Johnnie Gilbert and ten other black police officers presently or formerly employed by *1393the Little Rock Police Department appeal from a final judgment entered in the District Court1 for the Eastern District of Arkansas denying them relief on their individual claims of racial discrimination in the Police Department’s promotional practices and work atmosphere. Appellees are the City of Little Rock, the Police Chief and the' City Manager (collectively referred to here-, in as the City). For reversal appellants argue that the district court erred in (1)1 finding that the racial atmosphere in the work place did not constitute a violation of1 Title VII, (2) finding that the City did not: discriminate on the basis of race in the imposition of disciplinary measures, (3) finding that the promotional system did not operate to exclude black police officers from higher-level positions on the basis of race, and (4) refusing to grant appellants’ motions for order of recusal and for class certification. For the reasons discussed below, we hold that the district court erred in its analysis of the evidence of discrimination in the City’s promotional practices and remand for further proceedings on that issue. We affirm the district court’s judgment in all other regards.
On September 27, 1978, appellants Johnnie Gilbert, Horace Walters, Andrew Lockhart, and Billy O’Donald filed the original complaint in this action under Title VII of the Civil Rights Act of 1964, 42 U.S.C. § 2000e et seq., and 42 U.S.C. §§ 1981 and 1983. On February 22, 1979, the remaining appellants, Julius Bryant, Jack Matlock, Mixie Alexander, Jesse Briscoe, Grady Alexander, Marcella Wilson and Larry Baz-zelle,2 filed their complaint in intervention under the same provisions as the original ' complaint and under the additional provision of 31 U.S.C. §§ 1221, 1242 and 1244 (Fiscal Assistance Act of 1972), and 42 U.S.C. § 3789D (Crime Control Act of 1968).
Appellants’ complaints as to the City’s policies and procedures alleged to be racially discriminatory were as follows:
(1) Black officers were subjected to numerous incidents of racial harassment by white officers and the personnel in charge did nothing to prevent this.
(2) A greater severity of discipline was imposed on black officers than on white officers for similar infractions.
(3) The Police Department’s system for promotion of officers to the positions of sergeant and lieutenant operated to exclude blacks from these high level positions.
(4) Job transfer and in-service training decisions were made on the basis of race and adversely affected black police officers.
• The case was tried to the district court without a jury. The record in the case is voluminous and the district court filed a detailed 58-page opinion. Although other causes of action were alleged in the complaints, trial of this case centered around alleged violations of Title VII. The issues raised on appeal are also framed in terms of a Title VII action, and accordingly our discussion and analysis will focus on this cause of action.3
Individual actions brought under Title VII may seek to remedy either disparate treatment or the results of a disparate impact upon a protected group. Disparate treatment occurs where “[t]he employer simply treats some people less favorably than others because of their race.” International Brotherhood of Teamsters v. United States, 431 U.S. 324, 335-36 n. 15, 97 *1394S.Ct. 1843, 1854 n. 15, 52 L.Ed.2d 396 (1977). On the other hand, disparate impact “involve^] employment practices that are facially neutral in their treatment of different groups but that in fact fall more harshly on one group than another.” Id. Aspects of both of these theories of discrimination are involved in the present case.
A court of appeals may only reverse a district court’s finding of discrimination if it concludes that the finding is clearly erroneous under Rule 52(a). Pullman-Standard v. Swint, 456 U.S. 273, 290, 102 S.Ct. 1781, 1791, 72 L.Ed.2d 66 (1982); Coble v. Hot Springs School District No. 6, 682 F.2d 721, 723 (8th Cir.1982) (Coble). The clearly erroneous standard has been characterized as follows: “A finding is ‘clearly erroneous’ when although there is evidence to support it, the reviewing court on the entire evidence is left with the definite and firm conviction that a mistake has been committed.” Pullman-Standard v. Swint, 456 U.S. at 284-85 n. 14, 102 S.Ct. at 1788 n. 14 (quoting United States v. United States Gypsum Co., 333 U.S. 364, 395, 68 S.Ct. 525, 542, 92 L.Ed. 746 (1948)).
Because the individual claims allege similar discriminatory practices, we will discuss the district court’s findings in terms of the challenged practices.
