Hershel CLADY, Et Al., Plaintiffs-Appellants, v. COUNTY OF LOS ANGELES, Defendant-Appellee

EUGENE A. WRIGHT, Circuit Judge: This class action for damages and injunctive relief under Title VII of the Civil Rights Act of 1964, as amended, 42 U.S.C. §§ 2000e et seq. (Title VII), and 42 U.S.C. § 1981 challenges the 1979 selection procedures used by the Los Angeles County Fire Department. Plaintiffs allege that the written examination, education requirement (ER), and physical agility test (PAT) had an adverse impact on black and Hispanic applicants. Following a seven day bench trial, the district court entered judgment for the County of Los Angeles (County). We affirm the judgment. FACTS AND PROCEEDINGS BELOW To understand the parties’ positions, we provide in some detail the history of prior legal challenges to the County’s firefighter selection procedures. 1969 Exam In 1969, firefighter applicants were required to take a civil service examination and a physical agility test. They were ranked according to their performance on both tests. Following successful job interviews, candidates were placed on a hiring eligibility list. The County voluntarily discontinued these procedures after it was determined that they had a disparate adverse impact on minority hiring. See County of Los Angeles v. Davis, 440 U.S. 625, 627-28, 99 S.Ct. 1379, 1381-82, 59 L.Ed.2d 642 (1979) (describing 1969 selection procedures). 1972-73 Exam Hiring procedures were changed in 1971 but court challenges prevented their implementation. The County proposed to eliminate ranking candidates by written exam score. A new written exam was designed to eliminate cultural bias. It was to be graded on a pass/fail basis. Five hundred applicants who passed the written exam were to be selected at random for oral interviews and physical agility tests. The candidates then would be ranked for appointment to available openings. *1424 In January 1972, the County conducted a written examination pursuant to this plan. Ninety-seven percent of the applicants passed. Before further selection was made, a lawsuit was filed charging that the random selection procedure violated civil service regulations and the county charter. The County was enjoined from further selection of firefighters. When staffing needs became critical, the County proposed to interview the top 544 applicants based on written exam scores. Of this number, 492 were white, 10 black, and 33 Mexican-American. Minority applicants objected and this plan, too, was abandoned. In 1973, unsuccessful black and Mexican-American firefighter applicants brought a Title VII class action against the County, its Board of Supervisors, and its Civil Service Commission challenging selection procedures used in 1969 and 1972-73. District Judge William P. Gray held for the plaintiffs, enjoining future discrimination and mandating good faith affirmative action efforts. Davis v. County of Los Angeles, 8 Fair Empl.Prac.Cas. (BNA) 239 (C.D.Cal.1973), aff 'd in part, rev’d in part, 566 F.2d 1334 (9th Cir.1977), vacated as moot, 440 U.S. 625, 99 S.Ct. 1379, 59 L.Ed.2d 642 (1979). Judge Gray found that while minorities comprised 29.1% of L.A. County, they comprised 3.3% of the firefighters. Id. at 240. Moreover, the County failed to validate the 1972 exam as predictive of job performance. Id. at 240-41. The court entered a remedial hiring order requiring that at least 20% of all new recruits be black and 20% be Mexican-American until their percentages equalled the corresponding percentages for those minorities in Los Angeles County. Id. at 242. Between 1973 and 1979 when Judge Gray’s order was in effect, all applicants who had passed the 1972 written exam were interviewed. An eligibility list was formulated using the oral interview as a ranking device. The civil service rule then in effect, the “rule of three,” required the County to consider only the top three candidates on the eligible list for the next firefighter opening. Pursuant to the court order, the County permitted the first of the top three blacks and the first of the top three Mexican-American candidates to be certified as eligible for any given opening. These procedures yielded a 55% minority hiring level during this period. Judge Gray’s injunction was dissolved by the Supreme Court in March 1979. Specifically, the Court held that any discrimination from the 1972 selection procedures had been vitiated by the more than five years of hiring under Judge Gray’s order. Davis, 440 U.S. at 633, 99 S.Ct. at 1384. 1979-80 Exam When Judge Gray’s order was vacated, the County developed the selection procedures at issue here. The process began by posting bulletins which announced filing dates, minimum job qualifications and examination procedures. Persons who had filed “notice of interest” cards with the County since the last application period also were notified of openings. In addition to the recruitment efforts by County personnel, defendant contracted with an outside organization, the International Association of Firefighters, to recruit and prepare minority candidates. Applicants had to be at least 18 years of age, but not older than 29. They had to have a high school diploma or general equivalency degree (GED). A written test was administered to all applicants who appeared on the scheduled testing date and who met these minimum