We use cookies to distinguish you from other users and to provide you with a better experience on our websites. Close this message to accept cookies or find out how to manage your cookie settings.
Dustin Tingley (Harvard University), Teppei Yamamoto (Massachusetts Institute of Technology, Kentaro Hirose (Princeton University), Luke Keele (Pennsylvania State University), and Kosuke Imai (Princeton University)
Citation
Many social scientists are increasingly interested in identifying the role of particular causal mechanisms. Recent work on causal mediation analysis has developed a new set of procedures for identifying causal mechanisms using minimal assumptions within the potential outcomes framework \citep{mediate1,mediate2,mediate3,mediate4}. The ``mediation'' R package, described in ``mediation: R Package for Causal Mediation Analysis,'' ({\em Journal of Statistical Software} 2014: Dustin Tingley, Harvard; Teppei Yamamoto, Massachusetts Institute of Technology; Kentaro Hirose, Princeton; Luke Keele, Pennsylvania State University; and Kosuke Imai, Princeton), provides an extensive set of tools for performing this analysis. This software allows users to investigate different causal mechanisms while employing different types of data and statistical models, to explore the effect of relaxing identification assumptions, and to examine model-based inference as well as design-based inference, embracing both observational as well as experimental research. A wide range of mediator models may be used, including generalized linear models, ordered categorical models, generalized additive models, quantile regression models, and survival models. To probe the robustness of results to violations of the causal assumptions, the package also implements sensitivity analysis. The R package is available via CRAN and is widely used.