Discriminatory Work Environment
Appellants claimed that they were repeatedly subjected to racial slurs and that a racially biased atmosphere pervaded the work place. Appellants further alleged that this situation was brought to the attention of the City and that no attempt was made to correct it.
Title 42 U.S.C. § 2000e-2(a)(l) provides that an employer may not “discriminate against any individual with respect to ... terms, conditions or privileges of employment, because of such individual’s race.” A working environment dominated by racial hostility and harassment constitutes a violation of Title VII, regardless of any other tangible job detriment to minority employees. An employer violates Title VII simply by creating or condoning an environment at the work place which significantly and adversely affects the psychological well-being of an employee because of his or her race. Walker v. Ford Motor Co., 684 F.2d 1355, 1358 (11th Cir.1982); Johnson v. Bunny Bread Co., 646 F.2d 1250, 1257 (8th Cir.1981) (Bunny Bread); EEOC v. Murphy Motor Freight Lines, Inc., 488 F.Supp. 381, 384 (D.Minn.1980).
More than a few isolated incidents of harassment must have occurred to establish a violation of Title VII. Bunny Bread, 646 F.2d at 1355; Cariddi v. Kansas City Chiefs Football Club, Inc., 568 F.2d 87, 88 (8th Cir.1977). Whether the degree of harassment is sufficiently severe and persistent to trigger Title VII sanctions is to be determined from a totality of the circumstances. Henson v. City of Dundee, 682 F.2d 897, 904 (11th Cir.1982).
Appellants presented much evidence on the extent to which racial slurs were used and condoned at the Little Rock Police Department. Each of the present or former black police officers involved in this case testified to racial remarks and derogatory epithets used by white officers, including those in high-level positions. There was testimony that anti-black sentiments were expressed in other ways as well, such as racially oriented graffiti in the restrooms and a racial cartoon posted on a bulletin board at police headquarters.
The district court nevertheless found that appellants’ evidence was insufficient to establish a pattern of harassment and that the allegations only involved isolated events rather than a “standard operating procedure.” 544 F.Supp. at 1239. These findings were findings of fact which must be upheld unless clearly erroneous. Walker v. Ford Motor Co., 684 F.2d at 1359 (whether use of racially abusive language is repeated and continuous enough to rise to a Title VII violation is a question of fact). We express our strong disapproval of the degree of racial harassment at the Little Rock Police' Department. However, under the clearly *1395erroneous standard we must affirm the district court’s finding that this degree never rose to the level of a Title VII violation.
Discipline
Appellants Matlock, O’Donald, Walters, Bryant and Lockhart relied on a disparate treatment theory to establish discrimination in the City’s imposition of disciplinary measures against them. A case proceeding on this theory has three phases.! First, the plaintiff must prove a prima facie case of disparate treatment; second, the burden shifts to the defendant to rebut the! prima facie case by articulating a legiti-! mate, nondiscriminatory reason for the challenged action; and third, the plaintiff may show that the defendant’s rebuttal is a pretext for a discriminatory motive. The plaintiff retains the burden of persuasion throughout. Texas Department of Community Affairs v. Burdine, 450 U.S. 248, 252-54, 101 S.Ct. 1089, 1093-1094, 67 L.Ed.2d 207 (1981); McDonnell Douglas Corp. v. Green, 411 U.S. 792, 802, 93 S.Ct. 1817, 1824, 36 L.Ed.2d 668 (1973).
The district court reviewed each challenged disciplinary action and found that none was implemented on the basis of race. 544 F.Supp. 1253, 1255, 1257, 1259, 1260. These findings are not clearly erroneous and must be approved. See Soria v. Ozinga Bros., 704 F.2d 990, 997-98 (7th Cir.1983); Pope v. City of Hickory, North Carolina, 679 F.2d 20 (4th Cir.1982) (findings that municipal employer met requirement of articulating legitimate reason for suspension and subsequent. discharge of black police officer, and that former officer had not proved reasons were pretextual, were not clearly erroneous); Bunny Bread, 646 F.2d at 1255-56 (disciplined black employee did not show similarity between his conduct and that of white employees not disciplined).
