Preference elicitation and preference uncertainty: an application to noise valuation

The valuation of environmental impacts through Choice Experiments (CE) has been increasing applied in order to estimate the cost of environmental externalities.
While this valuation technique offers several advantages over other methods, a crucial problem lies in representing the attributes in a manner that can be easily understood by the respondents. Another problem associated with this valuation technique is the assumption that respondents have known and consistent preferences. This thesis relaxes the restraint by allowing respondents to indicate their level of preference certainty. The effect of different attribute representation techniques especially in context of traffic noise is also examined in relation to the level of preference certainty, while the effect of preference elicitation methods on certainty levels is also scrutinised.

Several CE surveys were conducted to evaluate the impact of traffic noise under a residential setting. In order to examine the effects of attribute representation
method on the respondents, two different surveys were undertaken using the location and the linguistic representation techniques. This has been carried out in
conjunction with three different methods of preference elicitation: the binary choice, one stage Likert and two stage Likert methods. Thus for each of the attribute representation methods, different preference elicitation techniques have been employed.

The main purpose of the analyses has been to examine the variation in error structure and the need for error flexibility due to the different preference elicitation
and representation techniques. The results reveal that these components of choice design significantly affect respondents’ decision making and subsequent valuation.
Moreover, different methods of representation also influence the level and cause of preference uncertainty as well the decision process.