The priority evaluator is a stated response format, which is in many situations much superior to simple stated choice experiments, as it enables the survey designer to observe trade-offs across multiple choices and as it enables the respondents to express their preferences both categorically as well as continuously. Still it has been neglected ever since the early work of Hoinville (1971) due to the lack of an analysis framework of comparable sophistication and theoretical strength to the multinomial logit model and its extensions.

The recent development of a multiple discrete continuous choice framework by Bhat (2005) seems to fill this gap. This framework allows the analyst to describe the trade-offs between multiple competing uses for the same budget and the size of their use. It was originally developed for the analysis of time use within a day, but is obviously applicable for other structurally similar situations. Still, some elements of the model formulation might be strongly tied to its time use context to make its transfer feasible.

Two recent studies at IVT employed the priority evaluator to gain insights into two other structurally similar but not identical situations: the composition of the fleet of mobility tools (Cars of different types, public transport season tickets for different gasoline price levels) and the annual mileage with each of them and the choice of methods to reduce personal CO₂ emissions of home owners to a externally specified amount. The methods range from home improvement, changes in the mobility tools, reductions in flying to acquiring scare and expensive CO₂ emission certificates.

Both studies involved complex survey protocols and both customised the priority evaluator experiments to the personal situation of the 400 and 220 respondents respectively. The samples were limited, if large for studies of their kind due to the budget constraints of the studies. In both cases stated choice experiments were also conducted, which shed partial light on the issues as hand.

The paper will present the instruments, some descriptives and the results of the multiple discrete analyses, but will focus on the analysis of the residuals and on the validation exercises undertaken. This analysis is directed at improving the model formulation and at identifying the conditions under which it cannot capture the underlying trade-off processes, if any. 

Hoinville, G. (1971) Evaluating community preferences, Environment and Planning, 3 (1) 33-50.

Bhat, C.R. (2005) A multiple discrete-continuous extreme value model: Formulation and application to discretionary time-use decisions, Transportation Research B, 39 (8) 679-707.