Statistical Mediation Analysis
June 2-4, 2008
David P. MacKinnon, Ph.D.
The Methodology Center announces its 13th Summer Institute on Longitudinal Methods. The Institute, sponsored by The Pennsylvania State University and the National Institute on Drug Abuse, provides researchers who have varying levels of methodological training the opportunity to familiarize themselves with longitudinal data analysis. In addition, the Institute is designed to facilitate the exchange of ideas among substantive and methodological researchers. This year's three-day Institute features a workshop on mediation analysis by expert David MacKinnon of Arizona State University.
The goal of the workshop is to describe statistical, methodological, and conceptual aspects of mediation analysis. The workshop consists of four parts. Part I covers definitions, history, and applications for the mediation model. The purpose of this section is to provide an overview of the research questions the mediation model can answer. Examples from mediation analysis of prevention and treatment programs are described. In Part II, the conceptual model described in Part I is quantified in the estimation of mediation in single and multiple mediator models. Estimation of mediation effects including assumptions of the methods, different statistical tests, effect size, and construction of confidence limits for the mediated effect are covered. The methods described in this section serve as the foundation for more advanced methods. In Part III, advanced mediation models are covered, including mediation in path analysis, longitudinal mediation models, mediation with multilevel data, mediation for categorical variables, and mediation in the context of moderation. In Part IV, general issues in the investigation of mediation including causal inference for mediation models, additional approaches to identifying mediating variables and future directions are described.
David MacKinnon is a Professor in the Department of Psychology at Arizona State University. He received a BA from Harvard University in 1979 and a Ph.D. in measurement and psychometrics from UCLA in 1986. He was an Assistant Professor of Research at the University of Southern California's Institute for Prevention Research from 1986 to 1990. He has been at Arizona State University since 1990 and is affiliated with the Prevention Intervention Research Center and the Research in Prevention Laboratory. Dr. MacKinnon teaches graduate analysis of variance, mediation analysis, and statistical methods in prevention research as well as undergraduate psychological statistics classes. He has served on federal review committees including a 5-year term on the Epidemiology and Prevention review committee. He is on the editorial board of the journals Prevention Science and Psychological Methods. He is also a member of the Prevention Research Methodology Group. Dr. MacKinnon has been principal investigator on several National Institute on Alcoholism and Alcohol Abuse and National Institute on Drug Abuse grants. His primary interest is in the area of statistical methods to assess how prevention and treatment programs achieve their effects.