Researchers typically desire that one, or each, subset be ‘‘repre- sentative’’ of the set of all target products. As such, they usually rely on the recommendation of trained assessors (see Hersleth et al., 2005, for example). However, it is possible that the assessors could be unintentionally subjective and struggle to agree on what qualifies as a truly representative subset. Herein, we say a subset of products is representative of the set of all target products if its elements are as dissimilar as possible, where dissimilarity is determined using the Euclidean distance measure. Formally, using their sensory profile, we calculate the Euclidean distance between each target product and argue that the subset of k > 1 target prod- ucts that maximize the Euclidean distance best represents the set of all target products. We form a sensory informed design (SID) by maximizing distance between each target product while also maintaining overall position balance, so that each product occurs the same number of times in each position (or as close to the same number of times as the total number of panelists allows). Of course, we also require that no product is presented to the same panelist more than once.
The remainder of the paper is outlined as follows. In Section 2, we formulate an SID. In Section 3, we review a model-based approach developed for analyzing data with missing values. In Section 4, we apply this model-based approach to two SIDs collected at Compusense Inc., and we conclude with a summary and suggestions for future work (Section 5).