Promotions
We come now to the most difficult aspect of this ease. Appellants Gilbert, O’Donald, and Walters claimed that they were not promoted to the position of sergeant because of racial discrimination on the part of the City. Appellant Anthony made the same assertion with respect to the position of lieutenant. The promotional system is attacked under a disparate impact theory. To establish a prima facie case on a disparate impact claim, a plaintiff need only show that a particular practice, though neutral on its face, has an adverse impact upon a protected class. The burden then shifts to the employer to show that the practice is related to job performance and justified by business necessity. If the employer meets this burden, the plaintiff may then show that other practices, which lack a similarly discriminatory effect, would satisfy the employer’s legitimate interests. Griggs v. Duke Power Co., 401 U.S. 424, 431, 91 S.Ct. 849, 853, 28 L.Ed.2d 158 (1971); Kirby v. Colony Furniture Co., 613 F.2d 696, 703 (8th Cir.1980); Firefighters Institute for Racial Equality v. City of St. Louis, 549 F.2d 506, 510 (8th Cir.1977) (Firefighters), cert. denied, 434 U.S. 819, 98 S.Ct. 60, 54 L.Ed.2d 76 (1977).
The job classifications of the uniformed personnel at the Little Rock Police Department, beginning with the entry level category, are patrolman, sergeant, lieutenant, captain, assistant chief and chief. Promotion within the department is governed by procedures established by the City of Little Rock Civil Service Commission.
To be eligible to compete for a promotion, the candidate must be a permanent police officer serving on active status in a certain classification for a given amount of time. Thus, to be eligible for promotion to sergeant, a candidate must have a minimum of three years of service as a patrolman, and to be eligible for promotion to lieutenant, a candidate must have at least one year of service as a sergeant.
The promotion selection process for the positions of sergeant and lieutenant are *1396conducted at least once a year. A certified list of candidates for each position is prepared by the Civil Service Commission in rank order based on the candidates’ composite scores comprised of a written examination, an oral interview, performance appraisal, chief’s rating,4 and seniority. A candidate must achieve a certain minimum score on the written examination in order to complete the selection process. Each of the component factors is assigned a certain percentage of the composite score. Promotions must be made, as vacancies arise, from the certified lists in order, beginning with the candidate in first standing. At the end of each year, the certified lists automatically expire and the process starts over again.
The district court treated the issue of discrimination in promotions under both a disparate impact and disparate treatment analysis. The court found that appellants made a prima facie case on the issue of disparate impact of the promotional process, but that the City’s rebuttal evidence undermined the accuracy and significance of appellants’ statistical evidence. Based on the record as a whole, the district court made the following conclusions: “The only part of the [Little Rock Police Department] promotional examinations for sergeant and lieutenant that serves as a pass/fail barrier is the written test, and it has no adverse impact on blacks.... Other analyses also reflect that there is no adverse impact in any of the subpárts of the promotional process.” 544 F.Supp. at 1247. This latter conclusion was based on comparisons between the average composite, oral and service scores for whites and for blacks.
In reviewing the individual claims, apparently under a disparate treatment approach, the district court found that several appellants did make prima facie cases of discrimination in their failure to be promoted, but that the City successfully rebutted these “on the basis of the promotion system.” Id. at 1254.
We believe that the district court’s analysis was incorrect because it did not focus on the interrelationship of the component parts of the promotional procedure.
During the years 1975-1979 inclusive, the Little Rock Police Department had approximately 215 uniformed personnel. Approximately 14, or 6.7%, of the force were black. These numbers have remained rather constant. Of the 26 officers promoted to sergeant during this period, one (appellant Anthony) was black.5 At the time of trial in 1982, one black was employed as a sergeant (appellant Anthony) out of a total of 35 sergeants. None of the force’s 14 lieutenants was black, nor were any of the higher ranking officers.
Because promotions were made from within, general population statistics are not probative for determining whether an inference of discrimination arose from the low number of blacks in the higher levels of the uniformed police force. Paxton v. Union National Bank, 688 F.2d 552, 564 (8th Cir.1982), cert. denied,-U.S. -, 103 S.Ct. 1772, 76 L.Ed.2d 345 (1983). We have noted above the small absolute number of blacks on the police force. Furthermore, because a minimum number of years in service was required for promotion eligibility, statistical comparisons should be based on the still smaller group of black officers who possessed this necessary qualification. Hazelwood School District v. United States, 433 U.S. 299, 308 n. 13, 97 S.Ct. 2736, 2742 n. 13, 53 L.Ed.2d 768 (1977); Coble, 682 F.2d at 733.
