There are lies and damn lies, and then there are statistics

published in Oklahoma Employment Law Letter | October 1, 2012

How do employees go about proving discrimination in cases involving multiple employment decisions, like reductions in force (RIFs)? Typically, they use a combination of evidence and arguments — for example, assertions that selection criteria were subjective (e.g., “attitude” and “adaptability”), evidence of managers’ biased remarks (e.g., we “need to get rid of the old farts”), and statistical evidence (e.g., a larger number of workers in the protected group were terminated than would be expected if truly neutral selection criteria were used). The 10th Circuit recently analyzed the value of these kinds of evidence in a large-scale age discrimination lawsuit, focusing on the employees’ use of statistical evidence to prove discriminatory intent. The court held that although there was a statistically significant disparity in the treatment of older and younger workers, in practical terms, the disparity was too small to prove a pattern of age discrimination or a significant disparate impact on older workers.


In June 2005, Boeing terminated the entire 10,671-member workforce at its Wichita, Kansas, division and sold the division to Spirit AeroSystems. The next day, Spirit rehired 8,354 employees selected for rehire by Boeing managers. Although the workforce was relatively old both before and after the sale (with an average age of 48), Spirit rehired a lower percentage of older workers, and the average age of the workforce after the sale decreased by about five months.

Former Boeing employees filed a class action on behalf of 700 employees who weren’t rehired by Spirit. They filed claims for both pattern and practice (intentional discrimination) and disparate impact (adverse impact on older employees under a neutral employment policy) age discrimination. The employees argued that selection criteria were subjective, there was evidence of managers’ remarks showing age bias, the companies had a financial motive to reduce the age of the workforce to reduce pension costs, and there was significant statistical evidence of age bias. The district court dismissed the case in favor of Boeing and Spirit, and the employees appealed to the 10th Circuit.

10th Circuit’s opinion

The 10th Circuit held that although the employees did indeed introduce evidence of age bias, they still failed to prove that the companies engaged in a widespread pattern or practice of age discrimination or that their hiring practices had a statistically significant disparate impact on older workers. In other words, the court held that although there was evidence of isolated age discrimination in certain employment decisions, that wasn’t enough to win under the employees’ theory of the case; instead, they had to come up with evidence of systemwide discrimination, which they failed to do.

The former employees presented expert statistical evidence showing that the difference between the number of employees over age 40 who should have been recommended for rehire in the absence of discrimination (8,028) and the number who actually were recommended for rehire (7,968) was greater than five standard deviations. In addition, the difference between the number who should have been rehired in the absence of discrimination (7,285) and the number who actually were rehired (7,237) was over four standard deviations. In the opinion of the employees’ expert witness, there was only a 1 in 50,000 chance that the hiring decisions were the result of an unbiased process.

However, the 10th Circuit found that the disparities identified by the former employees lacked practical significance. The court explained that in light of the thousands of total recommendations and hiring decisions being analyzed, the number of disparities in terms of real numbers was small. The court noted that if Boeing had recommended just 60 more employees over age 40 or if Spirit had hired just 48 more people over 40, the discrepancies would have disappeared altogether. And when you consider those numbers in the context of 8,000 recommendation decisions and 8,000 hiring decisions, they are insignificant as a practical matter — even though the statistical analysis painted a different picture.

The court stated that the companies recommended and hired more than 99 percent of the older employees they would have been expected to recommend and hire in the absence of discrimination. Accordingly, the 10th Circuit concluded that no reasonable fact finder could have found that discrimination was the companies’ “standard operating procedure — the regular rather than the unusual practice.”

The 10th Circuit affirmed the dismissal of the employees’ disparate impact claim for the same reasons it affirmed the dismissal of the pattern or practice claim, noting that although their statistics revealed a “highly unlikely disparity in the treatment of older and younger workers,” the disparity was “very small” in absolute terms. As a result, the employees also failed to show that the companies’ sale and rehiring decisions caused a significant disparate impact on older workers.

The former employees introduced evidence of statements by company managers demonstrating concerns about an aging workforce and the goal of reducing pension costs. The 10th Circuit used the same type of analysis for this evidence, concluding that although the comments supported the idea that discrimination “may have occurred,” the statements were only evidence of “isolated or sporadic discriminatory acts by the employer, which are insufficient to establish a prima facie [minimally sufficient] case of a pattern or practice of discrimination.” Apsley v. Boeing Co., No. 11-3238, 2012 WL 3642800 (10th Cir., Aug. 27, 2012).

Importance of the decision

The 10th Circuit rejected evidence of superficial statistical disparities as proof of age discrimination, relying instead on the practical implications of the statistical evidence. The court specifically held that even though a statistical analysis may support a claim of a biased process, courts have to put those numbers into a real-world context. This decision is important because too many times in the past, courts have been swayed by complex statistical evidence that, frankly, is poorly understood by lawyers and judges alike and therefore has been manipulated by paid experts to support meritless claims.