How the Rankings Are Actually Calculated — Indicator Weights, Methodological Controversies, and HKU's Strategic Response

# How the Rankings Are Actually Calculated — Indicator Weights, Methodological Controversies, and HKU’s Strategic Response

One-sentence conclusion: QS, THE, and U.S. News use three completely different yardsticks to measure “how good a university is”^※: QS draws nearly half its weight from reputation surveys, THE deploys 29 granular indicators spanning teaching, research, industry and internationalisation, and U.S. News is almost purely bibliometric. The University of Hong Kong (HKU) places 11th in the world on QS^※, 33rd on THE^※, and 40th on U.S. News^※ (all 2025–2026 latest editions) — a gap of up to 29 places that arises precisely because the three yardsticks are measuring different things.

---

# What Each Ranking Measures: A Guide to the Indicator Systems

Before reading any HKU ranking number, you need to understand how that number was arrived at. The three tables below are the starting point for everything that follows.

# QS’s Nine Indicators (from the 2024 Edition)

From the 2024 edition, QS expanded its indicators from six to nine^※. The five “lenses” and their weights are:

The single most consequential line: the two reputation components together account for 45% — nearly half the total. Citations per faculty contributes 20%, and the faculty–student ratio has been reduced to 10%. This distribution of weights changed significantly in the 2024 edition — more on that in the next section.

# THE’s Five Pillars (2026 Edition)

The THE World University Rankings 2026 edition uses 18 refined indicators^※ grouped into five pillars. A comparison of the core weights:

THE and QS diverge most sharply in two respects. First, THE’s “research reputation” (18%) is far lower than QS’s academic reputation (30%); reputation’s influence in THE is diluted across more dimensions. Second, THE includes faculty–student ratio (4.5%) and institutional income indicators, while its internationalisation component accounts for only 7.5% — far below QS’s combined 15% from three internationalisation metrics.

# U.S. News’s 13 Indicators (Almost Purely Bibliometric)

U.S. News Best Global Universities uses the Clarivate Web of Science database to support 13 indicators^※, with the 2026–27 edition’s data window covering publications from 2020–2024^※. There is no faculty–student ratio, no internationalisation headcount, and “reputation” is limited to “research reputation”. The top three weighted items:

In a single sentence comparing the three systems: QS asks, “What do scholars and employers around the world think of you?”; THE asks, “How do you perform comprehensively across teaching, research, industry and internationalisation?”; U.S. News asks, “Over the past five years, how many papers have you published and how deeply have they been cited?”. The same HKU receives three starkly different answers to these three questions — and there is no contradiction.

---

# The Great QS 2024 Overhaul: How a Methodology Shift Made HKU’s Ranking Jump

The QS 2024 edition was the largest methodology adjustment in nearly two decades, and it is the essential key to understanding HKU’s wild ranking swings in 2023–2025.

# What Changed in the Weights? A Before-and-After Comparison

QS expanded from six indicators to nine in the 2024 edition, redistributing weights substantially^※:

Twenty per cent of the old weight was carved out (academic reputation –10%, faculty–student ratio –10%) to make room for three new indicators. On the surface, this was “adding dimensions”; in substance, it was “swapping out half the yardstick”.

# Why Did HKU Drop Five Places in a Single Year in the 2024 Edition?

HKU placed 21st in the world in QS 2023 (released June 2022)^※, then fell abruptly to 26th in QS 2024 (released June 2023)^※ — a single-year drop of five places, and the media ran stories of an “HKU slide”. But the fall had structural causes, not a decline in institutional strength: data for the three new indicators takes time to accumulate, and the down-weighted faculty–student ratio was precisely an area where HKU had previously been a steady scorer.

Even more telling is this: multiple Hong Kong universities fell simultaneously in the same cycle — HKUST 40→60, CityU 54→70, CUHK 38→47. For four of a city’s top universities to collectively slide in the same year can scarcely reflect a simultaneous weakening of all four; it almost certainly means “the yardstick changed”. This mirrors a more extreme parallel case from the same period: 52 South Korean universities jointly protested the QS 2024 methodology and formed a “University Rankings Forum” to declare a collective boycott^※ — precisely because the faculty–student ratio was cut from 20% to 10%, dealing a heavy blow to South Korean universities that had historically scored highly on that measure, with some institutions dropping two or three hundred places in a single year. QS’s response was that “the methodology changes were transparently disclosed in advance, and a similar approach had already been used in the Asia rankings for several years.”

