A Comparative Study of Ordinal Probit and Logistic Regression for Affective Product Design

Abstract

Affective product design, which concentrates on customers’ affective responses and aspirations, is arousing increasing attention. In this paper, ordinal probit regression (OPR) is introduced into this field to discover mapping knowledge from design elements to customer affect, and a comparative study is always recommended between OPR and ordinal logistic regression (OLR) for available data to choose a better fitted model. The discovered mapping relations could facilitate the handling of affective information and assist the designer to make trade-off decisions. A case study of cell phone design was conducted. Four generic affective dimensions and six key product attributes of the cell phone were identified. OPR and OLR were applied successively to reveal the quantitative relations from design elements to customer affect. For the two models, five classes of indexes were compared. The findings show that OLR is superior to OPR to fit the collected data.