11.4.4.2 Comparison of Products with IdealA study might include CATA evaluations for both real products and a (hypothetical) ideal product. Cowden et al. (2009) compare the proportion of elicitations for the real products with the ideal product. The approach involves using a confidence interval for the ideal but not the real products, which ignores the statistical uncertainty about the ideal product. Meyners et al. (2013) build on this approach, taking the difference between the elicitations for the ideal and the real products and a confidence interval for the difference. Effective sample size has to be taken into account: Only assessors that discriminate between the products provide pertinent information; thus, confidence intervals are ofdiffering widths. Visual display of these data reveals the attributes for which the real products differ from the ideal.Additionally, Cowden et al. (2009) propose to ask consumers to rank the three attributes they consider to be most important in their ideal prod- uct. The percentage of respondents that ranked an attribute first (called strength) is then plotted against percentages selected in CATA (called interest), enabling a prioritization of high-strength, high-interest attributes to be identified and refined in the subsequent reformulation process.11.4.4.3Penalty Analysis
đang được dịch, vui lòng đợi..
