Summary
accepting 1.96 standard deviations as the threshold for statistical significance
Summary of this case from Jones v. City of Bos.Opinion
Nos. 05-56061, 05-56064.
Argued and Submitted May 8, 2007.
Filed May 16, 2007.
Dan Stormer, Esq., Hadsell Stormer, Pasadena, CA, for Plaintiff-Appellant/Plaintiff-Appellee.
Delia Bahan, Esq., Bahan Associates, Pasadena, CA, for Plaintiff-Appellee.
Joel M. Cohn, Esq., Akin Gump Strauss Hauer Feld, LLP, Washington, DC, Celine M. Cooper Fax, Office of the California Attorney General, Los Angeles, CA, for Defendants-Appellees/Defendants-Appellants.
Richard J. Romans M., Esq., Litigation Solutions Law Group LLP, Redwood City, CA, for Defendant.
Appeal from the United States District Court for the Central District of California, Consuelo B. Marshall, Chief District Judge, Presiding. D.C. No. CV-94-00083-CBM.
Before: SILVERMAN, WARDLAW, and BYBEE, Circuit Judges.
MEMORANDUM
This disposition is not appropriate for publication and is not precedent except as provided by 9th Cir. R. 36-3.
Jeff D. Paige, on behalf of himself and similarly situated minority employees in the California Highway Patrol (CHP), appeals the district court's judgment in favor of the State of California in his Title VII action. A jury found in favor of the State on a claim of disparate treatment and the district court rejected a claim of disparate impact after a bench trial. We now affirm.
I. Disparate Treatment
Plaintiffs had the burden of demonstrating a violation of Title VII between August 1, 1992 and January 5, 1994. They raise a host of evidentiary objections to the manner in which the district court admitted or excluded testimony about events occurring outside of that period.
We find no abuse of discretion in any of the district court's rulings. The court has considerable discretion in ruling on evidentiary objections, particularly when it comes to questions of relevancy. See United States v. Alvarez, 358 F.3d 1194, 1205 (9th Cir. 2004). Furthermore, many of plaintiffs' claimed errors were not objected to at trial and have, therefore, been waived.
II. Disparate Impact
To prevail on a claim of disparate impact, "the plaintiffs must offer statistical evidence of a kind and degree sufficient to show that the [employment] practice in question has caused the exclusion of applicants for jobs or promotions because of their membership in a protected group." Watson v. Fort Worth Bank Trust, 487 U.S. 977, 994, 108 S.Ct. 2777, 101 L.Ed.2d 827 (1988).
The district court rejected plaintiffs' statistical evidence of disparate treatment on two independent bases.
The district court's analysis was vaguely reminiscent of the burden-shifting framework outlined in McDonnell Douglas Corp. v. Green, 411 U.S. 792, 93 S.Ct. 1817, 36 L.Ed.2d 668 (1973): plaintiffs failed to prove their prima facie case of disparate treatment, but if they had, the State failed to rebut it. Such terminology is more appropriate for the summary judgment stage. See Bouman v. Block, 940 F.2d 1211, 1223 (9th Cir. 1991) ("Once a Title VII case proceeds to judgment the issue is no longer whether plaintiff has established a prima facie case, but whether there was discrimination."). However, it is apparent that, despite its use of these terms, the district court was addressing the ultimate question of whether plaintiffs had proven their case by a preponderance of the evidence.
A. Failure to clear statistical significance threshold
The court rejected a statistical analysis by plaintiffs' expert showing a disparity of 2.18 standard deviations in all CHP promotional tests from 1988 until May 1993. It also discounted other studies showing a statistically significant disparity of slightly over two standard deviations in other promotional tests.
To the extent that the district court rejected as insignificant any of plaintiffs' statistical results showing a disparity of greater than 1.96 standard deviations, we find error. First, the threshold was not in dispute. Both parties' experts testified that any disparity above 1.96 standard deviations was indicative of statistical significance. Second, the 1.96 threshold conforms with accepted conventions in the social science field and with the federal government's internal standards. See Segar v. Smith, 738 F.2d 1249, 1282-83 (D.C. Cir. 1984). Third, we have relied upon the 1.96 standard deviation standard unless the factual circumstances of the case warranted departure. Compare Bouman v. Block, 940 F.2d 1211, 1225 n. 1 (9th Cir. 1991); Eldredge v. Carpenters b.6 N. Cal. Counties JATC, 833 F.2d 1334, 1340 n. 8 (9th Cir. 1987) with Gay v. Waiters' Dairy Lunchmen's Union, 694 F.2d 531, 551-52 (9th Cir. 1982) (requiring higher threshold for inference of intentional discrimination); Contreras v. City of Los Angeles, 656 F.2d 1267, 1273 (9th Cir. 1981) (rejecting statistical significance at the 1.96 level because sample size was too small). Here, there is nothing in the record to justify the district court's use of a higher threshold. As a result, the court committed clear error by not using the 1.96 mark. B. Credibility of plaintiffs' statistical expert
We are also troubled by the district court's reliance on our opinion in Gay for the proposition that "[c]ourts should be `extremely cautious' of drawing any inferences from standard deviations in the range of 1 to 3." 694 F.2d at 551. The quoted material was taken out of context. In Gay, we discussed the threshold of statistical significance necessary to "support an inference of intentional discrimination," which, we reasoned, must be higher than the traditional 1.96 marker for a disparate impact case. Id. (emphasis added).
The district court also gave "little weight" to plaintiffs' statistical expert because of concerns raised by the defendant's expert about the validity of his data set and methodology. In this the district court did not clearly err. First, the expert offered only a single-pool analysis for some of his studies — a statistical technique criticized by some as outdated and inaccurate. Second, he relied on records that improperly coded the race of the CHP applicants — a decision that purportedly invalidated his conclusions.
The district court was the factfinder in this pitched battle of the experts. Plaintiffs' expert claimed that his data and methods were sound and the State's expert vigorously disagreed. There is no universal truth to this dispute; statistics are amenable to interpretation and manipulation by each party. See Int'l Brotherhood of Teamsters v. United States, 431 U.S. 324, 340, 97 S.Ct. 1843, 52 L.Ed.2d 396 (1977) ("[Statistics] come in infinite variety. . . . [T]heir usefulness depends on all of the surrounding facts and circumstances."). Here, the district court sided with the State in finding that the plaintiffs expert's methodology was flawed, and the record does not compel the alternative.
Plaintiffs identified, and the State conceded, violations of the "four-fifths" rule during several promotional processes. See 29 C.F.R. § 1607.4(d). The rule is a rough proxy for a statistical measurement of disparate impact. See Clady v. County of Los Angeles, 770 F.2d 1421, 1428 (9th Cir. 1985) ("a rule of thumb"). Therefore, although the State's violations of the rule may have relevance to the disparate impact inquiry, the district court's decision to emphasize the statistical evidence over the rule was not clearly erroneous.
There was no clear error in this decision. Because one of the two grounds cited by the district court to discredit the plaintiffs' statistical evidence was valid, remand is not necessary. AFFIRMED.
In light of this ruling, it is unnecessary to address the State's cross-appeal.