Towards greater transparency in selecting cost vectors for discrete choice experiments in the context of food choice

This study aims (i) to increase the evidence base on cost vector effects with a focus on food choice; (ii) to offer a novel and more transparent approach to defining cost vector bounds via scanner data; and (iii) tests the sensitivity of choices and WTP estimates to different cost vector definitions across two datasets (i.e., lamb chops and porridge). The data come from two web-based surveys for a choice experiment (one on lamb and the other one on porridge). The modelling approach to analyse the choice data for different treatments is based on the random utility theory and uses a random parameter logit model. The main findings are: compared to a ‘baseline’ treatment (T1 - regular prices), (i) increasing the upper bound of the cost vector results in an increase in WTP; and (ii) decreasing the lower bound of the cost vector also results in an increase in WTP. We find a greater number of significant differences in marginal WTP if the difference in the upper bound between ‘baseline’ (T1) and other treatments is greater.