Data presented at trial regarding the testing procedures for promotion to ser*1397geant6 reveals that, as the district court observed, the passing rate for the black candidates taking the written examination was comparable to that of the white candidates. The percentage of blacks who passed the written examination and who were actually placed on the certified lists was also comparable to the percentage of whites. It was primarily based on these statistics that the district court concluded the testing procedure did not have a disparate impact on black officers seeking promotion.
The critical factor in being promoted, however, was not to be placed on the list, but to be placed at a high standing on the list. This is especially critical in view of the fact that the certified lists automatical-' ly expired at the end of one year. Thus by controlling the number of vacancies which would come up each year, the City could effectively exclude black officers from the higher level positions in the Police Department. Such a practice would clearly be in violation of Title VII which makes it unlawful for an employer “to limit ... his employees ... in any way which would deprive or tend to deprive any individual of employment opportunities or otherwise adversely affect his [or her] status as an employee, because of such individual’s race.” 42 U.S.C. § 2000e-2. There was some evidence that such manipulation of vacancies occurred,7 and the statistical evidence gives some support to this claim.8 However, the district court found this evidence on discriminatory intent unconvincing and we cannot say that such a finding was clearly erroneous. See Firefighters, 549 F.2d at 516.
By failing to focus on the ranking aspect of the promotional system, the district court neglected to adequately consider the interrelationship of the component factors and, more specifically, whether the oral interview and performance appraisal factors, which accounted for between 40%-50% of a candidate’s composite score, had a disparate *1398impact on black officers’ scores and thus on their rank on the list. It is not sufficient for such a determination to compare the average scores of whites and blacks on these subparts of the procedure. Rather, the correlation between scores on the written exam on the one hand and on the oral exam, performance appraisal, and chief’s rating on the other needs to be examined. Thus, if there is a low correlation, i.e. those blacks scoring high on the written exam do not score as relatively high on the other components, these other components would operate in a racially disparate manner. The relative weight accorded to each component factor would also affect a candidate’s composite score and hence his position on the eligibility list. See id., 549 F.2d at 509 n. 3.
In 1978, appellant Gilbert made 99 on the written exam, but placed 15th on the sergeant’s list. (There were 11 sergeants promoted that year.) Two other black officers (Walters and Lockhart) made 99 on the written exam but did not make the certified list. In 1979, appellant Bryant scored 97 on the written test, but was not placed on the list. There was evidence that the procedures did not preclude oral panelists’ access to a candidate’s score on the written examination. Moreover, the relative weight accorded each component fluctuated from year to year. Standardized responses were not introduced into the oral interview process until 1979.
We note that all these elements of the promotional process, the oral interview, the performance appraisals, the chief’s rating, and the relative weight of each component involve decisions made by white supervisors on the basis of largely subjective criteria and must therefore be closely scrutinized because of their susceptibility to discriminatory abuse. Paxton, 688 F.2d at 563 n. 15; Coble, 682 F.2d at 726.
Because we believe that the district court’s focus in analyzing the statistical evidence in this case was not correct, we remand for further proceedings so that the district court may address the specific concerns we have identified. Upon remand it is incumbent upon the district court to consider the statistical evidence against the background of the evidence on racial harassment at the Little Rock Police Department. Although we affirm the district court’s finding that appellants did not prove a Title VII violation based on a discriminatory work place theory, clearly evidence of a racial atmosphere is relevant to appellants’ claim of discrimination in the promotional process.
Recusal
John Walker was the original attorney representing appellants in this case. In November 1979, the Honorable Elsijane T. Roy, the district judge in the instant case, entered an order in an unrelated case recus-ing herself in all cases involving the John Walker law firm because of comments made to her by Walker. In September 1980, one year before this case was tried, Walker was replaced by appellants’ present attorney. Based on the 1979 order of recu-sal, appellants filed a motion for order of recusal by Judge Roy in the present case on the ground that Walker and one of his former associates were expected to be called as witnesses. Judge Roy denied the motion. This motion was renewed after judgment in appellants’ motion for new trial, which was denied.
