Is mplus able to handled truncated regression. That is, where the outcome variable (criterion y) and the independents (predictors) are both truncated.
For example, where observations with values in the outcome variable below or above certain thresholds are systematically excluded from the sample. Therefore, whole observations are missing, so that neither the criterion value nor the predictor variable value is known.
An example would be where individuals are selected to a job but only if they have a aptitude test score over a certain threshold such that people with low test scores are not employed. The result is both the measure of job performance (criterion y) does not include very poor performers and the independent (aptitude test) does not include low scores.