he did not know the height of his truck, and that a company dispatcher had concurred in the choice of the route taken. The first excuse compounded the company's belief in plaintiff's irresponsibility, since the height of the truck was a readily ascertainable fact which other company drivers had taken the trouble to learn. At trial, the company dispatcher also denied having concurred in plaintiff's delivery route, and noted that such consultation was infrequent. Moreover, plaintiff appeared to acknowledge the depth of his dereliction and its natural consequences when, in returning to the company yard, he told the dispatcher, "This will probably be my last load, I just hit a viaduct." When a company manager met with plaintiff after the accident and heard the plaintiff's excuses and plaintiff's statement that he refused to drive the type of truck involved in the accident again, the manager rebuked him for his lack of responsibility and carelessness. Plaintiff was indefinitely laid off, and after a consultation among management in which the plaintiff's history of noncooperation and refusal to accept basic responsibility was discussed, it was decided to permanently discharge plaintiff.

One week after his discharge, plaintiff filed a grievance with a labor-management grievance committee empowered to reinstate him,4 in which, significantly, he did not allege any religious or ethnic discrimination. The committee rejected his demand to be reinstated. Several months thereafter, plaintiff filed a discrimination charge with the Equal Employment Opportunity Commission (EEOC), claiming discrimination on the basis of his Italian ancestry and Catholic affiliation; the EEOC rejected his claim. Finally, plaintiff instituted this action under Title VII of the Civil Rights Act of 1964, 42 U.S.C. Sec. 2000e-2(a)(1), changing his discrimination theory to allege that he was victimized as a non-Dutch, non-CRC affiliated employee rather than specifically as an Italian Catholic.

At trial, plaintiff attempted to demonstrate unlawful discrimination against him through two doctrinal avenues--"disparate impact" and "disparate treatment." First, he alleged that the company's informal, subjective system of discipline resulted in markedly more severe discipline being applied to non-Dutch and non-CRC drivers. Second, he alleged that his specific discharge resulted from discriminatory treatment based on his non-Dutch, non-CRC background. To establish his claim of disparate impact and to paint as pretextual the company's claim of legitimate, nondiscriminatory discharge, plaintiff relied chiefly upon comparative and statistical evidence purporting to demonstrate worse treatment of non-Dutch, non-CRC drivers. The district court, however, found the statistical data presented by both sides irreparably flawed and of "minimal assistance;" it also found the comparative evidence indeterminate. The district court concluded that plaintiff had proven neither disparate impact nor disparate treatment. Instead, the district court found that plaintiff had been discharged for "careless and unsafe attitudes, noncooperation with management, and lack of responsibility toward his job." From that determination, plaintiff appeals.

To establish unlawful disparate impact, the plaintiff must point to employment practices that are based upon facially neutral criteria but that in fact fall more harshly on one group and cannot be justified by business necessity; no proof of discriminatory motive is necessary. Teamsters v. United States, 431 U.S. 324, 336 n. 15, 97 S. Ct. 1843, 1854 n. 15, 52 L. Ed. 2d 396 (1977). The company challenges at the outset the application of the disparate impact paradigm to a disciplinary system which, as here, does not rely on neutral and objective criteria but rather is concededly based on somewhat subjective and individualized treatment. However, we need not reach that objection here, for, even assuming arguendo that disparate impact analysis may be applied to the present facts, we agree with the district court that the plaintiff failed to meet his burden of proof.

Here, the viability of plaintiff's disparate impact claim hinges solely on his proffered statistical and comparative evidence. Thus, the plaintiff argues on appeal that the district court erred in finding that this evidence was "not compelling" and "adequately rebutted by the defendant" and thus insufficient to alone establish disparate impact.5 We find no clear error in the district court's refusal to consider as sufficiently reliable the proffered evidence.6

At first glance, the plaintiff's statistical evidence appears to tell a dramatic story. As presented by the plaintiff's expert witness, for example, the figures reveal that while just over half of the company's 61 drivers were of Dutch origin and predictably were involved in an equally proportionate number of accidents, Dutch drivers were the subject of only 13.3 per cent of the recorded disciplinary actions. More strikingly, although over a third of the company's drivers were members of the Christian Reformed Church and were responsible for a third of the accidents, these CRC drivers were the subject of no recorded disciplinary actions.

