SEQUENTIAL DESIGN OF SENSORY TRIALS

The design of sensory trials is examined in the context of using infor mation from previous trials to improve the information from each new t rial. The methodological arguments are illustrated by results from fou r large trials. However, the central argument of this paper is that fo r any new trial there will usually be a set of previous trials capable of providing the information on which design decisions can be made. D esign aspects considered are replication, block size, position and car ry over effects and explicit controls. In the analysis of sensory tria l data it is important to examine the assumptions made about how asses sors use measurement Scales. A self-adjusting scale is described which overcomes some problems in the practical use of measurement scales.