This study comprehensively compared and examined the model estimation and market share prediction accuracy of the preference regression model, the DEFENDER model, the hierarchical elimination model, and the Batsell and Polking (1985) model, using choice data obtained through questionnaires as actual market shares. According to the research results, the hierarchical elimination model, which considers all possible choice sets, best reflects actual market shares, while the preference regression model shows the lowest predictive power. In applying the DEFENDER model, the assumption of homogeneous perceptions was insufficient to satisfy actual choice data, and along with the preference regression model, it was found to underestimate market shares for products in relatively inferior positions on the perceptual map. Gensch and Ghose (1992) proposed the EBD model using the approach of Batsell and Polking (1985), claiming that its predictive power was higher than that of the hierarchical elimination model estimated based on paired comparison data. They pointed out as a structural weakness of the hierarchical elimination model that pairwise preference ratios equal market share ratios. However, this study argued that their claims originated from an inadequate understanding of both the hierarchical elimination model and the Batsell and Polking (1985) model, and demonstrated that the lower predictive power of the hierarchical elimination model using only paired comparison data is not attributable to a structural weakness of the model but rather stems from differences in the number of choice sets utilized. This study presented the possibility, unlike prior research, of utilizing all possible choice sets for parameter estimation in the hierarchical elimination model. Methods for marketing managers to easily estimate each model in practice were presented, and the strengths and limitations of each model were discussed.