thumbnail

Structural equation modeling for observational studies

Journal of Wildlife Management

By:
DOI: 10.2193/2007-307

Links

Abstract

Structural equation modeling (SEM) represents a framework for developing and evaluating complex hypotheses about systems. This method of data analysis differs from conventional univariate and multivariate approaches familiar to most biologists in several ways. First, SEMs are multiequational and capable of representing a wide array of complex hypotheses about how system components interrelate. Second, models are typically developed based on theoretical knowledge and designed to represent competing hypotheses about the processes responsible for data structure. Third, SEM is conceptually based on the analysis of covariance relations. Most commonly, solutions are obtained using maximum-likelihood solution procedures, although a variety of solution procedures are used, including Bayesian estimation. Numerous extensions give SEM a very high degree of flexibility in dealing with nonnormal data, categorical responses, latent variables, hierarchical structure, multigroup comparisons, nonlinearities, and other complicating factors. Structural equation modeling allows researchers to address a variety of questions about systems, such as how different processes work in concert, how the influences of perturbations cascade through systems, and about the relative importance of different influences. I present 2 example applications of SEM, one involving interactions among lynx (Lynx pardinus), mongooses (Herpestes ichneumon), and rabbits (Oryctolagus cuniculus), and the second involving anuran species richness. Many wildlife ecologists may find SEM useful for understanding how populations function within their environments. Along with the capability of the methodology comes a need for care in the proper application of SEM.

Additional Publication Details

Publication type:
Article
Publication Subtype:
Journal Article
Title:
Structural equation modeling for observational studies
Series title:
Journal of Wildlife Management
DOI:
10.2193/2007-307
Volume
72
Issue:
1
Year Published:
2008
Language:
English
Larger Work Type:
Article
Larger Work Subtype:
Journal Article
First page:
14
Last page:
22
Number of Pages:
9