Measurement of Fidelity of Implementation of Evidence- Based Practices: Case Example of the IPS Fidelity Scale Gary R. Bond, Department of Psychiatry, Dartmouth Medical School Deborah R. Becker, Departments of Community and Family Medicine and Psychiatry, Dartmouth Medical School Robert E. Drake, Departments of Psychiatry and Community and Family Medicine, Dartmouth Medical School Advances in the scientific study and dissemination of evidence-based practices (EBPs) require psychometri- cally valid fidelity scales, which measure adherence to program models. We examined the literature on the Individual Placement and Support Fidelity Scale to illus- trate the strengths and limitations of this methodology. We found that this scale had excellent psychometric properties. Nine of 10 studies assessing its predictive validity found positive associations with employment outcomes. Its use in quality improvement was sup- ported by positive reports from seven multisite pro- jects. Although not yet evaluated as an accreditation tool, three states have adopted the scale for reimburse- ment purposes. Development of valid fidelity scales is necessary for the field to advance. Technical and logisti- cal challenges impede the replication of these findings for other EBPs. Key words: evidence-based practice, fidelity scales, individual placement and support, predictive validity, supported employment. [Clin Psychol Sci Prac 18: 126– 141, 2011] Increasing the use of evidence-based interventions is the central theme of current mental health reforms (Institute of Medicine, 2001; New Freedom Commis- sion on Mental Health, 2003; U.S. Surgeon General, 2000). Fidelity, defined as adherence to evidence-based program models (Bond, Evans, Salyers, Williams, & Kim, 2000), has also emerged as a central concept in these efforts. Within the community mental health field, the need for systematic methods to assess fidelity became clear in the 1970s and 1980s (Drake, Bond, & Essock, 2009). During this period, the National Institute of Mental Health funded the dissemination and evaluation of promising models to promote community integration of individuals with severe mental illness (Turner & TenHoor, 1978). Many of these projects produced dis- appointing results, in part attributable to poor model specification (Brekke, 1988). The absence of clear, objective program standards—criteria for implementing a practice—interfered with many well-intentioned efforts to disseminate these practices and severely atten- uated the accumulation of scientific evidence in sup- port of these program models. Starting in the early 1990s, many researchers began to define program mod- els operationally and to develop tools to measure their implementation. Nonetheless, the lack of model speci- fication continues to the present; the large majority of randomized controlled trials of behavioral interventions continue to lack critical details about fidelity (Michie, Address correspondence to Gary R. Bond, Dartmouth Psychiatric Research Center, Rivermill Commercial Center, 85 Mechanic Street, Suite B4-1, Lebanon, NH 03766. E-mail: gary.bond@dartmouth.edu All three authors are affiliated with the Dartmouth Psychiatric Research Center. Ó 2011 American Psychological Association. Published by Wiley Periodicals, Inc., on behalf of the American Psychological Association. All rights reserved. For permissions, please email: permissionsuk@wiley.com 126