- Nov 3, 2017
In many research situations, researchers want to show the extent to which a predictor variable, such as education, influences an outcome variable, such as income, through what is called a mediator, in this case, perhaps job type. Statistical tests for mediation in research typically use a regression-based framework. But there are many research situations in which the assumptions of regression tend not to hold, including when one or more variables may be categorical (e.g., high school/community college/university), or relationships among constructs may be nonlinear. To assist researchers, He, Wang, and Curry detail how and when their new test can best complement existing methods, and they offer software that performs the new test for any combination of continuous and categorical variables.
He, Jiaxiu, Xin (Shane) Wang, and David J. Curry (2017), “Mediation Analysis: A New Test When All or Some Variables are Categorical,” International Journal of Research in Marketing, in press.