Applications of Two-Stage Least Squares in Causal Analysis and Structural Equations.
Abstract
A procedure, two-stage least squares (2SLS), for analyzing structural equations when one or more explanatory variables are correlated with the error or disturbance term is reviewed. A brief introduction to structural equations and the use of ordinary least squares in causal analysis is presented initially. This is followed by an introduction of 2SLS and the application to 2SLS to designs in which (a) two or more variables are reciprocal causes of each other, (b) one or more variables contain random measurement error, and (c) lagged values of one or more dependent variables are used as predictors. A major objective for the review of these particular applications of 2SLS is to demonstrate how salient problems in psychology can be addressed by use of structural equations. (Author)
Document Details
- Document Type
- Technical Report
- Publication Date
- Mar 11, 1977
- Accession Number
- ADA038348
Entities
People
- B. Krishna Singh
- Lawrence R. James
Organizations
- Texas Christian University