requirements. The 1979 written exam measured mechanical comprehension, spatial perception, and verbal ability. Applicants who scored less than 53 correct out of 95 multiple-choice questions were excluded from further consideration. Next was the administration of an oral interview to all applicants qualifying on the written exam. At that time, a panel of three persons assessed the applicant’s education, experience and personal suitability for performing firefighter duties. *1425 An “eligible list” was then created which ranked the applicants who qualified on both the written test and oral interview. The relative ranking was determined by weighting the written exam score by 25% and the oral interview by 75%. A ten point credit was added for qualifying military veterans. As openings arose, the candidates were administered a physical agility test on a pass/fail basis. It consisted of six parts, 1 all of which had to be passed. Selection of candidates from the eligible list was made in accordance with the “rule of three” civil service requirement. In April 1981, the “rule of three” was replaced with a new civil service rule which authorized a “banding” system for selecting candidates from an eligible list. Banding permitted discretion to select for hire any candidate whose final score was within a band of scores. 2 Appointments were made from the highest ranking band until fewer than five persons were available for appointment from that band. Following appointment from the eligible list, all newly hired firefighters were required to complete a training program at the Fire Academy and a one year probation period. Present Lawsuit In October 1979, plaintiffs filed this lawsuit alleging a pattern of illegal discrimination by the County based on race, color and national origin. Pretrial orders limited this action to the 1979 firefighter selection procedures. In May 1980, the court conditionally certified a plaintiff class of: (1) All black and Mexican-American applicants for the position of firefighter with the Los Angeles County Fire Department whose applications were processed under the 1979 selection procedures and who were not hired by defendant; and (2) All black and Mexican-American applicants for the position of firefighter with the Los Angeles County Fire Department who will apply in the future. Plaintiffs’ motion for preliminary injunction was granted in April 1981, ordering defendant to select 35% blacks and Hispanics 3 for any future classes. Seven recruit classes, totalling 280 persons, were hired pursuant to the 1979 selection procedures. The parties stipulated that the selection rate for blacks and Hispanics was 18% for the first four classes. This rate jumped to 40% for the last three classes hired pursuant to Judge Lucas’ preliminary injunction and the civil service banding rule. The County conducted extensive validation studies covering the period from the imposition of Judge Gray’s order until commencement of this action. The three studies were: (1) Firefighter Validation Study: Los Angeles County, dated February 1980 (Study); (2) 1979 Firefighter Validation Study Supplement: Los Angeles County, dated June 1980 (Supplement); and (3) Analysis of the Job Relatedness of Academy and Recruit Follow-Up Examinations at the Los Angeles County Fire Department, dated June 8, 1981 (Academy Study). The Study compared test scores from the written exam received by the first three *1426 recruit classes with their performance on four criteria: (1) academy scores, (2) station scores, (3) deficiency points, and (4) academy graduation. 4 Through its validation studies, the County sought to demonstrate a significant correlation between the exam components and important elements of firefighter training. Correlation coefficients 5 derived for the 1979 written test were: -.7162 for deficiency points .6435 for Academy score .3838 for graduation/termination Sample size was insufficient to obtain a significant correlation for station score. The County’s personnel analyst who had primary responsibility for conducting the studies testified that the negative number on deficiency points meant that those who had more points had lower test scores. Other witnesses for the County testified that all three studies met professionally accepted standards. Plaintiffs did not contest the appropriateness of the performance criteria against which the exam was correlated. The Supplement included a utility analysis which estimated that use of the written exam would result in an increase in the selection of qualified recruits from 9% to 48%. Based on this estimate, the County determined that it could save between $39,-000 and $180,000 in direct training costs per class. 