Mplus

Mplus is a statistical modeling program with a focus on the estimation of latent variables developed by Linda Muthén and Bengt Muthén. It is a highly flexible program capable of estimating a wide variety of models and offering a range of estimators, options, and special features. Thus, Mplus can be useful for researchers with diverse needs related to their particular data and analytic framework. Mplus utilizes latent variables—unobserved variables inferred from observed data—to model hypothesized associations among concepts. The particular utility of Mplus is that the program models both continuous and categorical latent variables. Continuous latent variables are used in structural equation modeling (SEM) and similar techniques (e.g., path analysis, confirmatory factor analysis), as well as growth modeling, survival analysis, and time-series analysis. Categorical latent variables are utilized to estimated mixture models and similar techniques (e.g., latent class analysis, latent class growth analysis).

Because of its focus on latent variables, Mplus has been used fairly extensively by researchers employing SEM. Mplus has a number of utilities that make it useful to these researchers, including a variety of options for handling missing data, multiple estimators and algorithms, the ability to conduct Monte Carlo simulation studies, graphical tools, and a language generator. The Mplus website and user guide provide a large pool of sample studies and syntax. The publishers themselves are highly responsive to users, often providing fairly individualized responses in the Mplus discussion boards. However, there are also important limitations researchers should consider, including the closed-source nature (and therefore cost) of the program, the restricted compatibility with non-Windows operating systems, and data management weaknesses. This entry first describes the framework and capabilities of Mplus and then highlights its advantages and limitations.