Structural equation modeling (SEM), a scientific methodology with origins in biology, has seen much of its development take place in the human sciences. Increasingly ecologists and evolutionary biologists are finding utility in applying SEM to questions about natural systems. In this paper, we discuss the importance of bridging the gap between SE models, which are ultimately quite specific in their details, and general ecological models.
Results/Conclusions
To better bridge the gap between SE models and general ecological models, we propose to formalize the use of structural equation meta-models (SEMMs). SEMMs are a type of conceptual model that represents the general theoretical ideas of interest in a form that can be readily related to specific SEMs, but that leaves out statistical details and other study features that lack generality. An example is presented to illustrate how a SEMM of plant diversity can represent the general features of SEMs that differ in their specifics. We show that theoretical concepts may be modeled in a variety of ways in SE models and that latent variables have a more narrow application than often implied. As part of this discussion, we consider the use of composite variables, which can function as alternatives to latent variables in cases where the concepts being represented are heterogeneous. We end by showing how SEMMs relate to broader ideas such as general and constituent theories of ecology.
