ERIC Number: ED420154
Record Type: RIE
Publication Date: 1998-Jul-16
Reference Count: N/A
The Ten Commandments of Good Structural Equation Modeling Behavior: A User-Friendly, Introductory Primer on SEM.
This paper provides an introduction to basic issues concerning structural equation modeling (SEM), a research methodology increasingly being used in social science research. First, seven key issues that must be considered in any SEM analysis are explained. These include matrix of associations to analyze, model identification, parameter estimation theory, multivariate normality, model misspecification and specification searches, sample size, and measurement model adequacy. Second, heuristic SEM analyses involving structural models are presented to demonstrate how SEM takes score measurement reliability into account and how SEM may shed light on causal issues. Finally, ten commandments for proper SEM use are presented, among which are the following: (1) never conclude that a model has been definitely proven; (2) for specification searches that require larger samples, test the re-specified model with a "hold-out" or independent sample and never change a specification without a theoretical justification; (3) test multiple plausible rival models; and (4) don't use SEM with small samples. Appendices provide five examples of statistical analyses using SEM. (Contains 66 references.) (DB)
Publication Type: Reports - Descriptive; Speeches/Meeting Papers
Education Level: N/A
Authoring Institution: N/A
Note: Paper presented at the Annual Meeting of the U.S. Department of Education, Office of Special Education Programs (OSEP) Project Directors' Conference (July 16, 1998, Washington, DC).