A new model for preference aggregation in resource allocation problems

• Grant number: 25/02902-9
• Support Opportunities: Scholarships abroad - Research
• Start date: July 26, 2025
• End date: September 26, 2025
• Field of knowledge: Engineering - Production Engineering
• Principal Investigator: José Geraldo Vidal Vieira
• Grantee: José Geraldo Vidal Vieira
• Host Investigator: Gilberto Montibeller Neto
• Host Institution: Centro de Ciências em Gestão e Tecnologia (CCGT). Universidade Federal de São Carlos (UFSCAR). Campus de Sorocaba. Sorocaba , SP, Brazil
• Institution abroad: University of Bristol, England

Abstract

Designing preference aggregation for resource allocation problems, which involve competing objectives (i.e., resource attributes) and account for individual priorities, is a challenging task. Ensuring adequate representativeness for a large group of people adds another layer of complexity. To achieve this representativeness, preferences are [increasingly] collected through [online] surveys. As a result of this procedure, preference judgments typically lead to an individual ranking of the resources provided by each respondent. In decision analysis, various methods can be used to derive community attribute weights from individual priorities. This can be achieved through pairwise comparisons (for ordinal rankings) or direct ratings, which yield cardinal rankings (indicating the ratios between weights). The former is rank-based methods, which are straightforward, require limited cognitive effort and can be applied to a large-scale survey. However, rank-based methods only provide ordinal rankings for individual priorities. The latter are swing methods, which may be prone to poor-quality responses due to their sensitivity to range.Additionally, swing methods often lack specific trade-offs among resource attributes, potentially leading to underestimating the weight of less critical attributes. While rank-based methods may offer more reliable individual priorities, aggregating these preferences is challenging, as they typically rely on mean operators or procedures to convert ordinal data into cardinal numbers-leaving the aggregated preferences open to legitimate scrutiny. Therefore, this research will focus on the preference aggregation of individual priorities for resource attribute allocation using pairwise comparisons. We base this proposal on the previous research, "A survey-based priority elicitation protocol for community-based resource allocation decisions", which captures strict individual preferences from the community using a large survey by using multi-attribute choices. This proposal suggests a new aggregated preference model in resource allocation decisions that considers strict preference relations. For methodological support, we aim to investigate decision analysis methods, preference learning theory, and methods based on economics theory. Applying these theories, many socio-economic and demographic characteristics of individuals may influence their preferences and the community members' priorities, such as different levels of education, gender, employment, income, etc. These characteristics may be appropriate for preference aggregation analysis. The potential benefits of this research lie in providing an aggregated preferences model that considers the representativeness of the community in the resource allocation context. Additionally, the research could contribute to the important field of online surveys for priority elicitation.