However, even a cursory look behind the charts dispels their apparent conclusiveness. First, as the district court noted, the usefulness of these statistics was severely impaired by the small sample size involved. While the company employed 61 drivers, plaintiff was one of only four drivers who had been discharged, and one of only two drivers who were terminated for stated reasons. Indeed, an examination of this category of severest discipline would suggest that its application was random, not ethnically biased: of the four discharges, two involved Dutch drivers, and two involved non-Dutch drivers. But even if the statistical universe is expanded to include all drivers receiving any form of recorded discipline, as plaintiff's calculations do, the sample size is less than 15 drivers and covers only 15 incidents. Courts in Title VII actions have almost uniformly rejected statistical conclusions based upon such small samples, especially in discharge cases. See, e.g., Whack v. Peabody & Wind Engineering Co., 595 F.2d 190, 194 (3rd Cir.1979) (no "meaningful" statistical case where plaintiff is the only discharged member of allegedly disfavored group to be discharged); Williams v. Tallahassee Motors, 607 F.2d 689, 691-92 (5th Cir.1979), cert. denied, 449 U.S. 858, 101 S. Ct. 159, 66 L. Ed. 2d 74 (1980) (disciplinary disparities to be viewed with "caution" in small workforce of 68 employees); Harris v. Group Health Association, Inc., 662 F.2d 869, 872 (D.C.Cir.1981) (where all discharged employees number only 14, data "too fragmentary and speculative" to support claim).

Plaintiff argues that, in the present case, the small sample size was compensated for by the massaging of raw data into standard deviation measures rather than raw percentages. But it is still undeniable that even a minor change in the raw disciplinary data--as the expert demonstrated through his additional calculations (see infra )--would eliminate the statistical significance of any standard deviation. Cf. Friend v. Leidinger, 446 F. Supp. 361, 367 (E.D.Va.1977), aff'd 588 F.2d 61 (4th Cir.1978) (twofold differing treatment of black and white employees in 22 vehicular accidents involves data too "sparse" for conclusions, as shift of treatment in only 9 cases would eliminate disparity); Contreras v. City of Los Angeles, 656 F.2d 1267, 1273 & n. 4 (9th Cir.1981) ("Statistics are not trustworthy when minor numerical variations produce significant percentage fluctuations.").

Even were the sample size adequate, however, the district court noted a still more decisive flaw in the sample: its radical incompleteness. The plaintiff's data base includes only recorded disciplinary actions. Since the company's disciplinary system is largely informal and unwritten, the few recorded disciplinary incidents could scarcely provide a balanced picture of the company's overall disciplinary pattern. Moreover, the plaintiff significantly chose to omit from its analysis thirteen additional incidences of unwritten discipline recalled by the company management in deposition. This court has previously recognized that such deposition evidence from company officials is especially crucial in Title VII actions involving small employers who do not keep extensive personnel records. Kephart v. Institute of Gas Technology, 630 F.2d 1217, 1220 (7th Cir.1980), cert. denied, 450 U.S. 959, 101 S. Ct. 1418, 67 L. Ed. 2d 383 (1981).7

Finally, as the company's expert pointed out, the usefulness of the plaintiff's proffered statistical proof is diminished by its failure to properly categorize the nature of the accidents, the severity of the discipline administered, and the previous history of the disciplined driver. First, the accidents included in the plaintiff's sample occupied a spectrum from the comparatively trivial (knocking over a mailbox) to the serious (collision with an automobile with the possibility of injury). The effect of a proper categorization of serious (i.e. involving a threat to safety) and non-serious accidents was demonstrated by the company's expert witness: after considering the disciplinary actions evidenced by both records and deposition, and categorizing them according to seriousness, no statistically significant ethnic disciplinary disparity existed. Second, the plaintiff's statistics failed to categorize the severity of the recorded discipline, which ranged from mere verbal reprimands to actual suspension. Finally, the plaintiff's statistics do not account for the previous attitudinal and accident history of the disciplined drivers, which information the company has insisted is necessary to properly evaluate the propriety of any discipline administered. Any of these failings alone make manifest the dangers of relying on plaintiff's raw, undifferentiated data. Together, these flaws, along with the small and incomplete starting sample, provide ample evidence from which the district court could conclude that the plaintiff's statistical proof had been "adequately rebutted" even without embracing the alternative figures proposed by the company's expert witness.8

In addition to offering this severely flawed statistical evidence, plaintiff attempted to show disparate treatment through the production of driving records of five Dutch/CRC drivers who, he claimed, had demonstrably worse driving records than plaintiff but who were not discharged. However, as the district court noted, this comparative evidence was as inconclusive as the statistical evidence, since the company also produced the driving records of five non-Dutch, non-CRC drivers who had considerably worse driving records than the plaintiff but were also not discharged.9

Because there was no clear error in the district court's finding that plaintiff's statistical or comparative evidence adequately represented the reality of the company's disciplinary pattern, we can find no error in its legal conclusion that discriminatory disparate impact was not adequately demonstrated.

