Tests of assumed mediation models are common in the organizational sciences. However, the validity of inferences about mediation is a function of experimental design and the setting of a study. Regrettably, most tests of mediation have relied on the application of so-called ``causal modeling'' techniques to data from nonexperimental studies. As we demonstrate, inferences about the validity of assumed mediation models are highly suspect when they are based on the findings of nonexperimental research. One of the many reasons for this is the failure of the model being tested to be consistent with reality. Valid research-based inferences about mediation are possible. However, inferences from such tests are most likely to be valid when they are based on research that uses randomized experimental designs. Strategies for conducting research using these and other designs are described. Finally, we offer a set of conclusions and recommendations that stem from our analysis.

Heteroscedasticity refers to a phenomenon where data violate a statistical assumption. This assumption is known as homoscedasticity. When the homoscedasticity assumption is violated, this can lead to increased Type I error rates or decreased statistical power. Because this can adversely affect substantive conclusions, the failure to detect and manage heteroscedasticity could have serious implications for theory, research, and practice. In addition, heteroscedasticity is not uncommon in the behavioral and social sciences. Thus, in the current article, we synthesize extant literature in applied psychology, econometrics, quantitative psychology, and statistics, and we offer recommendations for researchers and practitioners regarding available procedures for detecting heteroscedasticity and mitigating its effects. In addition to discussing the strengths and weaknesses of various procedures and comparing them in terms of existing simulation results, we describe a 3-step data-analytic process for detecting and managing heteroscedasticity: (a) fitting a model based on theory and saving residuals, (b) the analysis of residuals, and (c) statistical inferences (e.g., hypothesis tests and confidence intervals) involving parameter estimates. We also demonstrate this data-analytic process using an illustrative example. Overall, detecting violations of the homoscedasticity assumption and mitigating its biasing effects can strengthen the validity of inferences from behavioral and social science data.

Experimental tests of assumed mediation models have been recommended by numerous methodologists. However, they may prove problematic for a number of reasons. Thus, this article considers (a) experimental designs that can be used in tests of mediation, (b) problems that may arise in manipulating mediator variables, (c) distributional issues in experimental tests of mediation, (d) statistical approaches for testing mediation models, including strategies for testing for direct versus indirect effects in experimental tests of mediation models, and (e) a set of conclusions stemming from our analysis.