Robustness of taxometric analysis with skewed indicators: II. A Monte Carlo study of the MAXCOV procedure.

Abstract

A small Monte Carlo study was conducted to determine whether MAXCOV analysis, a taxometric method for testing between discrete ("taxonic") and continuous models of latent variables, is robust when indicators of the latent variable are skewed. Analysis of constructed data sets containing three levels of skew indicated that the MAXCOV procedure is unlikely to yield spurious findings of taxonicity even when skewness is considerable. However, care must be taken to distinguish low base-rate taxonic variables from skewed nontaxonic variables.