It is often the case that I have data which lends itself to questions about within person (intraindividual) variability and what variables predict individual differences in within-person variation (i.e., interindividual differences in intraindividual variability). It seems to me that I should be able to regress predictors directly on the within-person variance component of a multilevel model in which occasions of measurement are nested within persons, but I can't figure out how to do this in Mplus. The user's guide states that, "For convenience, no distinction is made in how variances and residual variances are referred to in the model command," but I'm not convinced that means you can't directly test group-level variation in residuals.
Mplus does not allow you to have a variance as a dependent variable in a regression equation where the variance is predicted by a covariate. You can however let a variance be different in different groups - so in that sense you can handle a dichotomous covariate (where the regression has no error). This latter case is handled in multiple-group analysis in Mplus.
Where in the output do you find the within person variance and the level 2 covariance in multilevel modeling? I have the variance in initial status and rate of change but not the others as you would have in SAS.