Measuring polarization in the context of ordinal scales

J. L. García Lapresta, D. Gómez González, J. A. Guevara Gil

How to measure polarization in a society is a problem that has received attention from different disciplines and approaches. In this contribution, we focus on the case of individuals of a society evaluating an issue through an ordinal scale formed by linguistic terms (e.g., ‘very bad’, ‘bad’, ‘fair’, ‘good’, ‘very good’ and ‘excellent’). Sometimes these scales are not uniform. To manage non-uniform ordered qualitative scales in a purely ordinal way, García-Lapresta and Pérez-Román (2015) introduce the notion of ordinal proximity measure. In this context, we introduce polarization measures as functions that assign a number between 0 and 1 to each profile of individual assessments, satisfying some conditions. We propose a parameterized class of functions that aggregate the ordinal proximities between the assessments of each pair of individuals and the highest and lowest terms of the ordinal scale. This aggregation process is managed through different power means in several steps.

Palabras clave: polarization, ordinal scales, ordinal proximity measures, power means

Análisis de Datos en CCSS I
3 de septiembre de 2026  09:00
Aula 21

R. de Andrés Calle, J. M. Pavía, P. Pérez-Asurmendi

E. Delahoz-Domínguez, L. M. Carmona Armenta, M. Serrano-Hernandez

L. Escot Mangas, A. Pérez Alonso, M. R. Quispe Quispe

A. Recio Alcaide, C. CASTELLANOS SERRANO, M. Pazos Morán