Structural equation modeling (SEM) is becoming an increasingly popular data analytic technique in communication studies. Reports of SEM analyses are published in communication journals (including Communication Research Reports) allowing for… Click to show full abstract
Structural equation modeling (SEM) is becoming an increasingly popular data analytic technique in communication studies. Reports of SEM analyses are published in communication journals (including Communication Research Reports) allowing for hypothesis testing with latent variables, estimation of direct and indirect causal effects, and validity testing for measurement instruments. Too often, though, serious mistakes are made by authors of SEM studies that cancel out the potential benefits of SEM. Highlighted in this work are five of the most common mistakes made by communication researchers in analyzing and reporting about structural equation models. These problems concern descriptions of model specification, model identification, and methods to evaluate the degree of model-data correspondence, or fit, and lack of replication that continue to plague the empirical SEM literature. This current work is intended as a primer, one that outlines best practices in contrast to widespread but poor practices in each of the areas just mentioned. The hope is that communication researchers yield benefits from the correct application of SEM while avoiding common pitfalls.
               
Click one of the above tabs to view related content.