USE OF INDIVIDUAL DIFFERENCES SCALING FOR SENSORY STUDIES: SIMULATED RECOVERY OF STRUCTURE UNDER VARIOUS MISSING VALUE RATES AND ERROR LEVELS

Abstract

ABSTRACTThe use of Individual Differences Scaling in sensory studies has been limited because the number of pairwise comparisons in a complete design can be prohibitively large. This study examines the effect of different levels of error and missing data on recovery of true structure by Individual Difference Scaling for the normal range of assessors and stimuli (8–16) used by sensory workers. Recovery was very good for a missing value rate of 20%. It was good for 40% provided that the number of stimuli was high, but deteriorated sharply at 60% missing data. The effect of superimposing random error on to the true structure indicated that a large error rate gave poor recovery. It is concluded that up to 40% of the distances in a complete design could be excluded, provided that replicate measurements indicated error levels were not severe.