## Latent Variable Path Analysis

Latent Variable Path Analysis is one of the popular statistical methodologies that is extensively used by quantitative social scientists. The popularity of this statistical technique is confirmed by the creation of a dedicated scholarly journal and existence of SEMNET (a very popular and active electronic discussion list) that focuses on Latent Variable Path Analysis and related issues.

Latent Variable Path Analysis is a widely used statistical methodology for theoretical and applied statistical research. There has been remarkable developments in the statistical theory underlying structural equation modeling during the last 40 years, including software developments that allow flexible and sophisticated modeling under nonstandard conditions of the data. In addition, recent developments have allowed traditionally different approaches to statistical modeling to be specified as special cases of structural equation modeling.

The advantages of the structural equation modeling perspective are its tremendous flexibility as well as the incorporation of explicit measurement models into more general statistical models. One of the important modern developments has been the extension of new estimation methods to handle non-normal distributions, it is now possible to estimate the parameters of complex structural equation models when the data are nonnormal—including mixtures of dichotomous, ordered-categorical, and continuous variables. In addition to estimation with nonnormal variables, a long history of methodological developments in structural equation modeling now allows researchers to estimate models in the presence of other data-related problems. For example, one can use standard structural modeling software to estimate the parameters of structural equation models when missing data are not missing completely at random.

Other developments in structural equation modeling have resulted from specifying the general model in a way that allows a “structural modeling” approach to other types of modeling. The most recent example of this development is the use of structural equation modeling to estimate multilevel data— including longitudinal data for the estimation of growth curve parameters. Finally, the merging of categorical latent variable modeling and models for continuous latent variables in cross-sectional and longitudinal contexts constitutes the current “state of the art” in structural equation modeling.

Reference:

Kaplan, David. Structural equation modeling: Foundations and Extensions.

Click Here to Read About **Statistical Significance: The p-value**