6 The Academy Study concluded that the Academy training tests and the recruit follow-up exams “were representative of important or critical work behaviors performed by Firefighters.” Academy Study, Defendant’s Exhibit 202, at 14. Trial Plaintiffs’ case in chief was brief. There was no direct evidence of discriminatory treatment of blacks or Hispanics. The entire case was founded on stipulated personnel department data. Plaintiffs relied on Dr. James J. Kirkpatrick, an industrial psychologist and experienced expert witness, to interpret the stipulated facts. They also called the County’s chief personnel analyst, Frank V. Frazier, and Fire Chief Clyde A. Bragdon as adverse witnesses. Defendant called 13 expert and lay witnesses, mostly Fire Department personnel, to prove job-relatedness of the selection devices and absence of legally sufficient alternative selection procedures. The court concluded that plaintiffs failed to establish a prima facie case of disparate impact for the overall 1979 selection procedures. The court further concluded that plaintiffs failed to establish disparate impact against blacks and Hispanics for the ER and PAT. Nor did plaintiffs prove disparate impact against Hispanics for the written exam. The only issue on which plaintiffs established a prima facie case was disparate impact against blacks on the written exam as a discrete selection device. However, the court found that the County carried its burden of proving the exam’s job-relatedness. Plaintiffs failed to identify alternative selection devices that would serve the County’s hiring needs with less adverse impact. Finally, the court concluded plaintiffs lacked standing to challenge any subsequent firefighter exam. STANDARD OF REVIEW Appellants assert that the findings below are not to be accorded the same weight as *1427 findings written by the court. They allege that Judge Lucas incorporated 171 of 178 sub-parts from the County’s proposed findings. Only certain proposed credibility determinations were rejected. The recent Supreme Court decision in Anderson v. City of Bessemer, 105 S.Ct. 1504 (1985), is instructive. There, the district court adopted petitioner’s findings and conclusions following issuance of a preliminary memorandum. The Court rejected the Fourth Circuit’s suggestion that the record should be closely scrutinized. Id. at 1510-11. Because the party’s findings were not “uncritically accepted,” the clearly erroneous standard was retained. Id. at 1511. Similarly, we note that Judge Lucas did not uncritically accept the County’s proposed findings. He was careful to reject certain credibility findings with which he disagreed. Under these circumstances, “[tjhere is no reason to subject those findings to a more stringent appellate review than called for by the applicable rules.” Id. We review the district court’s findings of fact under the clearly erroneous standard of review. Id. at 1511-13; Fed.R.Civ.P. 52(a). “If the district court’s account of the evidence is plausible in light of the record viewed in its entirety, the court of appeals may not reverse it even though convinced that had it been sitting as the trier of fact, it would have weighed the evidence differently.” Id. at 1512. It is uncertain what standard of review applies to plaintiffs’ establishment of a prima facie case. Compare Peters v. Lieuallen, 746 F.2d 1390, 1392 (9th Cir. 1984) (clearly erroneous) with Moore v. Hughes Helicopters, Inc., 708 F.2d 475, 480-81 (9th Cir.1983) (de novo). We believe de novo review is compelled because “the question requires us to consider legal concepts in the mix of fact and law.” United States v. McConney, 728 F.2d 1195, 1202 (9th Cir.) (en banc), cert. denied, — U.S. -, 105 S.Ct. 101, 83 L.Ed.2d 46 (1984). ANALYSIS Title VII “proscribes not only overt discrimination but also practices that are fair in form, but discriminatory in operation.” Griggs v. Duke Power Co., 401 U.S. 424, 431, 91 S.Ct. 849, 853, 28 L.Ed.2d 158 (1971). It does not mandate special treatment in hiring merely because a person “was formerly the subject of discrimination, or because he is a member of a minority group.” Id. Indeed, “[discriminatory preference for any group, minority or majority, is precisely and only what Congress has proscribed.” Id. To succeed under disparate impact theory, appellants must establish that a facially neutral employment practice produces a significant adverse impact on a protected class. Connecticut v. Teal, 457 U.S. 440, 446, 102 S.Ct. 2525, 2530, 73 L.Ed.2d 130 (1982); Dothard v. Rawlinson, 433 U.S. 321, 329, 97 S.Ct. 2720, 2726, 53 L.Ed.2d 786 (1977). Appellants can meet this burden through reliable statistics. See Spaulding v. University of Washington, 740 F.2d 686, 703 (9th Cir.), cert. denied, — U.S.-, 105 S.Ct. 511, 83 L.Ed.2d 401 (1984). Intent to discriminate need not be shown. International Brotherhood of Teamsters v. United States, 431 U.S. 324, 335 n. 15, 97 S.Ct. 1843, 1854 n. 15, 52 L.Ed.2d 396 (1977). If a plaintiff fails to meet the initial burden of establishing a prima facie case of substantial adverse impact, validity of the selection device is irrelevant. Moore v. Southwestern Bell Telephone Co., 593 F.2d 607, 608 (5th Cir.1979) (per curiam). If substantial impact is demonstrated by a preponderance of the evidence, the burden shifts to the employer to validate the selection device, that is, to show that it has “a manifest relationship to the employment in question.” Griggs, 401 U.S. at 432, 91 S.Ct. at 854. When scored tests are involved, this circuit has required a showing that the test actually measure “skills, *1428 knowledge or ability” required for successful performance on the job. Contreras v. City of Los Angeles, 656 F.2d 1267, 1283 (9th Cir.1981), cert. denied, 455 U.S. 1021, 102 S.Ct. 1719, 72 L.Ed.2d 140 (1982). Contra Rivera v. City of Wichita Falls, 665 F.2d 531, 538 n. 9 (5th Cir.1982) (test is legally sufficient if it validly predicts performance at the training academy). If the