Using a Rank-Rating method, d' values were obtained for stimuli from a
model system using 15-point, 9-point, 7-point and 5-point category sc
ales. For an apple/grape juice system, d' values were compared for 9-p
oint and 7-point scales. There was general agreement on d' values amon
g the scales, except where stimuli had their scores 'bunched' into one
category: this tended to occur for the strongest and weakest stimuli
for Rank-Rating and for the scales with fewer points. Bunching altered
distribution variances and distorted the calculated d' values. Furthe
r study gave d' values for the discrimination of confusable stimuli, u
sing paired-comparison and rating data, thus providing numerical measu
res of their relative sensitivity.