Employment Litigation · Primary-source case analysis

Watson: Adding an Interview Did Not Shield a Selection System From Impact Review

By Aneeta Mishra and Timothy Chan

Published August 11, 2026

Decision: Watson v. Fort Worth Bank & Trust, 487 U.S. 977 (1988).

Watson v. Fort Worth Bank & Trust, 487 U.S. 977 (1988), rejected a proposed loophole: A selection system did not become immune from disparate-impact review merely because it included subjective judgment. The Court illustrated the problem with a written test followed by a brief interview. It did not hold that interviews or supervisory discretion are inherently discriminatory.

Objective and subjective methods can work together

If adding an interview to a test placed the whole process outside impact review, an employer could keep giving the test nearly decisive weight while avoiding the doctrine applied to objective screens. That was the majority’s reason for rejecting a categorical objective-versus-subjective boundary. The inquiry could reach mixed selection systems as well as wholly discretionary ones. Watson, 487 U.S. at 989–990.

Delegating judgment was not itself proof of bias

The majority recognized that a small business may reasonably leave promotion decisions to supervisors who know the jobs and candidates. Delegation alone did not imply discriminatory conduct. But that observation did not guarantee that each decision was unbiased or eliminate the possibility of effects produced by subconscious stereotypes. Watson, 487 U.S. at 990–991.

Impact and intentional discrimination remained different questions

The lower courts had evaluated Clara Watson’s individual promotion claims through intentional-discrimination proof and rejected pretext. The Supreme Court held that subjective practices could also be examined through disparate impact in appropriate cases. Failure to prove an intentional motive therefore did not answer whether the practice had the prohibited effect alleged. The Court did not simply reverse the finding on intent. Watson, 487 U.S. at 983–989, 999.

The remand did not establish a reliable statistical case

Neither lower court had assessed the statistics under impact analysis. The Court even recognized that the database might be too small for meaningful statistical analysis and left the next step to the Court of Appeals. The vacatur allowed the correct theory to be considered; it did not establish causation or liability. Our companion analysis explains the separate opinions’ disagreement about proof. Watson, 487 U.S. at 999–1000.

Key takeaways

• Identify how tests, interviews, and supervisor recommendations actually combine.
• Do not assume an interview immunizes an otherwise test-driven system.
• Keep intentional-motive evidence separate from evidence of selection effects.
• Do not treat permission to analyze impact as a finding that the statistics prove it.