Applications of Structural Equation Modeling in Marketing and Consumer Research: Did Researchers Heed Baumgartner and Homburg's (1996) Advice?
Swimberghe, Krist, Issues in Innovation
Structural Equation Modeling (SEM) has been established as one of the choice data analysis methods among researchers in marketing and other disciplines. The popularity of SEM among academicians is exemplified by the publication of Structural Equation Modeling, a journal dedicated to the questions which confront researchers when using SEM. Yet, it was not until the monograph by Bagozzi on causal modeling in 1980 that SEM caught the attention of a wide audience of researchers.
Note: A list of articles included in the study is available upon request.
In a study which was published in 1996, Baumgartner and Homburg documented the increased popularity of SEM in marketing and consumer research. The authors cautioned however that with the increased use of SEM, comes a responsibility to use this method more appropriately. This concern is not merely an issue in marketing and consumer research, but has also been voiced by academicians in operations management (Shah and Goldstein, 2006), organizational research (Medsker et al., 1994), psychology (Hershberger, 2003), MIS (Chin and Todd, 1995; Gefen et al., 2000), logistics (Garver and Mentzer, 1999), and strategic management (Shook et al., 2004).
Baumgartner and Homburg (1996), as well as some of the other authors previously mentioned, provide guidelines for future applications of SEM. While taking into consideration suggestions made by researchers in other disciplines, this paper focuses on the guidance provided by Baumgartner and Homburg in their 1996 study, published in the International Journal of Marketing Research, and suggestions provided in the most recent study by Shah and Goldstein, published in the journal of Operations Management in 2006. The purpose of this paper is to check if authors who have used SEM in marketing research have heeded the advice by Baumgartner and Homburg (1996).
Specifically, my review has the following objectives. First, I document the number of marketing and consumer research applications that used SEM from 2000-2005 and classify them according to the purpose for which SEM is used. Second, I seek to evaluate to what extent the suggestions, provided by Baumgartner and Homburg in their 1996 published study, were taken into account by researchers and have been addressed in future marketing studies.
STRUCTURAL EQUATION MODELING (SEM)
Structural equation modeling is a statistical methodology that takes a confirmatory (i.e. hypothesis testing) approach to the analysis of a structural theory bearing on some phenomenon. Several aspects of SEM set it apart from the older generation of multivariate procedures (Fornell, 1982). First, it takes a confirmatory approach, rather than an exploratory approach to data analysis. Sound and strong theory guides the pattern of intervariable relations a priori. In contrast, traditional multivariate procedures are essentially descriptive. Second, SEM explicitly takes into account measurement error which traditional statistical analysis methods rooted in regression ignore. Error in the explanatory variables is especially relevant when dealing with survey data. Finally, traditional analysis methods are based on observed measurements. SEM can take into account both observed (manifest) and unobserved (latent) variables (Byrne, 2001). For these reasons SEM has become a very attractive and popular method of analyzing survey data.
SUGGESTED GUIDELINES FROM PREVIOUS RESEARCH
Baumgartner and Homburg in their 1996 review of SEM applications in marketing and consumer research provided three general guidelines to future users of these techniques. Shah and Goldstein (2006) made several similar suggestions. Combined, they provide the following guidance: First, before empirical data are ever collected thought should be given to model specification grounded in sound theoretical foundations. Structural equation models are most effective when used to confirm a well-defined set of structural relationships in a parsimonious framework. …