# Post-2024: Why Did HKU Rebound So Rapidly?

The new yardstick came with one structural advantage for HKU, and it drove the rebound over the next three years: the International Research Network (IRN) indicator measures the geographic diversity of cross-border research partners in published papers^※. HKU has an extraordinarily broad international research collaboration reach; more than 74% of professoriate staff are non-local^※, and the student body hails from 97 countries and regions — exactly the profile IRN rewards. Coupled with genuine accumulation of reputation (see the next section), HKU rebounded continuously from 26th in the 2024 edition to 17th in QS 2025^※, 11th in QS 2026^※, and held at 11th in QS 2027.

---

# Reputation Surveys: The Heaviest Indicator, and Where the Controversy Concentrates

# How Do the Reputation Surveys Actually Work?

QS’s academic reputation questionnaire is sent annually to around 70,000 academics worldwide, asking them to nominate “the top institutions in your field globally”; the employer reputation questionnaire poses a similar question to corporate recruitment heads. These two components, together accounting for 45% of the weight, constitute the largest single block of any QS ranking^※. THE’s “teaching reputation” (15%) and “research reputation” (18%) together add up to 33%, likewise forming its system’s largest block.

The problem is this: reputation surveys do not measure “how good this year’s teaching at this institution is”. They measure “the impression respondents hold of this institution” — and impressions often lag behind reality and are heavily shaped by historical accumulation, linguistic exposure, and brand visibility.

# Four Core Critiques from the Academic Literature

One: Self-referentiality. A 2025 peer-reviewed study published in Frontiers in Education used Australian university data to show that QS’s academic reputation has a correlation coefficient of r=0.98 with universities’ overall scores^※, and that employer reputation has a correlation of r=0.97 with overall scores. Reputation alone can almost perfectly predict the overall ranking; the other seven indicators have very limited independent explanatory power. In a sense, the rankings turn into “reputation rankings”, and reputation itself is often derived from the previous edition’s ranking — creating a loop.

Two: Reinforcing established hierarchies. Reputation surveys inherently favour historically prominent institutions with strong English-language exposure. Higher-education researcher Ellen Hazelkorn has pointed out that rankings are fundamentally “not objective” — the choice of indicators and the assignment of weights themselves reflect the producer’s value judgements, not neutral criteria^※. Universities in non-English-speaking countries face a systemic disadvantage in academic language and media visibility.

Three: Potential for gaming. Reputation surveys face the risk that institutions “guide” the responses. The most notorious case occurred at University College Dublin: its president publicly called on academics to nominate the university in the QS reputation survey; after the episode was exposed, QS specifically amended its policy to ban such behaviour^※. QS has since introduced machine-learning methods to detect anomalous voting patterns, but critics argue that the inherently subjective nature of reputation surveys cannot be fundamentally eliminated.

Four: Doubtful proxy validity of the faculty–student ratio. The same Frontiers in Education study found that QS’s faculty–student ratio has a correlation coefficient with overall scores of only r=0.11^※, the lowest of all indicators — meaning that this indicator, originally intended as a proxy for “teaching quality”, has virtually no predictive power in the actual data. It measures institutional scale and headcount ratios far more than it measures genuine teaching investment.

---

# “Optimising for the Rankings”: Strategic Responses and the Goodhart’s Law Trap

# What Are Universities Doing?

When rankings influence reputation, reputation influences student recruitment, and student recruitment influences funding, it is very hard for universities not to devote energy to “lifting ranking scores”. Strategically motivated behaviours with clear documentary records include:

• Intensive recruitment of highly cited scholars: In 2023 alone, HKU recruited more than 120 outstanding international academics from 13 countries and regions^※, directly boosting QS academic reputation, international faculty, and citations per faculty. In 2024, HKU had 53 academics named on the Clarivate Highly Cited Researchers list, ranking the institution 10th globally^※; in 2025, the number rose to 54, with the institutional rank rising to 13th globally^※.
• Expanding international research collaboration networks: THE International Ranking 2025 edition data showed that co-authored papers between Hong Kong and mainland Chinese universities grew by about 70% in the 2019–2023 period^※, and Hong Kong institutions clearly benefited in THE’s internationalisation indicators as a result. But scholar Futao Huang has noted that mainland academics’ “close research ties maintained with their home institutions” are an important driver of this data — raising a question: does the growth in collaboration volume represent genuine international academic exchange, or “proximal network inflation” driven by geographic adjacency?
• Prioritising English-language publication and global visibility: Ranking systems systematically favour English-language papers (Scopus and Web of Science index a far higher proportion of English-language articles), and institutions are therefore incentivised to channel research output towards English-language journals — sometimes at the expense of depth in locally applied policy research and Chinese-language publication.

