A Preliminary Assessment of the Robustness of Signal Detection Theory Estimates Using Monte Carlo Simulations for Human Factors Professionals

Abstract

Signal detection theory is commonly utilized in the field of human factors. Despite its common use, the assessment of the signal detection theory assumptions is not often cited. The purpose of this research was to provide a preliminary assessment of the impact of assumption violations on estimates of sensitivity commonly used in signal detection theory research. This assessment was performed using Monte Carlo simulations. Our research indicated that violating the homogeneity of variance assumption resulted in estimates of sensitivity varying with changes in the response criterion. However, an unequal number of signal and noise trials, which is common in vigilance research, did not impact the estimates of sensitivity. Based upon our findings, caution should be taken with regard to violations of homogeneity of variance. Future research aims to determine the impact of multiple assumption violations on estimates of sensitivity.