employer fails to meet its burden, use of the selection device will be deemed a Title VII violation. If it succeeds, the plaintiff then may attempt to rebut the defendant’s evidence by showing that although job-related, the test does not constitute a business necessity because an alternative selection device exists which would have comparable business utility and less adverse impact. Albemarle Paper Co. v. Moody, 422 U.S. 405, 425, 95 S.Ct. 2362, 2375, 45 L.Ed.2d 280 (1975). I. Impact Analysis Appellants urge rigid adherence to the so-called “80% rule” found in the Uniform Guidelines on Employee Selection Procedures (Uniform Guidelines), 29 C.F.R. Part 1607 (1984). The Uniform Guidelines apply to any selection procedure “used as a basis for making employment decisions.” Id., § 1607.2(C). For determining adverse impact, the Uniform Guidelines set 80 percent as a “rule of thumb.” Thus, a scored test or other selection device normally is considered to have adverse impact if it has a “selection rate for any race, sex, or ethnic group which is less than four-fifths (4/5) (or eighty percent) of the rate of the group with the highest rate____” Id., § 1607.-4(D). The rule does not require strict adherence. Smaller differences in selection rate may nevertheless constitute adverse impact, where they are significant in both statistical and practical terms or where a user’s actions have discouraged applicants disproportionately on grounds of race, sex, or ethnic group. Greater differences in selection rate may not constitute adverse impact where the differences are based on small numbers and are not statistically significant, or where special recruiting or other programs cause the pool of minority or female candidates to be atypical of the normal pool of applicants from that group. Id. The Uniform Guidelines are not legally binding. See General Electric Co. v. Gilbert, 429 U.S. 125, 141-42, 97 S.Ct. 401, 410-11, 50 L.Ed.2d 343 (1976). They have not been promulgated as regulations and do not have the force of law. Aquilera v. Cook County Police and Corrections Merit Board, 760 F.2d 844, 847 (7th Cir.1985). Additionally, the 80 percent rule has been sharply criticized by courts and commentators. See E. Shoben, Differential Pass-Fail Rates in Employment Testing: Statistical Proof Under Title VII, 91 Harv.L. Rev. 793, 805 (1978) (ill-conceived rule capable of producing anomalous results because it fails to take account of differences in sampling size). There is no consensus on a threshold mathematical showing of variance to constitute substantial disproportionate impact. Some courts have looked to Castaneda v. Partida, 430 U.S. 482, 496-97 n. 17, 97 S.Ct. 1272, 1281-82 n. 17 n. 17, 51 L.Ed.2d 498 (1977), which found adverse impact where the selection rate for the protected group was “greater than two or three standard deviations” from the selection rate of their counterparts. See, e.g., Rivera, 665 F.2d at 536-37 n. 7; Guardians Association of the New York City Police Dept. v. Civil Service Commission, 630 F.2d 79, 88 (2d Cir.1980), cert. denied, 452 U.S. 940, 101 S.Ct. 3083, 69 L.Ed.2d 954 (1981). Others have applied the 80 percent rule of the Uniform Guidelines. See, e.g., Firefighters Institute for Racial Equality v. City of St. Louis, 616 F.2d 350, 356-57 (8th Cir.1980), cert. denied, 452 U.S. 938, 101 S.Ct. 3079, 69 L.Ed.2d 951 (1981) (finding adverse impact where black selection rate was between 42.5% and 44.4% of whites). Without endorsing either method, this circuit analyzes whether the statistical disparity is “substantial” or “significant” in a *1429 given case. Contreras, 656 F.2d at 1274-75 (passing ratio of 2.45:1 established adverse impact); Craig v. City of Los Angeles, 626 F.2d 659, 661-62 (9th Cir.1980), cert. denied, 450 U.S. 919, 101 S.Ct. 1364, 67 L.Ed.2d 345 (1981) (failure ratio of 2.5:1 held to establish adverse impact). 7 The plaintiffs concede that there is no “bottom line” adverse impact on blacks and Hispanics when evaluating the total selection process. Significantly, the district court found that there was “[n]o statistically significant difference between hires of Blacks and Hispanics and their representation in the Civilian Labor Force____” Finding No. 14(a)(1). However, in Connecticut v. Teal, the Supreme Court held that discrete pass/fail barriers having adverse impact must be individually validated even if there is no “bottom line” disparate impact for the entire multicomponent selection process. 8 457 U.S. at 442, 452, 102 S.Ct. at 2528, 2533. Here, the written exam, education requirement (ER), and physical agility test (PAT) are pass/fail barriers under Teal. Accordingly, we must evaluate each device separately for adverse impact. If plaintiffs establish a prima facie case as to any discrete selection device, we must consider whether defendant met its burden of proving job-relatedness. A. Written Exam Considering the written exam as a discrete selection device, the district court found no actionable adverse impact for Hispanics while finding actionable impact for blacks. The parties disagree over the proper method to compute the exam’s impact. Plaintiffs use failure rates and figures based on self-identification by racial codes. The district court adopted defendant’s use of pass rates and figures based on visual survey and surname review by the