This study uses two studies about the role of managed-care programs in serving Medicaid long-term care clients in Florida to illustrate how different research designs can reach divergent conclusions. Two reports from different groups using essentially the same database to assess the impact of managed care on a group of older Medicaid clients served by a Nursing Home Diversion Program reached different conclusions. The report from Florida's Office of Program Policy Analysis and Government Accountability concluded that the Diversion program saved money, whereas the report from the Florida Policy Exchange Center on Aging at the University of South Florida reached basically the opposite conclusion. Both agreed that the capitation rate was too high. How the policy questions are framed and analyzed can affect the conclusions reached. A variety of factors can influence the apparent effects of programmatic interventions. Evaluations must take relevant confounding variables into account.