Several ranking agencies evaluate universities by assigning scores and creating composite indices based on attributes such as outputs, inputs, and quality proxies. These rankings reflect the three primary missions of universities: teaching, research, and the “third mission,” which encompasses activities such as spin-offs and community engagement. However, the values that form these indices are often presented as ordinal rankings, which involve technical decisions—particularly the weighting of each attribute—that can influence their reliability and comparability. The three most prominent global indices are ARWU, QS, and THE; they measure similar phenomena using different methods, variables, and weights, yet their results tend to be correlated. Which indicators most strongly explain the variance in global rankings, and what are their implicit importance weights? Moreover, to what extent can these rankings be replicated using publicly available indicators, and how can efficiency analysis identify opportunities for improvement for individual universities and for individual indicators? To address this, we employed Benefit-of-the-Doubt (BoD) models based on Data Envelopment Analysis (DEA), incorporating weight restrictions and a multi-directional extension (MDir_RBoD) to evaluate indicator-specific efficiency. Our analysis of 2024 datasets showed that ARWU emphasized publication outputs, THE focused on industry impact and research quality, while QS highlighted internationalization and sustainability. Unlike previous broad assessments, our approach provided detailed diagnostics of each indicator’s contribution and areas for improvement. Using the MDir_RBoD model, we proposed practical benchmarks for universities and policymakers, enhancing their understanding of efficiency and strategic improvements in ranking, thereby contributing to more transparent and comparable university evaluations.

Acceder al documento: https://link.springer.com/article/10.1007/s11192-026-05663-5