April 27: Melanie Wall, University of Minnesota

UNIVERSITY OF MINNESOTA

SEMINAR

School of Statistics
College of Liberal Arts

An Extension to Structural Equation Modeling

Melanie Wall
Division of Biostatistics
University of Minnesota

Thursday, April 27, 2000
4:00-5:00 PM, Room 240 Amundson Hall
Social at 3:30 PM in Room 300 Ford Hall

Abstract

In many applied science problems, the variables of interest represent underlying concepts or theoretical quantities, and are not directly observable. The standard statistical method used for such data analysis has been the structural equation modeling procedure. A discussion of the basic uses for structural equation modeling including factor analysis is given with examples coming from current projects in the School of Public Health.

Although the traditional structural equation modeling has been restricted to examining linear relationships between latent variables, a natural extension is to consider polynomial relationships. A systematic and complete estimation and inference procedure for polynomial structural equation modeling is presented. The new procedure applies a method of moments technique similar to the one used in the errors-in-variables regression.