test monitors. The tests were coded using the County’s method, which is the only method of comparing tests over time. The figures are as follows: Whites Blacks Hispanics Took Written Test 1816 525 436 Passed Written Test 1375 292 148 Passing Rate 75.7% 55.6% 33.9% Appellants failed to prove that the court’s method of comparison was clearly erroneous. (1) Hispanics Using the 80 percent rule, the court found that the difference between the passing rate achieved by Hispanics and 80 percent of the passing rate for whites was only 5 percent. Computing five percent of 525 Hispanic examinees, the court found the “relative small number differences ... [to be] unreliable as evidence of actionable adverse impact.” Finding No. 15(f). Appellants assert that the numerical differences here are large and statistically significant. They compare the exam pass rates here with those found to be actionable adverse impact in Craig, 626 F.2d at 661 n. 3. In that case, the pass rate for Mexican-Americans taking a test to select Deputy Sheriffs was 55 percent (compared to 55.6 percent here), while the white passing rate was 77 percent (compared to 75.7 percent here). Id. We find the appellants’ arguments persuasive. The court erred in finding no actionable adverse impact against Hispanics considering the written exam as a pass/fail device. Appellants also attack use of the written exam as a ranking device. However, they fail to prove that the district court’s finding 15(i) 9 is clearly erroneous. *1430 (2) Blacks Appellee does not dispute the finding of actionable impact against blacks taking the written exam. The battleground of this appeal concerns validation and less discriminatory alternatives, discussed later. B. Education Requirement (ER) The district court found that Plaintiffs have failed to show that the education requirement has an adverse impact on Blacks and Hispanics, whether in terms of actual applicants who were disqualified by the educational requirement or potential applicants who otherwise would have applied for the 1979 Firefighter Examination but for the educational requirement. Finding No. 20. To refute the court’s detailed findings concerning the ER, appellants rely on generalized census figures reflecting percentages of Californians with high school degrees. These statistics are not differentiated by age and, therefore, cannot be used to establish disparate impact for the ER. See Moore, 708 F.2d at 482-83 (proxy pool must be that of the local labor force possessing the requisite skills); Pack v. Energy Research and Development Administration, 566 F.2d 1111, 1113 (9th Cir.1977) (per curiam) (labor pool must possess requisite skills). Moreover, plaintiffs’ expert, Dr. Kirkpatrick, testified at trial that the census figures included illegal aliens, who could not be employed as firefighters. C. Physical Agility Test (PAT) The district court found that plaintiffs failed to show that the 1979 PAT had any adverse impact on Hispanics, the only group challenging it. Finding No. 23. The court further found that the statistical analysis performed by plaintiffs’ expert was unreliable. Appellants concede that the Hispanic pass rate was 81 percent of the white pass rate. They argue, however, that if only 15 more Hispanics had passed, their pass rate would increase by 25 percent. Appellants’ challenge to the PAT must fail. The numerical differences are too small for meaningful comparison. Contreras, 656 F.2d at 1273 n. 4 (statistics untrustworthy when minor numerical variations produce significant percentage fluctuations); see also Morita v. Southern California Permanente Medical Group, 541 F.2d 217, 220 (9th Cir.1976), cert. denied, 429 U.S. 1050, 97 S.Ct. 761, 50 L.Ed.2d 765 (1977) (statistical evidence derived from small universe has little predictive value and must be disregarded). II. Job-Relatedness Once plaintiffs have established a prima facie case of disparate impact for both blacks and Hispanics taking the written exam, the burden shifts to the County to prove the exam’s job-relatedness. This circuit has held that the employer’s burden is satisfied by showing “by professionally acceptable methods” that the test is “predictive of or significantly correlated with important elements of work behavior which comprise or are relevant to the job or jobs for which candidates are being evaluated.” Contreras, 656 F.2d at 1280. See also 42 U.S.C. § 2000e-2(h) (Title VII expressly authorizes the use of “professionally developed” scored tests). “A number of validation techniques are professionally acceptable.” Craig, 626 F.2d at 664. Some courts use the detailed technical standards for validity studies set forth in the Uniform Guidelines. See 29 C.F.R. § 1607.14. It is undisputed that the County’s validation studies in the present case do not meet the Guideline’s standards. Few validation studies do, however. See C. Richey, Manual on Employment Discrimination Law and Civil Rights Actions in the Federal Courts, August 1984, at A-44. This court has held that while noncompliance is “not necessarily fatal,” it “diminishes the probative value of the defendants’ validation study.” Craig, 626 F.2d at 665. *1431 The Supreme Court has approved validation studies which did not meet the standards of the Uniform Guidelines. 