Epsilon Archive for Student Projects

Abstract

Attribute non-attendance (ANA) is a common behavioural phenomenon in discrete choice experiments (DCEs) and can lead to biased estimates if not properly accounted for. This study investigates ANA in two applied contexts using data from farmer and investor choice experiments.
The analysis combines descriptive indicators of stated ANA with latent class models to infer non-attendance behaviour from observed choices. In addition, the relationship between stated and inferred ANA is examined to assess the consistency between self-reported and behaviourally derived measures. The results provide consistent evidence of attribute non-attendance in both datasets, with variation across attributes and contexts. For farmers, financial incentives play a central role in decision-making, while for investors, risk is a key determinant of choices. The comparison between stated and inferred ANA reveals limited alignment, suggesting that self-reported behaviour may not fully capture underlying decision processes. Overall, the findings highlight the importance of accounting for ANA in DCE applications and provide insights into how attribute processing differs across contexts.