Counting Heads or Measuring Internationalisation? Weighting and Normalisation in Uzbekistan’s National-Ranking Indicator of International Students

This article evaluates the international-student indicator used in Uzbekistan’s national ranking of higher education institutions. The indicator, K = (3T1 + 2T2 + T3)/Tt, weights international students by country group and converts the result into points by dividing each institution’s value by its category maximum. Using official institution-level data for the 165 scored institutions in 2025, the study decomposes K into an unweighted share and a composition factor, tests alternative and random weights, compares four normalisation rules, and examines sensitivity to small numbers and the relation between points and diversity. The unweighted share explains almost all variation in K: the variance of the log composition factor is about 3% of the variance of log K, and the rank correlation between K and the unweighted share is 0.99. Rank orders are robust to the weights (mean within-category ρ between 0.948 and 0.998), although alternative weights move up to 36 institutions by more than 0.25 points. The normalisation rule matters more: under the official rule, 69.1% of institutions receive less than 10% of the available points. Three quarters of institutions host fewer than 30 international students; five additional students can add up to 0.79 points, and the number of students needed for one point ranges from 6 to 3,782. Points are unrelated to diversity (ρ = 0.01). The indicator thus measures the volume of international enrolment rather than internationalisation more broadly. The article proposes a minimum-support rule, a less outlier-driven normalisation, a separate diversity component and annual sensitivity reporting.

Hali tarjima qilinmagan