Divergent orderings from a single provider: concordance between the QS World University Rankings and the QS Latin America and the Caribbean Rankings, 2024–2027

When one provider publishes global and regional tables from largely the same data, do their orderings agree? Earlier work compared the QS global and Latin American tables in single editions or, for two institutions, over four. Using the official result files of the QS World University Rankings and the QS Latin America and the Caribbean Rankings for the 2024 to 2027 editions, we match the Latin American institutions present in both tables (118 to 192 per edition) and measure the concordance of the two published orderings with Kendall's tau-b, treating institutions that share a published band as tied. Tau-b ranges from 0.650 to 0.706, and between 11.4 and 14.8 per cent of the institution pairs both tables order are placed in opposite order, with no consistent direction across editions; among institutions holding an exact global rank the share is 17.7 to 25.0 per cent. In a fixed panel of 111 institutions, tau-b rose from 0.547 to 0.704, mostly between the 2025 and 2026 editions, through a change in the global ordering that comes from changes in the distribution of the scores of a few global indicators, not in the order in which they rank the institutions, in editions in which QS changed its method. The four indicators the two tables share rank institutions almost identically (rank correlations from 0.945 to 1.000) but on different scales. An index of these four alone agrees more with the global ordering in their global versions than in their regional ones (0.863 to 0.891 against 0.771 to 0.805), yet within the published regional index replacing their regional versions with the global ones lowers concordance (by 0.095 to 0.154). The divergence therefore comes mainly from the indicators only the regional table uses, Staff with PhD, Citations per Paper, Papers per Faculty and, to a lesser extent, Web Impact, whose removal from the regional index raises its concordance with the global ordering, as does that of International Research Network, which carries twice its global weight there; the regional versions of the shared indicators offset part of that effect. The same exercise on the global side shows that these effects depend on both specifications. Over the period, Latin American institutions in the global table fell from 192 to 118 while the regional table grew from 430 to 517. The 59 that left the global table in 2026 came from its bottom bands, but within each band they left more often than the other institutions (88.7 against 69.4 per cent in the lowest), so their position does not explain their exit rate, and the data do not identify its cause. Analysis code and derived tables: https://doi.org/10.5281/zenodo.24293270

Authors

Institutions

Publication Details

Journal
Zenodo (CERN European Organization for Nuclear Research)
Published
2026-11-17
DOI
https://doi.org/10.5281/zenodo.24298512
Primary Topic
Higher Education Governance and Development
Type
preprint
Controls
|||
ALL TIME
JAN
FEB
MAR
APR
MAY
JUN
JUL
AUG
SEP
OCT
NOV
preprint

Divergent orderings from a single provider: concordance between the QS World University Rankings and the QS Latin America and the Caribbean Rankings, 2024–2027

Víctor Hugo Rocco-Toloza
1 links
Zenodo (CERN European Organization for Nuclear Research)
Higher Education Governance and Development
preprint

Divergent orderings from a single provider: concordance between the QS World University Rankings and the QS Latin America and the Caribbean Rankings, 2024–2027

Víctor Hugo Rocco-Toloza
preprint en

Abstract

When one provider publishes global and regional tables from largely the same data, do their orderings agree? Earlier work compared the QS global and Latin American tables in single editions or, for two institutions, over four. Using the official result files of the QS World University Rankings and the QS Latin America and the Caribbean Rankings for the 2024 to 2027 editions, we match the Latin American institutions present in both tables (118 to 192 per edition) and measure the concordance of the two published orderings with Kendall's tau-b, treating institutions that share a published band as tied. Tau-b ranges from 0.650 to 0.706, and between 11.4 and 14.8 per cent of the institution pairs both tables order are placed in opposite order, with no consistent direction across editions; among institutions holding an exact global rank the share is 17.7 to 25.0 per cent. In a fixed panel of 111 institutions, tau-b rose from 0.547 to 0.704, mostly between the 2025 and 2026 editions, through a change in the global ordering that comes from changes in the distribution of the scores of a few global indicators, not in the order in which they rank the institutions, in editions in which QS changed its method. The four indicators the two tables share rank institutions almost identically (rank correlations from 0.945 to 1.000) but on different scales. An index of these four alone agrees more with the global ordering in their global versions than in their regional ones (0.863 to 0.891 against 0.771 to 0.805), yet within the published regional index replacing their regional versions with the global ones lowers concordance (by 0.095 to 0.154). The divergence therefore comes mainly from the indicators only the regional table uses, Staff with PhD, Citations per Paper, Papers per Faculty and, to a lesser extent, Web Impact, whose removal from the regional index raises its concordance with the global ordering, as does that of International Research Network, which carries twice its global weight there; the regional versions of the shared indicators offset part of that effect. The same exercise on the global side shows that these effects depend on both specifications. Over the period, Latin American institutions in the global table fell from 192 to 118 while the regional table grew from 430 to 517. The 59 that left the global table in 2026 came from its bottom bands, but within each band they left more often than the other institutions (88.7 against 69.4 per cent in the lowest), so their position does not explain their exit rate, and the data do not identify its cause. Analysis code and derived tables: https://doi.org/10.5281/zenodo.24293270

Zenodo (CERN European Organization for Nuclear Research)
Universidad Alberto Hurtado (CL)
Higher Education Governance and Development
AI Navigator

Ask Laika to Summarize, Analyze, and Connect papers live on the map.

Summarize Papers & Methodologies

Extract key findings, datasets, and comparative methods across publications.

Benchmark Rankings & Visual Analytics

Rank top research institutions, authors, funders, topics, and journals by Field-Weighted Citation Impact (FWCI) and paper volume with instant charts.

Connect Distant Disciplines

Bridge topological clusters on the map to find hidden collaborative intersections.

Laika access coming soon