Structural Equation Models for the Study of Ecosystems and Socio-Ecosystems

International audienceStructural equation models (SEMs) provide a statistical framework for compact representations of theories concerning the relations between components in a socio-ecosystem. In this chapter, the authors show how SEMs maintain statistical power in cases with large numbers of observed variables, combining variables into groups and linking them to subjacent components through a measurement model; contribute to improved understanding of socio-ecosystems via progressive adjustments to the SEM structure throughout the analytical process. They highlight testing techniques based on resampling in order to avoid asymptotic hypotheses. This focus has inevitable implications in terms of the discussion of SEMs and their simplicity. Estimating an SEM is a numerical optimization operation, and can take several minutes using a standard computer