# Goodhart’s Law: “When a Measure Becomes a Target, It Ceases to Be a Good Measure”

A 2025 study published in Scientometrics characterised this as a concrete instantiation of Goodhart’s Law in academia^※: once citation counts and publication volumes are directly tied to rankings, institutions develop behaviours such as “super-prolific authors”, paper mills, gift authorship, and closed institutional mutual-citation circles — and these behaviours, while superficially boosting the indicators, substantively erode research integrity. Documented malpractices include: bulk publication in journals with extremely low thresholds, multi-institutional attribution for duplicate counting, gift authorship, and the purchase of ghostwritten papers. This is not merely a “question of ethics”; it is an institutional vulnerability of the ranking system itself.

For HKU, the above issues are more a matter of background risk than of specific documented violations. HKU’s accumulation of research reputation and recruitment of highly cited scholars, on the publicly available evidence, falls within legitimate academic pathways. But this backdrop serves as a reminder: upward movement in ranking numbers cannot be directly equated with proportional improvement in teaching quality or student experience. Research has found that metric-driven internationalisation can sometimes “divert resources away from genuine improvements in teaching and community engagement”^※ — a tension that HKU’s leadership and its students need to examine together.

---

# How HKU Fares Differently Across the Three Rankings: Three Yardsticks, Three Scoring Logics

# Quick-Reference Table: HKU’s Latest Positions in the Three Major Global Rankings

The following are HKU’s latest positions across the three major world composite rankings (2025/2026 release editions):

The three positions span 29 places, a direct reflection of the three systems’ core divergences. QS 11th is driven by HKU’s accumulated reputation and high degree of internationalisation; THE 33rd reflects HKU’s balanced composite performance across teaching, research and citations; U.S. News 40th is the purely bibliometric outcome of five years of paper output (2020–2024) and the proportion of highly cited work.

# HKU’s Relative Strengths and Weaknesses Across the Indicators

Across the three systems, HKU has several clear structural advantages and one persistent weakness:

Structural advantages: A highly internationalised profile. More than 74% of professoriate staff are non-local^※, from 97 countries and regions — this scores directly on QS “international faculty” (5%), “international research network” (5%), “international students” (5%), and THE “international outlook” (7.5%). Research reputation is likewise solid: 54 Highly Cited Researchers in 2025 (Clarivate count, November 2025)^※ provide a steady supply of points to QS citations per faculty and to multiple U.S. News bibliometric indicators.

Persistent weakness: Relatively small scale. As a “small but excellent” institution, HKU is at a volume disadvantage on indicators scored by absolute paper counts — U.S. News “publications” (10%) and THE “research productivity” (5.5%) — when compared with large comprehensive universities in the US, UK, or mainland China. This is the core structural reason why HKU’s U.S. News position (40th) lags 29 places behind its QS position (11th): strip out internationalisation and reputation, compare purely on papers and citations, and HKU’s “smallness” becomes visible.

---

# Three Critical Perspectives to Hold When Reading Rankings

# The Numerical Precision of a Rank Is an Illusion

Higher-education researchers have noted that only about 3% of the roughly 18,000 higher education institutions worldwide make it into the QS top 500^※, and the further down the list an institution sits, the smaller the differences between successive places and the more easily scores of places can jump due to minor data fluctuations. Even within the top 100, adjacent ranks often do not constitute statistically significant differences. To read “11th” and “15th” as two institutions of markedly different strength is a misuse of the apparent numerical precision.

# Year-on-Year Volatility Is Often Methodological Noise, Not a Change in Institutional Strength

The major QS 2024 methodology overhaul caused HKU to fall from 21st to 26th^※, and in the same year 52 South Korean universities lodged a collective protest — for any single year’s change in position, check first whether a methodology adjustment occurred before drawing conclusions about real changes in strength. Equally, the annual “highest ever ranking” news story is sometimes merely the comparative advantage created by a low base from a previous year’s methodological shift.

# Rankings Selectively Define What Makes a “Good” University

THE, QS and U.S. News methodologies all centre on research output; virtually none of the systems incorporate student satisfaction, undergraduate teaching quality, graduate median salaries, or contributions to social mobility^※. These rankings answer the question “how well does this institution perform as a research organisation?”, not “what is the experience of studying here like?”. For readers who are primarily concerned with the quality of undergraduate teaching, student support systems, or the depth of a specific discipline, the reference value of composite world rankings is quite limited — consulting the corresponding subject rankings and looking at specific on-the-ground institutional metrics is almost always more meaningful than comparing positions on a general league table.

---

Further reading: A five-year deep dive into two rankings A Deep Look at Five Years of QS and U.S. News Rankings · An overview of all four major world rankings World Composite Rankings · Analysis at the subject level Subject Rankings