A Likelihood-Based Approach to Detecting Aberrant Individuals in Confirmatory Factor Analytic Models