10 Appellants attempt to impeach the validation studies’ reliability because they were prepared as a defense to litigation. As the County noted, it was impossible to assess the written exam’s validity until after it was given and job performance of successful applicants measured. Validation studies are time-consuming and expensive. Teal, 457 U.S. at 463, 102 S.Ct. at 2539 (Powell, J., dissenting). Preparation for litigation should not affect their reliability. See Albemarle, 422 U.S. at 433 n. 32, 95 S.Ct. at 2379 n. 32; Contreras, 656 F.2d at 1283. Appellants also raise what they allege to be fatal flaws with the studies: (1) failure to utilize a representative sample, (2) failure to control for racial bias by training captains at the Fire Academy, and (3) failure to perform differential validation for blacks and Hispanics. These claims must fail. Appellants improperly rely on Blake v. City of Los Angeles, 595 F.2d 1367 (9th. Cir.1979), cert. denied, 446 U.S. 928, 100 S.Ct. 1865, 64 L.Ed.2d 281 (1980), for their contention that a study must include those who failed. Blake was criticized in Craig, which approves of restricting the sample to those who passed. See Craig, 626 F.2d at 665 (sheriff’s department not required to hire and train a sample of failing applicants to provide statisticians with more certain results). Concerning racial bias at the Academy, the district court relied on the credible testimony of defendant’s expert, Ms. Lorrayne Alstadt, who found no bias. Finding No. 17(d)(2)(b). The County established that subjective evaluation in Academy grading was minimized through written standards for scoring drills. See Albemarle, 422 U.S. at 432-33, 95 S.Ct. at 2378-79 (vague instructions to supervisors too subjective). Appellants cannot show that the finding of no racial bias is clearly erroneous. Finally, appellants rely on their expert, Dr. Kirkpatrick, for their contention that differential validation should have been performed for blacks and Hispanics. The district court rejected Dr. Kirkpatrick’s testimony as “contradictory; unsupported by sufficient data, analyses of data, or relevant authority; contra the consensus of professionals in his field of expertise; and contra the opinions of defendant’s witnesses, both expert and lay, which were supported by substantial relevant experience, sufficient data, analyses of data, or relevant authority.” Finding No. 17(d)(3)®. As the Supreme Court recently wrote: [W]hen a trial judge’s finding is based on his decision to credit the testimony of one of two or more witnesses, each of whom has told a coherent and facially plausible story that is not contradicted by extrinsic evidence, that finding, if not internally inconsistent, can virtually never he clear error. Anderson, 105 S.Ct. at 1513. Moreover, appellants’ insistence on differential validation lacks support in the technical community. See B. Schlei & P. Grossman, Employment Discrimination Law 154 n. 151 (2d ed. 1983). Defendant’s evidence demonstrated that the Academy teaches principles and techniques of firefighting including hose lays, ladder techniques, use of ropes and knots, use of breathing apparatus, salvage and rescue techniques, and emergency medical training. These are the basic skills utilized on the job. Moreover, department policy requires that these basic skills be drilled by firefighters for two hours each day. In conclusion, the County’s validation studies demonstrate legally sufficient correlation to success at the Academy and performance on the job. Courts generally *1432 accept correlation coefficients above +.30 as reliable. Id. at 129 n. 131; see Craig, 626 F.2d at 664 (accepting correlation coefficient of .60 as significant). As a general principle, the greater the test’s adverse impact, the higher the correlation which will be required. See, e.g., Guardians Association, 630 F.2d at 105-06. Here, the correlation coefficients are statistically significant and can be considered reliable under prevailing legal principles. III. Alternative Selection Devices Appellants assert that they identified two alternative selection devices which met the County’s legitimate hiring needs with less adverse impact on protected minorities: 1) the 1973-78 procedures, and 2) the banding procedure adopted in April 1981. Documentation of the County’s search for alternative selection devices is extensive. The County solicited information about selection procedures from 57 county and 44 city fire departments statewide. They surveyed professionally developed tests. Finally, they investigated alternative means of using the same procedures. An employer is not required to choose the selection device with the least adverse impact on minority applicants. Id. at 110. Appellants contend they met the “virtually impossible” burden of identifying alternative selection devices which satisfy the employer’s legitimate hiring needs. A. 1973-78 Procedures The district court held that plaintiffs’ “unspecified” alternative procedures did not meet the County’s needs. Finding No. 19(c). This finding is not clearly erroneous. 