The field of prevention relies heavily on understanding causal processes as a way of identifying potential targets for prevention and how interventions operate to achieve their effects. Statistical mediation analysis is a critical tool fr prevention research because it helps explain how an independent variable exerts its effect on a dependent variable. Furthermore, the use of multiple methods and/or multiple raters to assess the constructs of interest in prevention science is greatly valued, because multimethod studies are more informative than single method designs and allow for the assessment of convergent validity and method specificity. Despite the fact that many recent studies have used multi-method measurement designs to study mediated effects, many of the approaches used to integrate multiple methods in the statistical analyses have significant theoretical and empirical limitations. The current research aims to address this issue by integrating modern methods of statistical mediation analysis with modern approaches of multitrait-multimethod (MTMM) methodology. In particular, we propose to 1) examine the relative statistical performance of approaches currently used by prevention scientists (Aim 1) and 2) develop and evaluate new multimethod mediation models with latent variables that properly account for the types of methods used in the study (Aim 2). In line with Eid et al. (2008), we distinguish between interchangeable and structurally different methods in this regard and propose to develop models for each type of method as well as the combination of both. Simulation studies will be used to evaluate the performance of the new models in absolute terms as well as in relation to other, already established approaches. Based on our findings from the simulation studies in Aim 1 and Aim 2, we will apply the best performing MM mediation models to real prevention datasets (Aim 3). Finally, the ultimate goal of this research is to disseminate knowledge to applied researchers about how to most appropriately analyze mediated effects in the context of a multimethod measurement design (Aim 4). The successful fulfillment of the aims proposed in this project will impact public health because it will help to clarify the meaning of mediating effects in prevention studies, which is a critical element in designing effective preventive interventions.
Because the analysis of mediating mechanisms is critical to the development of preventive interventions, it is essential that methods to test mediation produce findings that are minimally biased. The results from the proposed study, along with the dissemination activities, will provide methods to improve the estimation of mediation results, thereby improving clinicians'capacity to design effective preventive interventions.
|Geiser, Christian; Keller, Brian T; Lockhart, Ginger et al. (2015) Distinguishing state variability from trait change in longitudinal data: the role of measurement (non)invariance in latent state-trait analyses. Behav Res Methods 47:172-203|
|Geiser, Christian; Koch, Tobias; Eid, Michael (2014) Data-Generating Mechanisms Versus Constructively-Defined Latent Variables in Multitrait-Multimethod Analysis: A Comment on Castro-Schilo, Widaman, and Grimm (2013). Struct Equ Modeling 21:509-523|