Home‎ > ‎

Applied Analysis Notes

These notes address applied data analysis challenges using Mplus. Mplus is a powerful modeling software package, but has extremely limited data management capabilities. We use Stata for our general data analysis and management tasks. Most example code reflects our integration of Mplus within the Stata environment.

Note 1: Computing model-implied or expected scores in a growth modeling context from SAVEDATA/FSCORES. 
Tags: latent growth curve model, Mplus, Stata, runmplus, savedata, runmplus_load_savedata, fscores, predicted values, model-implied values, centering, ado, example, simulation

Note 2: Computing model-implied or expected scores from parameter estimates. (coming soon)
Tags: model-implied values, centering, example, simulation

Note 3: Getting factor scores using a fixed set of parameter estimates.
Tags: factor analysis, factor scores, item response theory, runmplus, savedata, fscores, runmplus_load_savedata, Mplus, Stata
Tags: parameter estimates, latent growth model, runmplus, LaTeX, Mplus, Stata

My favorite measure of fit and rules of thumb for judging good fit

After running a latent variable model with covariates, use the r2 from the STANDARDIZED part of the output to re-calibrate the latent variable variance to 1 by constraining the residual variance to 1-r2. This results in more easily interpretable effects of covariates, makes converting to IRT parameter estimates easier (see Macintosh & Hashim 2003), and makes quicky generating plots using canned functions for theoretical standard normal distributions for the latent variable.
Tabs: IRT, DIF, MIMIC, parameter estimates, Mplus