International Choice Modelling Conference, International Choice Modelling Conference 2015

Font Size: 
Disentangling the Influence of Knowledge on Preferences and Processing Strategies
Erlend Dancke Sandorf, Danny Campbell, Nick Hanley

Last modified: 11 May 2015


This paper investigates the connections between three important strands of the choice modelling literature. The first concerns the effects of information and knowledge on stated preferences, an issue which has been of interest since the beginning of contingent valuation applications to environmental goods for which respondents often have no direct experience in consuming. The second concerns the ways in which respondents make choices under complexity, and the extent to which they fall back on heuristics to simplify their decisions. This has led to a focus on attribute non-attendance as one heuristic which seems quite common empirically. Attribute non-attendance is related to the third strand of the choice modelling literature, which is concerned with the nature of the utility function and whether people are indeed willing to make trade-offs between all attributes which are used to describe their choices.


It is common within environmental economics to elicit preferences for environmental goods for which there is both limited scientific knowledge and low public awareness. This poses a problem for the use of stated preferences as part of a benefit-cost analysis of public policy choice, since it implies making policy recommendations based on the preferences of “un-informed respondents”. For “goods” such as biodiversity conservation, it is therefore important to provide information about the relevant aspects of the environmental good prior to the valuation task in a way that is meaningful to respondents. “Valuation workshops” (MacMillan et al, 2002; 2006; Alvarez-Farizo et al, 2007) offer one way to achieve this, since they involve  presentation of relevant information on the topic at hand and provide an opportunity for respondents to ask clarifying questions, before answering the questionnaire. Another advantage of the valuation workshop is the possibility of questioning and quizzing the respondents on the material described to them, so that their knowledge and comprehension of it can be determined. This provides a way of measuring the ex ante knowledge of respondents before their choices are made, and of comparing information provided with knowledge acquired (LaRiviere et al, 2014).


In this paper, we carry out a choice experiment on the preferences of citizens for a very unfamiliar good, namely cold water corals. Such corals are found in deep waters around many coastlines, and are valued by ecologists as biodiversity hotspots in the deep sea. However, their condition world-wide is threatened by deep sea trawling, oil and gas exploration, and deep water mining. Using a series of valuation workshops, we explore the link between the answers the respondents gave to a quiz, yielding a measure of their knowledge, and their choices in a discrete choice experiment. More specifically we, we implement a discrete and continuous mixture model to disentangle the influence of knowledge on preferences and processing strategies. For this, we segment the data according to quiz score (i.e. above or below the median and then later into quartiles), and these are then used as covariates to explain membership to different latent classes.


The effort required by a respondent in a discrete choice experiment is increasing in choice task complexity (e.g. number of attributes, levels, alternatives, tasks and complexity of the good). The more effort that is required on the part of the respondent in a discrete choice experiment, the more likely it is that he or she will use a simplifying strategy and heuristic. One such strategy, which is the focus of this paper, is simply to ignore one or more of the attributes when choosing between alternatives in the choice task, also known as attribute non-attendance.


We argue that the respondent’s knowledge about the environmental good under evaluation influences perceived choice task difficulty, in that more knowledge reduces the effort required through reduced difficulty of the choice task. Following from this, we hypothesize that the degree to which respondents simplify by ignoring attributes is influence by knowledge of the environmental good. Specifically, the hypothesis to be tested is that the more knowledge one has about the environmental good, i.e. the higher one’s quiz score, the less likely an attribute is to be ignored in making choices. There is some evidence suggesting that prior knowledge of a good and the relevancy of the information provided prior to the discrete choice experiment may avoid the expression of discontinuous preferences. Our setup allows us to test this specifically.


As noted above, we test this hypothesis using data from a discrete choice experiment on cold-water coral in Norway – an environmental good for which public awareness is low. The data was collected in valuation workshops. In the valuation workshops, the respondents were given a presentation on cold-water corals and then given a quiz of eight questions over the material covered in the presentation, prior to completing the valuation task. This quiz provides us with a measure of the respondent’s knowledge.


Our results shed some light on the effect of knowledge on the use of processing strategies. This has implications for the provision of information in discrete choice experiments in the future, as well as for the modelling of the effects of knowledge on attribute non-attendance.



Conference registration is required in order to view papers.