THE DISPERSIONS OF ESTIMATES OF SENSITIVITY OBTAINED FROM 4 PSYCHOPHYSICAL PROCEDURES - IMPLICATIONS FOR EXPERIMENTAL-DESIGN

The dispersions of estimates of sensitivity obtained from the yes-no, two-alternative forced-choice (2AFC), matching-to-sample, and same-dif ferent tasks were examined to determine which task would be more appro priate to use in a given experimental context. Consideration was given to the effects of corrections for extreme sampled proportions. These corrections result in biased estimators, and hence the mean square dev iation of the sampled values about the population mean [MSD((d) over c ap')], rather than that about the mean of the estimates [VAR((d) over cap')], indicates more completely the extent of the error in the estim ator. For barely discriminable events (d' similar or equal to 0.5), th e yes-no and BAFC tasks had the lowest values of MSD(d'). However, for very discriminable events (d' > 3), the same-different and matching-t o-sample tasks had lower values of MSD((d) over cap').