ON THE ROBUSTNESS OF MAXIMUM-LIKELIHOOD SCALING FOR VIOLATIONS OF THEERROR MODEL

A Monte Carlo study was conducted to investigate the robustness of the assumed error distribution in maximum likelihood estimation models fo r multidimensional scaling. Data sets generated according to the logno rmal, the normal, and the rectangular distribution were analysed with the log-normal error model in Ramsay's MULTISCALE program package. The results show that violations of the assumed error distribution have v irtually no effect on the estimated distance parameters. In a comparis on among several dimensionality tests, the corrected version of the ch i(2) test, as proposed by Ramsay, yielded the best results, and turned out to be quite robust against violations of the error model.