The top internationally ranked universities in 14 countries, each coded for control (public or private), the standard annual charge to a domestic first-year student relative to GDP per capita, and whether means-tested support exists — by who funds it (government or institution) and by instrument (a reduction of the charge, or cash paid to the student). Dot size shows enrolment within each country. Country types follow a rule fixed in advance.
Shares of the selected institutions, three ways. Pooled: every institution counts once, so countries with more ranked institutions weigh more. Per country: the share within each country, averaged with equal weight per country. Per country, enrolment-weighted: within each country, institutions weigh by their enrolment, then countries are averaged with equal weight; countries with an institution lacking enrolment are left out of that column. Price is grouped into three categories that add up to 100% within each block (per-country columns up to rounding); the 7 institutions without an established price are excluded from every price denominator. Means-tested support (tested on household income; loans, merit-only and in-kind support excluded) is shown only for institutions whose price is above the free band — a free institution has nothing to reduce and is left out as missing. "Unknown" stays in the denominator (unknown is not no), so those shares are lower bounds. Control is strict (ownership or appointment); government-dependent private universities count as private unless stated.
Institutions without an established price are excluded. A country is placed in the cell that holds the largest number of its priced institutions (or, enrolment-weighted, the largest share of their students); a country tied between two cells counts half in each (marked ½). Uruguay is placed on Udelar alone because its three private universities publish no price. This grid complements, and does not replace, the country types below (public share × median band).
Control type from the public share of the selected institutions (4–5 of 5 public, 2–3 mixed, 0–1 private). Price type is the median band of the headline charge. The small 2×2 grid next to each country counts the institutions where means-tested support is documented, by funder (rows) and instrument (columns); hover a country for its sensitivity checks, including the enrolment-weighted variant.
Each dot is one university, placed by its headline charge (standard tuition + compulsory fees, before targeted aid) as a share of GDP per capita. The axis is divided into the five bands, each drawn at equal width, with values placed linearly inside a band; the top band is drawn up to 200% and anything beyond sits at the edge. Dot area is proportional to the institution's enrolment relative to the largest selected institution in the same country (sizes are not comparable across countries); enrolment is first-degree headcount where available, otherwise the flagged fallback shown in the tooltip. Hover or tab to a dot for details.
Every row is backed by documented sources (URL, verbatim quote, locator and a pointer to the collector packet) in sources.csv; the count is shown per row. Aid: yes / no (searched, not found) / ? (unknown) / n/a (fee reduction where the charge is zero). Enrolment marked † is a fallback measure (undergraduate, all-level headcount or FTE); hover for the measure and year.
Internationally ranked research universities, not the universe of nationally selective programmes. Three samples can be shown (switch at the top): the top 5 per country in the QS World University Rankings 2027 by overall rank, the top 5 by QS academic reputation (the survey of academics), and the top 5 in QS 2024 (the first release). Which universities are "top 5" depends on the edition and on the criterion: small private universities score high on QS measures such as the faculty/student ratio and can rank above large public ones in the overall rank, which pulls some countries towards "mixed" or "private". Each sample can be shown with 2023/24 prices or with the latest prices (switch at the top).
The academic-reputation sample ranks universities by the Academic Reputation score of the QS World University Rankings 2027, which carries 30% of the overall QS score. QS builds it from its Academic Survey:
It is therefore a measure of standing among academics, mostly international and pooled over five years. It is not a measure of how selective admission is. Source: QS, "Academic Reputation (Indicator)", support.qs.com, updated 3 March 2026, archived copy of 29 September 2026.
Standard tuition plus compulsory fees paid by every baseline student, after reductions that apply to everyone, before any targeted aid. Housing, insurance and application fees are excluded. Where prices vary by programme, the headline is the midpoint of the published range; where they depend on family income, it is the full-pay rate.
Non-repayable support tested on household or parental income, coded yes / no (searched, not found) / unknown in four cells: government- or institution-funded × fee reduction (waiver, income-scaled fee, or grant paid against the charge) or cash (money paid to the student, e.g. living-cost grants). Fee reduction is "not applicable" where the headline is zero. A general need-based grant usable against the charge (e.g. Pell) counts as fee reduction, and as cash only where a source says money reaches the student. Loans, in-kind support and merit-only awards are excluded. The share of students receiving each instrument is shown where documented. "Unknown" is not "no".
One figure per institution for 2023/24 (or calendar 2023): first-degree headcount where published (), otherwise the institution's undergraduate headcount, all-level headcount or FTE (flagged). Not found for: . Enrolment sizes dots within a country and feeds one sensitivity check (enrolment-weighted country type); the main country type stays one institution, one vote.
Gini index from the World Bank Poverty and Inequality Platform: the latest national survey between 2013 and 2023. Most countries measure income; measure consumption, which gives lower values, so an income-only sample is shown as a check. Argentina has only an urban survey (shown only in the regression table's widest sample); have no survey on the platform.
Each dot is a country: its Gini index (x) against the share of its selected institutions that are public, and the share whose headline charge is free (y). The line is an ordinary least-squares fit across countries (equal weight). With under 50 countries, shares bunched at 0 and 1, and strong regional clustering (Latin America unequal and mixed, Europe equal and public), this describes an association, not an effect of inequality.
Share = a + b × Gini (Gini in points, 0–100). Slope shown per 10 Gini points, with a 95% confidence interval from classical standard errors and a two-sided t-test. Spearman ρ is the rank correlation, with a permutation p-value (20,000 permutations) that does not rely on normal errors.