ARWU vs QS: Why Shanghai and London Disagree About the Same Universities

Two different evidence models

The site’s full treatment of ARWU’s six indicators is at ARWU methodology deep dive. ARWU’s 2026 score is built almost entirely from attributed academic achievement: Nobel and Fields winners, Clarivate Highly Cited Researchers, and publications in journals and citation indexes. Its per-capita measure is also derived from academic performance. QS constructs its flagship World University Rankings from a broader institutional picture. Research and Discovery receives 50%, so research remains central, but QS also assigns 20% to Employability and Outcomes, 10% to Learning Experience, 15% to Global Engagement and 5% to Sustainability.

ARWU has no indicator for academic or employer reputation, employability or graduate outcomes, teaching or the learning environment, internationalisation or sustainability. QS includes signals in those families, although Learning Experience is represented by Faculty Student Ratio rather than a direct measure of classroom quality. The reverse absence is equally important: the QS flagship table has no direct Nobel or Fields affiliation indicator, no Highly Cited Researchers count, no Nature and Science article-count indicator and no total SCIE/SSCI paper-count indicator. Its citation indicator measures something different.

Awards are attributed, dated and divided

Awards have direct and time-sensitive purchase within ARWU. Nobel and Fields winners among alumni account for 10% of the score, while winning staff account for 20%. These weights make prize affiliation a material part of the result, but they place it within a model still dominated by other research and output measures.

The alumni indicator uses the date on which a person obtained a bachelor’s, master’s or doctoral degree. Degrees obtained after 2011 receive a weight of 100%; degrees from 2001–2010 receive 90%; degrees from 1991–2000 receive 80%, with the weights continuing down to 10% for degrees from 1921–1930. An older degree connection still counts, but it contributes less within the alumni component. If a person has several degrees from the same institution, that institution is considered only once.

The staff award rule uses the period in which the prize was won and covers Nobel Prizes in Physics, Chemistry, Medicine and Economics and the Fields Medal in Mathematics. Winners after 2021 receive 100%; winners in 2011–2020 receive 90%; those in 2001–2010 receive 80%; those in 1991–2000 receive 70%, continuing down to 10% for 1931–1940. Staff status is determined at the time of the award. A recent staff winner therefore has greater direct weight than an earlier winner within the 20% staff component.

Affiliation is not duplicated across institutions. When a winner is affiliated with more than one institution, each receives the reciprocal of the number of affiliations rather than the full award value. When a Nobel Prize is shared, the winners’ weights also reflect their proportions of the prize. These rules make the award record more precise: they identify attributable value rather than treating every associated institution as if it had received the complete prize.

This is where ARWU and QS diverge most sharply. ARWU directly records prize-winning affiliations and discounts them by age. QS has no parallel prize-count indicator in the flagship table, so a major award does not enter that score as a directly counted affiliation.

Counts scale with size; per-capita performance only partly corrects it

ARWU’s other major components are also quantity-based. HiCi uses Clarivate’s December 2025 Highly Cited Researchers list, considers primary affiliations and counts recognition in multiple ESI fields separately. N&S counts Nature and Science articles from 2021–2025. Authorship position modifies the credit: corresponding-author affiliation receives 100%, first-author affiliation 50%, the next author affiliation 25% and other affiliations 10%. Only articles of the specified type qualify.

PUB counts SCIE- and SSCI-indexed articles from 2025 and gives social-science-indexed papers a double weight. These rules alter how publication credit is assigned, but they do not convert the indicators into measures of output per member of staff. Raw counts scale with institutional size: an institution with more eligible researchers, outputs or author affiliations has more opportunities to accumulate records.

Within each indicator, the highest-scoring institution receives a score of 100 and other institutions receive a percentage of that score. ShanghaiRanking also examines indicator distributions for distorting effects and may use statistical adjustments. Neither process removes the underlying distinction between an absolute count and a per-capita measure.

PCP is ARWU’s explicit size correction. It divides the weighted scores of the other five indicators by the number of full-time-equivalent academic staff and carries 10% of the total score. This allows a smaller institution with concentrated research and award records to score strongly relative to its staff even when its absolute counts are lower. It does not divide the original award and publication scores by staff in the rest of the table, so institutional scale continues to shape those components.

