Although it is possible for employer discrimination policies to be facially apparent in employment discrimination cases, usually employer policies are not overtly discriminatory. In those situations where it is not apparent, the only method of proving a prima facie case of discrimination in most systemic disparate treatment instances is to show a pattern of discrimination in the employer's hiring decisions. Since these patterns can usually "be identified and analyzed in quantitative form," cases rely primarily, and often exclusively, on statistics. If the plaintiff offers statistics that indicate a hiring practice that is unlikely absent discrimination, then the plaintiff is deemed to have created a prima facie case, thereby shifting to the defendant the burden of rebutting the plaintiff's statistical proof. The defendant can rebut this presumption by: (I) demonstrating that the plaintiff's statistics are inaccurate or insignificant; (2) offering his own statistical proof of nondiscriminatory hiring practices; or (3) offering nondiscriminatory explanations for the disparity in the statistics.
While courts sometimes view the statistical data initially presented as so one-sided that little further analysis is required, courts are more inclined to use complicated statistical methods to determine whether the statistical discrepancy in the under-hiring of a protected class is due to pure chance, or rather discriminatory reasons. A majority of courts use standard deviations of numerical distribution approximations based on binomial distributions as described in Hazelwood School District v. United States. This Note aims to prove that the assumption of independent trials in the usage of binomial distributions is unjustified in employment discrimination cases, and is therefore an inappropriate and inaccurate model to use as a primary basis for systemic disparate treatment cases.
Why Binomial Distributions Do Not Work as Proof of Employment Discrimination,
59 Hastings L.J. 1235
Available at: https://repository.uchastings.edu/hastings_law_journal/vol59/iss5/10