Disparate impact case for social workers' licensing exam is rejected on appeal

This race discrimination claim alleges that the licensing exam for certain social workers discriminates on the basis of race because Black and Hispanic test-takers perform significantly worse than white candidates but that the Association of Social Work Boards, which administers the tests, knew about these disparities but did not correct them, resulting the plaintiffs' failing test scores. The case loses even prior to the commencement of discovery.

The case is Alameda v. Association of Social Work Boards, a summary order issued on May 15. This is a disparate impact case, where the plaintiffs can win even if the testing board did not intend to discriminate on the basis of race. For a disparate impact case to succeed, however, the plaintiffs have to satisfy a demanding legal standard that employs statistics and particular formulas. But first we have a mundane legal issue under Title VII: can the plaintiffs even sue the testing company under this civil rights statute?

Under Title VII, you can only sue the employer for discrimination. The testing people were not plaintiffs' employers. Cases hold that you can hold a third-party liable under Title VII where your formal employer "has delegated one of its core duties to a third party." But the Second Circuit (Perez, Nathan and Katzmann [Court of Intern'l Trade]) finds the defendant here is not such a "third party" that would create an employee-employer relationship under Title VII. 

Plaintiffs also sue under Section 1981, which prohibits racial discrimination in making contracts. There is no employee-employer relationship requirement under Section 1981 in this case, but plaintiffs lose on a different ground relating to the merits of the case: they have not plausibly alleged a disparate impact claim.

Plaintiffs rely on statistics to assert their case, pointing to the disparate test results among white and Black and Latino/a test-takers. But under the cases, to assert such a claim based on statistics alone, you have to show the statistical disparity is "of a level that makes other plausible nondiscriminatory explanations very unlikely." In other words, you need a dramatic statistical disparity. The Court holds that the statistical disparities here do not satisfy that demanding test.