Study On PLS Algorithm Of Structural Equation Models

Structural equation models have become an important statistics tool. Many academic disciplines have utilized structural equation models, such as economics, management, psychology and so on.Based on basic principle of structural equation modeling, a corresponding Linear Structural Relations (LISREL) estimation algorithm was presented in this paper.At the same time, a further research on partial least squares (PLS) estimation algorithm was made. Aimed at the structural equation models with two latent variables and two blocks of indicators, the iterative process of PLS algorithm was analyzed and the algorithm geometric meaning has been demonstrated in this paper. Moreover, based on the arrow scheme, the structural equation models with more latent variables were analyzed and some special structural equation models have been demonstrated.According to the research result, at last the paper gives out some advices about how to improve the study on structural equation models.