11 Appellants rely on testimony from defendant’s Fire Chief that recruits hired under Judge Gray’s hiring order were equally as competent as those hired under the 1979 selection procedures. This testimony falls short of establishing legally sufficient alternative selection procedures. Appellee introduced evidence that the applicant population in the quota years was of higher quality than in 1979 due to the larger percentage of veterans (71% as opposed to 43% in 1979), who are well suited to the discipline of fire departments. The disparity between selection rates is partially explained by a high quality applicant pool during the quota years and an “atypical” pool during the period at issue here. It is undisputed that the applicant pool for the 1979 exam was atypical due to special minority recruitment efforts. The percentage of minority applicants was almost double the corresponding percentage in the Civilian Labor Force. Plaintiffs’ expert described the “stimulation effect” on minority applications created by the quota hiring years before 1979. Greater differences in the selection ratios are tolerated for an atypical applicant pool. 29 C.F.R. § 1607.4(D) (1984). Moreover, there was considerable trial testimony concerning the cost savings of the 1979 procedures. Financial concerns are legitimate needs of the employer. See Chrisner v. Complete Auto Transit, Inc., 645 F.2d 1251, 1263 (6th Cir.1981) (marginal cost of another hiring policy and implications for public safety are factors to be considered). The 1972-73 exam was shown to have less validity and greater financial impact than the 1979 exam. Moreover, the 1979 exam improved the chances of selecting qualified recruits. *1433 Appellants’ reliance on Blake for the proposition that compliance with Title VII may require modification of departmental policies is misplaced. Contreras recites the proper standards. In that case, the plaintiffs relied solely upon expert testimony that Hispanics generally do better in oral interviews than on written examinations. This court held that the use of oral interviews failed to satisfy the City’s civil service hiring needs. Contreras, 656 F.2d at 1285 and n. 11. Similarly, appellants cannot establish that the procedures used from 1973-78 satisfied the County’s civil service hiring needs in 1979. Lastly, we note that plaintiffs’ own expert criticized the validity of the oral interview when - used as a ranking device. When Judge Gray’s hiring order was in effect, final eligibility was based solely on ranked oral interview scores. B. 1981-82 Procedures Appellants also advance the banding system as a second less discriminatory alternative. They correctly assert that the court’s findings fail to address this alternative. District court findings are adequate if they are sufficiently comprehensive to provide a basis of decision and are supported by the record. See Irish v. United States, 225 F.2d 3, 8 (9th Cir.1955). Here, appellants failed to present evidence on the impact of the banding rule alone. The imposition of the civil service banding rule coincided with Judge Lucas’ injunction order. At oral argument, appellants conceded that it is impossible on the present record to evaluate the banding rule’s impact on minority hiring. 12 In conclusion, appellants have not sustained their burden of identifying alternative selection devices to the 1979 written exam that met the County’s legitimate hiring needs. IV. Evidentiary Rulings Lastly, appellants argue that the court erred in excluding defendant’s Exhibit 248 from evidence. This exhibit contains a legal opinion that the 1979 selection procedures had an adverse impact on minority applicants. Evidentiary rulings are not reversible absent clear abuse of discretion. See M.A.P. Oil Co., Inc. v. Texaco, Inc., 691 F.2d 1303,1310 (9th Cir.1982). We find no abuse of discretion. In May 1982, U.S. Magistrate James Penne ruled that the portion of Exhibit 248 entitled “Firefighter — EEO Liability Analysis” was not discoverable because the attorney-client privilege attached to it. He found also that the document had been produced inadvertently during discovery and that the privilege had not been waived. It is undisputed that the appellants never appealed the magistrate’s ruling. See Rule 4.1, Local Rules Governing the Duties of the Magistrates of the United States District Court, Central District of California (1981) (ten days for appeal of magistrate’s order on non-dispositive motion). At trial, the plaintiffs sought to elicit testimony about the exhibit from the author, Frank Frazier. The plaintiffs asserted that the document was included in the County’s exhibits offered at trial. The court sustained the defendant’s objection on the ground that the plaintiffs’ challenge to the magistrate’s ruling was not timely presented. Voluntary disclosure of a privileged attorney communication constitutes waiver. Weil v. Investment/Indicators, Research and Management, Inc., 647 F.2d 18, 24 (9th Cir.1981). Inadvertence “does not as a matter of law prevent the occurrence of waiver.” Id. It is only one factor to be considered. However, the County denies that the challenged exhibit was part of its case. Nor was the exhibit admitted into evidence on the first day of trial as appellants as *1434 sert. The court noted that the exhibit was undated, unsigned and unaddressed. In sum, appellants fail to allege facts or other proof that the court abused its discretion in rejecting the exhibit. 