This paper aims to contribute to the literature on non-attendance of choice attributes in discrete choice experiments. Using simulation experiments, we analyse the effects of choice task and serial non-attendance on the variances of the estimated parameters and, the reliability of inferred non-attendance based on respondent-specific coefficient distributions obtained through conditioning on observed choices. According to the simulation experiments results: firstly, choice task non-attendance of one attribute causes biases in the estimation of all other parameters; and, secondly, only serial non-attendance can be successfully inferred. Subsequently, these results are tested in a real data concerning the landscape externalities of wind power generation.