Plaintiff alternatively claimed that, apart from being a victim of the company's allegedly discriminatory discipline system, his specific discharge was tainted by discrimination based on his religion and national origin. Under "disparate treatment" theory, if the plaintiff proves a prima facie case (as the district court held he did here), the defendant must articulate some legitimate, nondiscriminatory reason for the employee's discharge; if such a reason is articulated, the plaintiff must in turn "prove by the preponderance of the evidence that the legitimate reasons offered by the defendant were not its true reasons, but were a pretext for discrimination." Texas Department of Community Affairs v. Burdine, 450 U.S. 248, 252-53, 101 S. Ct. 1089, 1093-94, 67 L. Ed. 2d 207 (1981), citing and quoting McDonnell Douglas Corp. v. Green, 411 U.S. 792, 93 S. Ct. 1817, 36 L. Ed. 2d 668 (1973). Plaintiff argues that the district court erred in failing to consider conclusive certain evidence, in finding that the company had met its burden of production by articulating "legitimate, nondiscriminatory" reasons for his discharge, and in finding that plaintiff had failed to demonstrate that the proffered reasons were pretextual.

First, plaintiff argues that the proffered reasons for his discharge--careless and unsafe attitudes, non-cooperation with management, and lack of responsibility toward his job, exhaustively documented and testified to--are not sufficiently "clear" and "specific" to meet the company's burden of production and adequately frame the factual issue for trial as required under Burdine, 450 U.S. at 255-56, 101 S.Ct. at 1094-95. This contention is plainly without merit. Indeed, it would be hard to conceive of reasons more facially supportive of a proper discharge than the plaintiff's belligerent and insolent conduct toward the company's management, his repeated refusal to deliver loads at required times, and his indifference to repeated attempts to discuss and improve upon his failings. Non-cooperation, belligerence and refusal to perform assigned tasks have been repeatedly found by this court and others to be sufficiently clear, legitimate, non-discriminatory reasons for employee discharge. See, e.g., Barnes v. St. Catherine's Hospital, 563 F.2d 324, 326 (7th Cir.1977); Potter v. Goodwill Industries, 518 F.2d 864, 865 (6th Cir.1975); Jack v. American Linen Supply Co., 498 F.2d 122, 124 (5th Cir.1974); Equal Employment Opportunity Commission v. Murphy Motor Freight Lines, Inc., 488 F. Supp. 381, 387-88 (D.Minn.1980); Kralowec v. Prince George's County, Md., 503 F. Supp. 985, 1001 (D.Md.1980). In Robbins v. White-Wilson Medical Clinic Inc., 660 F.2d 1064 (5th Cir.1981), cited by plaintiff, the court, while not holding the discharged employee's lack of a "pleasant personality" in itself legally insufficient to meet the McDonnell Douglas burden of production, refused to credit such an explanation only after cross-examination demonstrated that the defendant considered a "pleasant personality" to be synonymous with membership in the favored group. Id. at 1067. There was no such demonstration that the company's conduct standards applied in this case simply masked ethnic stereotyping.

Second, plaintiff contends that, even assuming arguendo that the district court properly found his prima facie case adequately answered, the district court nonetheless erred in finding that he failed to demonstrate that the company's proffered reasons for discharge were pretextual. While this factual issue was hotly contested at trial, the company presented ample evidence supporting its claim, noting plaintiff's long history of non-cooperation with and hostility towards management, and his propensity to refuse responsibility for his negligent behavior. In addition, plaintiff called as witnesses four other drivers, only one of whom was of Dutch origin; two of them flatly rejected the suggestion that plaintiff was fired because of his ethnic background, two indicated that they themselves had not experienced any discrimination as a result of their non-Dutch background, and one confirmed management's picture of plaintiff as having a bad attitude and as arguing with management and customers. None of the five non-Dutch, non-CRC employees (including an Italian manager) called as witnesses by the company claimed to have experienced any ethnic discrimination from the company.

