Inferred vs. stated attribute non-attendance in choice experiments: A study of doctors’ prescription behaviour

It is increasingly recognised that respondents to choice experiments employ heuristics such as attribute non-attendance (ANA) to simplify the choice tasks. This paper develops an econometric model which incorporates preference heterogeneity among respondents and allows the probability of non-attendance to depend on the respondents’ stated non-attendance. We find evidence that stated ANA is a useful indicator of the prevalence of non-attendance in the data. Contrary to previous papers in the literature we find that estimates of marginal rates of substitution derived from models which account for ANA are similar to the standard logit estimates.