Use of Analyst-Generated Stakeholder Preference Profiles in Multi-Criteria Decision Analysis — Experiences from an Urban Planning Case

Abstract

Multi-criteria decision analysis (MCDA) is a generic term for approaches supporting the systematic evaluation of alternatives in problems involving multiple criteria and stakeholders. One of the most challenging tasks is to gather preference information from stakeholders in a way that both reflects their true opinions and meets the theoretical requirements of the applied MCDA approach. Various techniques have been used in practice, including interviews and decision conferences. In this paper, we present a new cost-efficient approach in which an analyst generates weight profiles for various stakeholder groups. That is, instead of personally asking specific trade-off questions from the stakeholders, weight profiles are developed on the basis of more general preference information collected from the stakeholders. The potential advantages of this approach are: (i) the collection of the preferences using surveys is less laborious than personal interviews or decision conferences and (ii) the risk of cognitive biases in the weight elicitation can be reduced, because the most challenging task of MCDA — assigning weights to the criteria — is left to the analyst, who should be aware of typical biases and how to avoid them. We developed and tested the approach in a contested public decision-making situation related to the development of a new residential area. We utilised the data gathered from the participants of the workshops (21) as well as the data from a web survey including 177 responds via a randomly sampled closed survey, in addition to 484 responds via an open survey. Four preference profiles each having specific weight distributions to criteria were developed, using a multi-stage procedure. Four development alternatives were compared as based on the developed preference profiles. We were able to realise the MCDA process within a very tight time schedule, create plausible preference profiles and summarise each alternative’s pros and cons from different perspectives. However, we also identified several issues which have to be paid more attention in future cases or require further research.