When people make repeated choices in stable environments, decision-making may shift from deliberate evaluation to a more automatic, heuristic-driven process. This could lead to reduced attention to certain attributes, potentially resulting in suboptimal choices. Despite its relevance to real-world decisions—such as consumer behavior, policy interventions, and economic modeling—traditional preference estimation methods often assume stable attribute weighting over time, overlooking the possibility that attention may shift as choices become habitual.

In this study, we investigate whether and how attention to choice attributes evolves in a large-scale discrete choice experiment (DCE) with over 100 decisions per participant. Using eye-tracking data, we test whether attention to attributes decreases systematically as choices are repeated. We analyze response times and information sampling patterns, examining whether individuals become more selective in what they attend to. We also explore whether altering the order of information presentation can re-engage attention, suggesting that habitual decision-making may not be entirely fixed but responsive to contextual changes.

To formally model these effects, we use a latent class approach to estimate the probability of attribute non-attendance (ANA), incorporating eye-tracking data to capture within-individual variation in attention over time. Our results provide new insights into whether decision-making moves away from fully compensatory behavior in repeated choices and whether attentional adaptations can be influenced by simple contextual interventions. These findings have important implications for policy design, consumer research, and behavioral interventions aimed at improving decision quality by strategically guiding attention.