The denominator can itself be estimated. If academic-staff data cannot be obtained for a country, ARWU uses the average number of academic staff among the world’s top 1000 universities for every institution in that country. PCP then uses a country-level proxy rather than each institution’s observed full-time-equivalent staff count. In such cases, the ratio should be read as an approximation based on a shared denominator.

QS shares the broad design idea of normalising research by staffing, but applies it more narrowly. Citations per Faculty receives 20% and divides citation counts by the number of faculty members. That corrects research volume for faculty size; ARWU’s PCP divides a combined academic-performance score by academic staff.

The named humanities and social sciences exception

ShanghaiRanking explicitly names London School of Economics when explaining how it handles humanities- and social-science-specialised institutions. For these institutions, the N&S indicator is not considered. Its 20% weight is not simply left unallocated: the methodology says it is relocated to other indicators. The exception therefore changes the discipline-sensitive construction of the score rather than merely recording a zero for N&S.

This does not remove the institution from publication-based assessment. PUB still includes SSCI-indexed articles and gives them a double weight. The practical consequence is narrower than either total acceptance or total exclusion: the N&S component is removed, the weight is redistributed, and other publication indicators still shape the result.

The QS flagship weight table has no corresponding N&S article-count component that can be removed in the same way. That structural difference matters when comparing universities with different research and publication profiles.

Reputation records perceived standing

The survey mechanics behind that signal are covered at reputation survey methodology critique. Academic Reputation receives 30% in QS. QS describes it as a survey in academic experts nominate universities in their areas of expertise, evaluating nominations for approximately 7000 institutions each year. Employer Reputation receives 15% and measures institutional and programme reputation among employers. Together, the reputation indicators account for 45% of the QS score, so perceived standing has more direct influence than any single survey component has in ARWU’s model.

A reputation survey captures a different object from a prize count. It can register accumulated judgements about an established academic or employer-facing identity. Because the signal comes from respondents’ perceptions rather than from a specified prize total, an established name can receive credit for standing that ARWU’s award indicators do not directly measure. The survey does not itself establish why respondents hold that view.

The reverse limitation is equally clear. An ARWU award score records prize type, affiliation and period under explicit rules. It cannot reproduce the collective academic and employer judgements collected by QS. Survey standing may matter without a new prize, while a prize may be directly counted without implying anything about the full range of perceptions held by surveyed groups.

QS also separates reputation from other outcomes and conditions. Employment Outcomes receives 5% and concerns employability and graduates’ subsequent impact. Faculty Student Ratio receives 10% under Learning Experience, making a staffing relationship visible rather than claiming to measure classroom quality directly. International Faculty Ratio, International Research Network and International Student Ratio each receive 5% under Global Engagement; International Student Diversity is listed at 0%, so it adds no direct weight to the flagship total. Sustainability receives 5%. These choices allow an institution to be distinguished by signals for which ARWU has no indicator at all.

Match the table to the question

For a reader asking about accumulated elite recognition and absolute research production, ARWU is the relevant table. Its position needs to be read through prize affiliation, award age, authorship credit and institutional scale. A strong alumni or staff record affects a defined component, but recent work is not treated the same as work from earlier periods, and multiple affiliations do not each receive the full prize value.

For a reader asking how academics and employers perceive programmes, and how an institution performs across employability, staffing, global engagement and sustainability, QS is the relevant table. Its result is driven substantially by survey-based reputation and by institutional signals that ARWU does not measure.

For a reader asking about research intensity relative to staffing, both tables contain relevant evidence with different denominators. ARWU’s PCP covers the combined weighted scores of its other indicators. QS’s Citations per Faculty covers citation activity. Neither measure makes the entire ranking a per-staff model.

For a reader comparing humanities- and social-science-specialised institutions, ARWU’s named exception must remain visible. The absence of N&S is paired with a redistribution of weight, while SSCI-indexed output remains part of PUB.

Different questions, not conflicting answers

The two tables are not in conflict about a single measurable form of university quality. ARWU asks what prize, researcher and publication records can be attributed to an institution under research-focused rules. QS asks how institutions are judged across academic reputation, employer views, outcomes, learning experience, global engagement and sustainability.

A difference in order is therefore information about construction, not a contradiction. The useful reading is to identify which signal produced the position and whether that signal answers the question actually being asked.