Comparative Analysis of Normalization, Weighting, and Aggregation Methods for Composite Indicators: Evidence from Social Responsibility in Brazilian Municipalities

This study presents a comparative analysis of alternative methods for constructing composite indicators, addressing methodological challenges related to normalization, weighting, and aggregation. Using the five dimensional scores of the 2021 Social Responsibility Index for 853 municipalities in Minas Gerais, Brazil, we evaluate the TOPSIS, DP2, ADRIANA, and Mazziotta-Pareto Index methods, considering both their original formulations and versions incorporating D-CRITIC weights. The comparison combines score distributions, Spearman rank correlations, average absolute differences in ranking positions, and categorical classifications of municipalities. The results show that, despite distinct mathematical formulations, the methods produce similar score distributions, high rank correlations, and comparable classifications, indicating strong robustness to methodological choices. These findings highlight the stability and usefulness of unconventional composite indicator methods for evaluating socioeconomic and public policy conditions at the municipal level.