Plaintiff points to other evidence indicating that management had experienced problems with other drivers (unspecified, however, as to ethnic origin), that plaintiff was an average employee regarding truck maintenance, and that other drivers had on occasion made excuses for their accidents. In doing so, plaintiff is simply inviting this court to reweigh competing evidence and conclude that, despite overwhelming proof of plaintiff's non-cooperation, belligerence, and refusal to accept responsibility, the company did not truly view plaintiff's difficult behavior as uniquely intense and worthy of discharge. We must decline such an invitation.

Plaintiff also argues that the district court failed to examine the company's justification for plaintiff's discharge with appropriate suspicion in view of the especially abuse-prone nature of subjective, individualized disciplinary systems, especially when operated by managers of the supposedly favored group. However, this argument was fully aired at trial, and was simply one factor considered among others. Moreover, the evidence indicated that the company's management, including at least one non-Dutch, non-CRC employee, undertook a sober and careful assessment of plaintiff's record and only then decided to permanently discharge the plaintiff. The district court was not required to discard strong evidence of the genuineness of the family management's conclusions simply because they hailed from a different ethnic group. Sessions v. Rusk State Hospital, 648 F.2d 1066, 1071 (5th Cir.1981).

Finally, the plaintiff contends that the district court's evaluation of the company's proffered legitimate, non-discriminatory reasons was marred by its refusal to consider as conclusive the plaintiff's statistical and comparative proof regarding ethnic bias in the application of driver discipline, and by its refusal to consider at all information concerning the company's ethnic composition and hierarchy compared to that of the surrounding community.

We may dispose of the plaintiff's first point quickly, for, as the discussion at II, supra, indicates, the district court was well justified in considering plaintiff's flawed and incomplete disciplinary statistics to be of "minimal assistance" and in any event "adequately rebutted" by the company's well-founded criticisms.

We must pause somewhat longer, however, over the district court's refusal to consider plaintiff's statistical ethnic profile of the company's workforce and management. Evidence of discriminatory hiring and promotional practices has been considered relevant in Title VII cases alleging discriminatory discharge or refusal to rehire. See, e.g., McDonnell Douglas Corp. v. Green, 411 U.S. 792, 804-05, 93 S. Ct. 1817, 1825-26, 36 L. Ed. 2d 668 (1973); Marquez v. Omaha District Sales Office, Ford Division of Ford Motor Co., 440 F.2d 1157, 1160-61 (8th Cir.1971). Hence, the district court technically erred in ignoring this information. But this error was insignificant, for the evidence offered was so incomplete that it could not have assisted in the establishment of a discriminatory motive. Specifically, the plaintiff's expert never tightly defined the "relevant labor market" for the company's hiring, as required to provide a meaningful comparison, see Taylor v. Safeway Stores, Inc., 524 F.2d 263 (10th Cir.1975); Neloms v. Southwestern Electric Co., 440 F. Supp. 1353 (W.D.La.1977), instead positing four hypothetical geographical areas which were not demonstrated to be within the normal hiring radius of the company's plant. More importantly, the plaintiff's expert admitted that he knew nothing about the company's specific hiring procedures (e.g., whether jobs were advertised or who made hiring decisions) and no evidence was offered to demonstrate whether these hiring procedures or promotional policies had in fact worked to exclude a single non-Dutch, non-CRC applicant from employment or advancement. Moreover, plaintiff's expert admitted that he had no basis for linking the company's disciplinary procedures with the company's hiring or promotion process. Such an insubstantial offer of proof falls far short of the standard required for even an ancillary demonstration of a discriminatory employment milieu. See, e.g., Person v. J.S. Alberici Construction Co., 640 F.2d 916, 919 (8th Cir.1981) (in absence of more detailed evidence, hiring information "too attenuated" to establish discriminatory motive in termination); Metcalf v. Omaha Steel Castings Co., 507 F. Supp. 679, 689 (D.Neb.1981) (actual anecdotal evidence of discrimination required to establish ancillary finding of "racially oppressive work environment"). Although we admonish trial courts in the future to at least consider such oblique evidence of discrimination in non-challenged practices, the failure of the district court to do so here caused no harm in view of the fragmentary nature of the plaintiff's statistical offer of proof and the overwhelming record evidence that plaintiff was genuinely discharged for his uniquely troublesome work behavior.

For the foregoing reasons, the district court's entry of judgment in favor of the defendant is affirmed.

AFFIRMED.