The motivation for this comparison of approaches to sample size estimation for paired binary data came from a study examining how patient gender affected physicians’ treatment recommendations regarding total knee arthro-plasty (TKA) [16]. Specifically, two standardized patients (one man and one woman) differing only in gender with otherwise identical case histories underwent blinded assess-ments by family physicians and orthopedic surgeons located in Ontario, Canada. Both standardized patients pre-sented with chronic knee pain as their chief complaint and level of function, pain, and prior treatment appropriate for a patient with moderate knee osteoarthritis. At the end of the visit, standardized patients recorded the binary primary outcome: the physicians’ recommendations for TKA (yes5 1 and no5 0) on a postvisit checklist.