Using Individual-level Parameter Estimates to Investigate Evidence of Lexicographic Behaviour and Heterogeneity in Information Processing Strategies
J Dumont, Resource Systems Group Inc., US and ITS, University of Leeds, UK: T J Adler, Resource Systems Group Inc., US; S Hess, ITS, University of Leeds, UK; W C Neafsey, Ford Motor Company, US
In this paper, we propose a way to use conditional estimates from Hierarchical Bayesian (HB) models to infer the specific strategies used by individual respondents.
There is growing interest in understanding how different respondents may be processing the information presented to them in surveys in heterogeneous ways. In particular, the focus has been on the possibility of a given respondent making his/her choices on the basis of only a subset of attributes, i.e. ignoring some of the attributes. In the present paper, we propose a way to use conditional estimates from Hierarchical Bayesian (HB) models to infer the specific strategies used by individual respondents. In contrast with previous work, we do not treat parameters in isolation, but look in detail at what combination of parameters for a given respondent is sufficient to adequately explain their choices. We show how this approach then also allows us to draw conclusions as to the potential use of lexicographic decision rules by some respondents.
Association for European Transport