Dr. William Fairley, senior statistician for Analysis & Inference, discovered two key problems with the auditor’s estimate. First, the auditor’s sample of charges made to Medicare was not random: alleged overcharges were much greater than if the auditors had used a truly representative sample. Second, instead of simply totaling the overcharges, the auditors adopted a complicated statistical formula involving averages of charges for different periods of service.
Dr. Fairley worked closely with the client to demonstrate that the auditor’s formula was biased to find overcharges where none existed, and introduced an alternative “regression” method to accurately estimate any overcharges.
A+I proved that, aside from a few statistically insignificant exceptions, estimated overcharges were zero, and noted that other federal agencies tended to disregard statistically insignificant differences. Dr. Fairley’s analysis was instrumental in the negotiations.