Content area
Full Text
The research wheel has been used to describe the recursive process of exploratory, descriptive, and causal research by which we create knowledge about observed phenomena, related factors, and the populations involved. This process yields endless iterations of deductive and inductive reasoning as researchers identify phenomena, hypothesize relationships, and test theoretical assumptions. With each passing turn, the empirical methodology becomes much more rigorous, yielding better information for decision making.
As business research evolves across all disciplines, so do the quantitative techniques used for empirical analysis. In particular, statistical tools related to structural equation modeling have increased rapidly in recent years (Hair, Hollingsworth, et al., 2017). Structural equation modeling (SEM) is a second-generation statistical technique used to test hypothetical relationships in social science and research. This powerful tool simultaneously performs analyses similar to principal components analysis and path analysis. SEM captures direct and indirect effects, thereby explaining a larger portion of variance in the dependent variables and offering more statistical power than multiple regression (Hair, Matthews, et al., 2017). Traditionally, SEM programs were limited to covariance-based applications (CB-SEM), such as AMOS or LISREL. While powerful, CB-SEM is primarily useful for confirming theoretical models. But there are a host of other variance-based programs that can be used for both exploratory and confirmatory quantitative analysis.
Recent years have seen the emergence of a new way to approach structural equation modeling, Partial Least Squares Structural Equation Modeling (PLS-SEM) is a variance-based SEM approach useful for many different types of business research. For instance, technological advances in PLS software now provide the ability to empirically test hierarchical component models, analyze moderating effects, and examine non-linear functions for interactions between latent variables (Hair, Sarstedt, Ringle, et al., 2012). PLS-SEM also provides the flexibility, robustness, and precision to perform different types of analyses across all disciplines of the social sciences, often using a similar conceptual model.
While most researchers are familiar with CB-SEM, many are unaware of the potential advantages of PLS-SEM. The purpose of this article is to describe potential applications of variance-based structural equation modeling across many disciplines of business research. This article discusses and presents an overview of the many uses of PLS-SEM throughout the research cycle, be it exploratory, descriptive, or causal. We first provide a brief overview of...