In his paper, "On Conjoint Measurement and Quantal Choice Models," Albert Madansky (this issue) points out that conjoint analysis' usually deals with each respondent separately. It is an individual level analysis in the sense that the idiosyncratic parameters for each individual are estimated using only the preference judgments of that individual (for an illustration, see Parker and Srinivasan [1976, p. 1009]). On the other hand, quantal choice models are usually estimated at the aggregate level in the sense that a common set of parameters is estimated from the choice data of a sample of individuals (for an illustration, see McFadden [1976, p. 130]). Variation in the utility function across individuals is incorporated in such quantal choice models through a vector of measured characteristics of the individual. In this comment, we first discuss the rationale for the popular use of individual level analysis of preferences in marketing. Although conjoint analysis is usually carried out at the individual level and quantal choice approach is usually carried out at the aggregate level, it is possible to use conjoint analysis at the aggregate level (e.g., see Srinivasan and Shocker 1973) and use quantal choice analysis at the individual level (e.g., see Jain et al. 1979). The second part of this comment discusses some of the limitations of a popular quantal choice approach, namely, the LOGIT model, in individual level analysis.