Modeling discrete choices with augmented perception hurdles
提出感知障碍Logit模型，利用受访者在离散选择前提供的重要性评分来增强传统方法，虽样本内拟合不如条件Logit和混合Logit，但在预测保留样本的选择上表现更优，并揭示了受访者可能采用多种决策策略。
Abstract Through creating latent perception hurdles associated with each attribute considered in a stated conjoint experiment, this article describes a model that augments the conventional approach by utilizing the importance ratings provided by respondents prior to the discrete choice stage. The resulting perception hurdle logit (PHL) model has both advantages and disadvantages compared to the conditional logit (CL) and mixed logit models. Although the proposed model may not have the best within‐sample fit, it outperforms the other two models in predicting choices in a hold out sample. In addition, the proposed model is also used to reveal that depending on their own characteristics and the process of the survey, respondents may employ an array of different decision strategies.