Title 28 U.S.C. § 144 provides for the recusal of a district judge whenever a party to a proceeding files an affidavit that the judge before whom the matter is pending has a personal bias or prejudice either against the party or in favor of an adverse party. The term “party” as used in this section does not include counsel as such. Davis v. Board of School Commissioners, 517 F.2d 1044 (5th Cir.1975) (Davis), cert. denied, 425 U.S. 944, 96 S.Ct. 1685, 48 L.Ed.2d 188 (1976). Antipathy to an attorney is insufficient grounds for disqualification of a judge because it is not indicative of extrajudicial bias against a “party.” United States v. International Business Machines Corporation, 475 F.Supp. 1372, 1383 (S.D.N.Y.1979).
*1399Similarly, under 28 U.S.C. § 455(a), which requires that a federal judge disqualify himself or herself “in any proceeding in which his [or her] impartiality might reasonably be questioned,” a controversy between a trial judge and an attorney for parties to an action would not require disqualification of the judge in absence of showing of bias or personal prejudice to the parties. Davis, 517 F.2d at 1052.
The standard for determining the appearance or fact of particular grounds for recusal or disqualification of a trial judge is the same under both statutes. A judge is to take into consideration all circumstances both public and private and determine if a reasonable, uninvolved observer would question the judge’s impartiality. Idaho v. Freeman, 507 F.Supp. 706 (D.Idaho 1981). A recusal or disqualification motion is committed to the sound discretion of the trial judge and the standard of review on appeal is whether the judge abused his or her discretion. In re Federal Skywalk Cases, 680 F.2d 1175, 1183 (8th Cir.1982), cert. denied, - U.S. -, 103 S.Ct. 342, 74 L.Ed.2d 383 (1983). No such abuse is found here.
Class Certification
On the last day of trial, appellants moved for class certification, for promotion purposes only, in order that a former black police officer, who did not file an individual claim, could be included in the suit. Appellants admit that the requirements of Fed.R. Civ.P. 23 have not been met. A district court has broad discretion in determining whether to certify a class, and its determination will not be overturned absent a showing that it abused its discretion. Shapiro v. Midwest Rubber Reclaiming Co., 626 F.2d 63, 71 (8th Cir.1980), cert. denied, 449 U.S. 1079, 101 S.Ct. 860, 66 L.Ed.2d 802 (1981). We find no abuse of discretion in the district court’s denial of appellants’ motion for class certification.
The judgment is reversed and remanded on the issue of discrimination in promotions. The judgment is affirmed in all other regards.
. The Honorable Elsijane T. Roy, United States District Judge for the Eastern District of Arkansas. The opinion of the district court is reported at 544 F.Supp. 1231 (E.D.Ark.1982).
. Another intervenor, Finis Lowe, took a non-suit at the close of plaintiffs’ case. At the time this action was instituted, the Little Rock Police Department had 206 uniformed personnel of whom 14 were black. Thus the plaintiffs and intervenors represented almost the entire black contingent of uniformed officers.
.For a discussion of the other statutory provisions in the complaints and their relationship to the Title VII claim, see the district court’s opinion, 544 F.Supp. at 1236-38.
. This rating factor was not used after 1978.
. In addition, two blacks (appellants Walters and O’Donald) and six whites were promoted to acting sergeants. Their status, however, could not be changed to regular sergeant without participating in the promotion process again.
.
[[Image here]]
. The 1974 sergeant list was effective April 1974-April 1975. One black officer, appellant Gilbert, was on the list. He ranked seventh. Six promotions were made from the list, all in the first four months of the list’s life. Thus, the one black candidate on the list was in the highest rank for eight months without being promoted before the list expired. One month later, in May 1975, the next vacancy arose. Appellant Gilbert testified that the white officer whose position became available in May 1975 planned to leave several months earlier but was asked to stay until the certified list for that year expired.
Although any discrimination in the promotion of police officers from April 1974-April 1975 would have no present legal consequences because not made the basis of a timely charge, it may constitute relevant background evidence in a proceeding in which the status of a practice within the relevant time period is at issue. United Airlines, Inc. v. Evans, 431 U.S. 553, 558, 97 S.Ct. 1885, 1889, 52 L.Ed.2d 571 (1977).
. Sergeant list:
[[Image here]]
Appellant Anthony ranked fourth on the 1978 lieutenant list. Two promotions were made from that list.