13 CONCLUSION Appellants asserted at oral argument that they should prevail because they established a prima facie case of adverse impact and the existence of an alternative selection device that selected equally competent firefighters. We disagree. Through a meticulous validation analysis, conducted as soon as the 1979 recruit classes could be tested, the County met its burden of showing that the challenged exam was sufficiently correlated with successful job performance. The County also demonstrated that the exam met its needs by accurately screening candidates on the basis of their proficiency in a relatively inexpensive fashion. Finally, plaintiffs failed to establish that less discriminatory alternatives available to the County were equally tailored to its legitimate hiring needs. “Congress did not intend by Title VII, however, to guarantee a job to every person regardless of qualifications.” Griggs, 401 U.S. at 430, 91 S.Ct. at 853. “The touchstone [of Title VII] is business necessity.” Id. at 431, 91 S.Ct. at 854. Unlike the situation in Griggs, the County adopted the challenged procedures after meaningful study of their relationship to job performance ability. Appellants would have us overlook those efforts. In conclusion, appellants established a prima facie case of disparate impact against blacks and Hispanics for the written exam. However, they failed to demonstrate that the court’s finding of job-relatedness was clearly erroneous. Nor did they establish an alternative selection device having comparable business utility and less adverse impact. AFFIRMED. Footnotes [1] The PAT was composed of six parts: 1) a hose line extension task, 2) hydrant manipulation, 3) ladder climb, 4) hose loading, 5) hose tower rope pull, and 6) a hose and ladder storing exercise. [2] The hiring bands, in order, were these: (1) 100 plus (called the veteran’s band because only those with the 10 point veteran credit can score over 100); (b) 95-100; (c) 89-94; (d) 83-88; (e) 77-82; and (f) 70-76. [3] We note that Mexican-Americans are a class within the larger Hispanic population. The complaint was filed on behalf of black and Mexican-American applicants, and the plaintiff class was certified on this basis. As the case proceeded to trial, however, the focus of impact analysis shifted from Mexican-Americans to Hispanics. [4] Academy score was based on four written tests given early in the training program at the academy. Station score represented the score received on a written exam taken six months after becoming a firefighter. Deficiency points were given by training captains at the academy for observed mistakes or omissions in exercises. Academy graduation signified whether or not a candidate was graduated or terminated. [5] Criterion-related validation is established when there is a significant positive correlation between comparative success on a test and comparative success on some measure of job performance. B. Schlei & P. Grossman, Employment Discrimination Law, 114 (2d ed. 1983). The degree of this relationship is expressed as a “correlation coefficient,” which ranges from -1.0 (completely negative relationship: the better one does on the test, the worse one performs on the job) to +1.0 (total identity of relative test scores and relative job performance). [6] In September 1982, the direct cost of training a recruit at the Academy was $9,515.35. [7] Method of comparison affects whether adverse impact appears "substantial.” Most courts compare selection rates; some compare failure rates. As noted above, this circuit has used both. [8] Significantly, the Uniform Guidelines take the opposite view. They provide that if the total selection process for a job does not have adverse impact, individual components having impact need not “in usual circumstances” be validated. 29 C.F.R. § 1607.4(C). [9] "No evidence was adduced sufficient to establish the additional adverse impact of the written test as a weighted rather than pass/fail selection device.” [10] See National Education Assoc, v. South Carolina, 434 U.S. 1026, 98 S.Ct. 756, 54 L.Ed.2d 775 (1978), aff'g mem. United States v. South Carolina, 445 F.Supp. 1094, 1113 n. 20 (D.S.C.1977); Washington v. Davis, 426 U.S. 229, 262-63, 96 S.Ct. 2040, 2058-59, 48 L.Ed.2d 597 (1976) (Brennan, J. and Marshall, J., dissenting). [11] The court found also that the proposed alternative was not legally available to the County in 1979 because it violated County ordinances and federal funding requirements. Finding No. 19(b)(3). We need not review this finding because of our holding that Finding No. 19(c) is not clearly erroneous. Specifically, we do not decide whether a less discriminatory alternative necessarily fails to meet an employer’s hiring needs if it violates the employer's own regulations. As shown below, under these facts the court’s finding that appellants have not satisfied their burden of establishing that another selection device without a similar discriminatory effect would serve the County’s legitimate interest in efficient and trustworthy workmanship is not clearly erroneous. See Dothard, 433 U.S. at 329, 97 S.Ct. at 2726. [12] The County asserts as a defense that the banding rule was not legally available in 1979. We decline to adopt this "legal unavailability" defense because it is unnecessary to our holding. [13] Appellants include other exhibits in their Excerpt of Record at pages 178-255 which were not admitted into evidence. They do not argue reversible error for the exclusion of these exhibits. This court has not considered them.