Multiple Indicators and Multiple Causes (MIMIC) Models as a Mixed-Modeling Technique: A Tutorial and an Annotated Example

摘要

Formative modeling of latent constructs has produced great interest and discussion among scholars in recent years. However, confusion exists surrounding researchers’ ability to validate these models, especially with covariance-based structural equation modeling (CB-SEM) techniques. With this paper, we help to clarify these issues and explain how formatively modeled constructs can be assessed rigorously by researchers using CB-SEM capabilities. In particular, we explain and provide an applied example of a mixed-modeling technique termed multiple indicators and multiple causes (MIMIC) models. Using this approach, researchers can assess formatively modeled constructs as the final, distal dependent variable in CB-SEM structural models—something previously impossible because of CB-SEM’s mathematical identification rules. Moreover, we assert that researchers can use MIMIC models to assess the content validity of a set of formative indicators quantitatively—something considered conventionally only from a qualitative standpoint. The research example we use in this manuscript involving protection-motivated behaviors (PMBs) details the entire process of MIMIC modeling and provides a set of detailed guidelines for researchers to follow when developing new constructs modeled as MIMIC structures.