Explore the scholarly record.

Abstract

Eye-tracking provides valuable insights into how respondents allocate visual attention and compare attributes in discrete choice scenarios. Such information can complement conventional choice data by revealing latent decision-making strategies and improving understanding of preference formation. However, eye-tracking data are typically collected in small-scale laboratory experiments due to financial and time constraints. The resulting limited sample sizes often lack the statistical power to detect the effects of key predictors and limit the generalizability of the results to broader populations. This paper proposes a framework to scale eye-tracking insights from the laboratory to large-scale online studies without requiring gaze data collection. Using laboratory-collected eye-tracking data, we identify how heterogeneous attention patterns are shaped by individual-level determinants that can be collected in both lab and online settings, such as socioeconomic characteristics and attitudes. These relationships are then used to infer latent constructs of decision-making strategies in a larger online sample. The framework is applied to examine how decoy-based nudging strategies influence the decision-making process and, in turn, shape electric vehicle preferences among Singaporean ride-hailing drivers. Our results show that i) individual-level determinants can effectively segment respondents with distinct attention patterns, and ii) transferring visual fixation patterns to a larger online study substantially improves the model fitting performance and ability to predict preference shifts under decoy effects.