Research design options for testing mediation models and their implications for facets of validity

Purpose - Tests of assumed mediation models are common in research in many disciplines, including managerial psychology, industrial and organizational psychology, organizational behavior, and organizational theory. Thus, the purpose of this paper is to detail experimental design options for conducting such tests in a manner that has the potential to yield results that have high levels of internal and construct validity.Design methodology approach - The paper presents a logical analysis of strategies for testing mediation models so as to insure valid inferences about causal relations between variables.Findings - The most appropriate strategy for testing assumed mediation models is research that uses randomized experimental designs.Practical implications - Managers should base their actions on valid evidence about phenomena. More specifically, managerial actions should be predicated on research results that have high levels of internal, construct, and statistical conclusion validity. Thus, this paper encourages managers to base decisions about organizational policies and practices on well-designed experimental research.Originality value - This paper addresses a number of points about issues involving internal and construct validity in tests of assumed causal models that have not been covered in previous work.