To assess satisfactory job performance of superintendents on the basis of school districts' high-stakes testing outcomes, existing teacher models were reviewed and critiqued as potential options for retrofit. For these models, specific problems were identified relative to the choice of referent groups. An alternate referent group (statewide population), that addressed these shortfalls, was proposed and tested via regression procedure. The results indicate that 72% of the variance in student outcomes is beyond their control because of certain student and organizational characteristics. We demonstrated how this information can be used to identify like-type comparisons yielding a better leveled playing field.