In the United Kingdom, the designs used in value of time studies have typically been very simplistic, presenting respondents with a choice between a limited number of alternatives (often only two) described by a limited number of attributes (often only time and money). While there are many compelling reasons for not wanting to burden respondents with too much information, there are also grounds to suggest that simplistic designs may not replicate real world behaviour. Furthermore, information relevant to respondent's choices that is not captured in SC surveys is forced into the unobserved effects of estimated models, and as such, simple designs that omit relevant information may place more emphasis on the unobserved effects of discrete choice models. Conversely, the opposite may also be true with too much information also impacting on the unobserved effects of these models. Thus, a balance between providing too little relevant information or too much information to respondents is important when conducting SC surveys. Unfortunately, most studies do not vary the information provided or vary only one dimension of information at a time and hence, tests of the impacts of such information cannot be conducted. Evidence is presented from a studymaking use of state of the art survey techniques in a route choice context in the United Kingdom. This represents a departure from the traditional studies undertaken which typically employ ad hoc attribute level combinations or orthogonal designs, and employ designs whereby only a small number of attributes and alternatives are presented. In this study state of the art experimental design methods are employed in constructing the SC survey.In a further departure from currently employed approaches, the study, which is being carried out in Bristol, presents respondents with varying levels of design complexity, changing the number of choice sets, alternatives and attributes as well as the number of attribute levels. This enables theinvestigation of the differences in behaviour depending on choice task complexity, with results on preliminary data suggesting that respondents areable to cope with far more complex choice situations than those traditionally used in UK studies. For the covering abstract see ITRD E145999
Samenvatting