Investigating heterogeneity in food risk perceptions using best-worst scaling

Abstract

The psychometric paradigm has dominated the field of empirical work analysing risk perceptions. In this paper, we use an alternative method, Best-Worst Scaling (BWS), to elicit relative risk perceptions concerning potentially unsafe domestic food behaviours. We analyse heterogeneity in those risk perceptions via estimation of latent class models. We identify 6 latent segments of differing risk perception profiles with the probability of membership of those segments differing between experts and the lay public. The BWS method provides a practical approach to assessing relative risks as the choices made by the participants and subsequent analysis have a strong theoretical basis. It does so without the influence of scale bias, the cognitive burden of ranking a large number of items or issues of aggregation of data, often associated with the more commonly used psychometric paradigm. We contend that BWS, in conjunction with latent class modelling, provides a powerful method for eliciting risk rankings and identifying differences in these rankings in the population.