Sie befinden Sich nicht im Netzwerk der Universität Paderborn. Der Zugriff auf elektronische Ressourcen ist gegebenenfalls nur via VPN oder Shibboleth (DFN-AAI) möglich. mehr Informationen...
To make valid statistical inferences from mediation analysis, a number of assumptions need to be assessed. Among the assumptions, 2 frequently discussed ones are (a) the independent variable, mediator, and outcome variables are measured without error; and (b) no confounders of the effects in the mediation model are omitted. The impact of violating either assumption alone on statistical inference of mediation has been discussed in previous literature. In practice, violations of the 2 assumptions often co-occur. In this study, we analytically investigated the effects of measurement error and omitting confounders on statistical inference of mediation effects, including both point estimation and significance testing. Based on the analytical results, we proposed sensitivity analysis techniques for assessing the robustness of mediation inference to the violation of the 2 assumptions. To implement the techniques, we developed R functions and a user-friendly web tool. Simulated-data and real-data examples were provided for illustrations. We hope the developed tools will help researchers conduct sensitivity analyses of mediation inference more conveniently.
Translational Abstract
Mediation analysis is widely used in substantive research to investigate the underlying causal mechanism between the independent and outcome variables. Valid statistical inference of mediation requires some assumptions. Two frequently discussed assumptions are (a) the independent variable, mediator, and outcome variables are measured without error; and (b) no confounders of the effects in the mediation model are omitted. In practice, either assumption can be challenging to meet; furthermore, violations of the two assumptions often co-occur, especially in mediation studies of psychological and behavioral sciences. The extant literature on the consequences of jointly violating the two assumptions in mediation analysis is limited. Especially, how statistical testing of mediation would be influenced has not been well understood, although estimation bias has been studied before. Therefore, to fill in the research gap, we systematically investigated the effects of measurement error and omitting confounders on statistical inference of mediation in this study. Both analytical and graphical tools were used to demonstrate the effects. Furthermore, we proposed a sensitivity analysis approach using the analytical results to help researchers gauge the robustness of mediation analyses under the potential violation of the two assumptions. Easy-to-use tools including R functions and a web application were developed to facilitate the implementation. For illustration, both simulated-data and real-data examples were provided. We hope this study can provide further insights into the impact of violating the two assumptions in statistical mediation analysis, and help researchers conveniently assess the robustness of empirical inference of mediation.