| In the social research fields, we often need to deal with the relationship between multiple variables, and the variable that can not be directly observed, the so-called latent variable. For example, the ability to study management, trust, self-esteem, motivation, success concepts are latent variables. In fact, these variables are the basical concept that people to understand the establishment of the society. In general, they do not exist for the direct measurement methods of operation. The traditional statistical methods of dealing with such issues with latent variables exist big limitations there. This article describes the latent variable statistical modeling techniques, which can be used to effectively deal with testing significant variables, latent variables and the relationship between the latent variables. The multivariate statistical methods will work in economics, management science, psychology, behavioral sciences and other wide range of areas.This article describes the latent variable statistical modeling techniques in various stages of development. In the factor analysis the latent variables are extraced. In the canonical correlation analysis, linear relationship between the latent variables is analyzed. In the principal component regression analysis, qualitative analysis of latent variable dependent variable and partial least-squares regression analysis, latent variables are used as an intermediate variable. In the structural equation model, the regression between latent variables is analyzed. The latent variable modeling techniques were being consummating. The paper also used the example of the business environment illustrates the application of latent variable. At last, the latent variable modeling approach for the development of evolution is summarized, the characteristics of various methods were compared and the latent variable modeling methods and applications for further development outlook were mentioned. |
