Optimal Approximate Choice Designs for a Two-step Coffee Choice, Taste and Choice Again Experiment

Abstract This work deals with consumers' preferences about coffee. Firstly, a choice experiment is performed on sample of potential consumers. Following this, sensory test involving the tasting two varieties coffee carried out respondents, after which same supplied to them again. An innovative approach for building heterogeneous designs specifically developed case-study, based approximate design theory and compound criterion. Panel Mixed Logit models are used, thereby allowing inclusion correlation among responses; choice-sets proportion respondents according optimal weights. The estimation results model satisfactory, confirming validity proposed approach.