Distinguishing between dimensions and categories is often not the sole purpose of a study. Class-specific parameters, between-class differences, class proportions or the assignment of individuals to classes are usually of great interest. There is a clear relationship between the assumptions that are acceptable and the level of information that can be gained from an analysis. If prior research provides information on how to specify a structural equation model within each class, then the payoff in terms of information gained from an LVMM analysis can be substantial. Parameter estimates of the within-class factor structure, factor means and variances, class proportions and covariate effects directly result from fitting a model to the data. Taxometric procedures, which can be appropriate in case of more limited knowledge about the data, are also more limited regarding their output. The taxon and complement proportions are deducted from the average hitmax of the different plots, and means and variances of observed variables within taxon and complement have to be computed post hoc by a hard partition of the sample at hitmax (Meehl, 1995).
