Inferring attribute non-attendance using eye tracking in choice-based conjoint analysis

Abstract

Traditionally, the choice-based conjoint analysis relies on the assumption of rational decision makers that use all available information. However, several studies suggest that people ignore some information when making choices. In this paper, we build upon recent developments in the choice literature and employ a latent class model that simultaneously allows for attribute non-attendance (ANA) and preference heterogeneity. In addition, we relate visual attention derived from eye tracking to the probability of ANA to test, understand, and validate ANA in a marketing context. In two empirical applications, we find that a) our proposed model fits the data best, b) the majority of respondents indeed ignores some attributes, which has implications for willingness-to-pay estimates, segmentation, and targeting, and c) even though the latent class model identifies ANA well without eye tracking information, our model with visual attention helps to better understand ANA by also accounting for differences in attribute processing patterns.

Does attribute order influence attribute-information processing in discrete choice experiments?

“…Most other studies (e.g. Balcombe et al, 2015;Yegoryan et al, 2018) suggest that visual attention based on eye tracking is not distributed equally across attributes. Balcombe et al (2015) showed that next to the last (monetary) attribute, the first and the last non-monetary attributes have the highest visual ANA.…”

Boundaries of Solidarity: Immigrants, Economic Contributions, and Welfare Attitudes

“…The experiment focused on unemployment support, a salient political issue in the countries. This should increase the likelihood that respondents consider the topic plausible and relevant, an important precondition to satisfy the assumption that decision makers in conjoint analysis use the information provided (Yegoryan, Guhl, and Klapper 2018).…”

“…The experiment focused on unemployment support, a salient political issue in the countries. This should increase the likelihood that respondents consider the topic plausible and relevant, an important precondition to satisfy the assumption that decision makers in conjoint analysis use the information provided (Yegoryan, Guhl, and Klapper 2018). I then analyzed the data using ordinary least square regressions with cluster-robust standard errors because each respondent evaluated four pairs of unemployed individuals.…”

Preferences as a Determinant of the Measurement-Unit Effect for Unit Prices

“…Since choosing a preferred product profile (from a set of alternative offers) is an everyday occurrence for consumers, the Choice Based Conjoint (CBC) approach is attributed to represent a real-life purchase situation more realistically than traditional rating- or ranking-based conjoint approaches (Papies et al, 2011; Toubia et al, 2004). Accordingly, the ability of CBC to identify consumer preferences is widely recognized in marketing research (Hauser et al, 2019; Hennig-Thurau et al, 2007; Meyer et al, 2018; Yegoryan et al, 2020). Moreover, the given context of pricing research as well as the small number of product attributes considered in the study design argued for preferring CBC to alternative methods.…”