How Does a University Get Ranked? The Data Submission Path From Institution to Score

From institutional sign-off to published position: who sets ranking indicators, how data is normalised and weighted, and how corrections are published.

A ranking score is the end of a publisher-defined production chain, not a number a university submits in finished form. The institution supplies observations; the ranking publisher determines what those observations measure, how they are adjusted, how much each component contributes and how the resulting values are ordered. That division of labour is why a raw value cannot be assumed to produce the same score in every ranking. The path runs from the rulebook to provenance, formal validation, normalisation, weighting, publication and correction.

The publisher, not the university, writes the rulebook

For Times Higher Education (THE), the World University Rankings 2026 methodology states that it uses “18 carefully calibrated performance indicators.” The methodology presents five pillars: Teaching, Research environment, Research quality, International outlook and Industry. THE sets the indicator set and the pillar allocation; an institution supplies the data that fit those categories. The university is therefore a data participant, not the author of the ranking formula.

QS uses a different hierarchy. Its methodology defines a lens as a group of indicators related to the same theme, an indicator as a measurement of one aspect of institutional performance, and a metric as a more specific calculation within an indicator. QS publishes category and metric weights and says that weightings are reviewed annually. ARWU uses an indicator-weighted model instead: Alumni, Award, HiCi, N&S, PUB and PCP each have a published weight, and each indicator is converted to a top-relative score. These are different publisher rules, not three names for a common score.

The input has provenance

Data entering a ranking must be separated by provenance. Institutional self-report means values supplied by the university through the ranking’s submission process. It is distinct from a bibliometric record or a survey response. THE’s participation page lists “Data submission (required)” and “All data submitted and validated” for the World University Rankings. That identifies the institutional stage, but a submitted value is still an input rather than a ranking score.

Third-party bibliometric data enters through an external supplier. THE says its bibliometric data supplier Elsevier provided 174.9 million citations to 18.7 million journal articles. The data include more than 28,700 active peer-reviewed journals indexed by Elsevier’s Scopus database and all indexed publications between 2020 and 2024. Those figures are third-party bibliometric inputs, not institutional self-report. ARWU’s HiCi input is separately identified: the Highly Cited Researchers list issued in November 2024 was used to calculate its HiCi indicator.

Reputation surveys form a third category. THE’s most recent Academic Reputation Survey underpinning its category was carried out between November 2024 and January 2025. The 2025 data were combined with the results of the 2024 survey, giving more than 108,000 responses. QS says its Academic Survey collects and distils the collective intelligence of academics from around the world while evaluating nominations for approximately 7,000 institutions each year; its framework also includes Employer Reputation. Survey responses are neither institutional self-report nor Scopus bibliometrics, and each publisher processes them under its own methodology.

Submission is a formal gate

THE makes the institutional stage formal. Its corrections policy states: “Every university in the rankings formally signs off its institutional data as correct.” Its participation page says the portal will be open from January to 30 March. The same page makes data submission a requirement and lists all data submitted and validated as a ranking criterion.

The signature means that the institution formally attests that its submitted institutional data are correct. It is not described as an independent audit, nor as a guarantee that every underlying item has been separately proved by THE. The policy places sign-off alongside its use of highly trusted third-party data providers where necessary and its extensive quality-control processes. Nor is the signature a final score or a bar to later correction. THE says all concerns about data accuracy or calculation errors are investigated by its data team. The institution acknowledges its submitted data at the submission stage; the publisher still controls the transformation, weighting and any later response.

Normalisation changes the unit of comparison

A submitted value is not automatically the value scored. THE gives specific examples of how raw data are made comparable. Its doctoral-awards indicator is normalised to account for a university’s unique subject mix, reflecting that doctoral volume varies by discipline. Research income is fully normalised to take account of each university’s distinct subject profile. Institutional income is scaled against academic staff numbers and normalised for purchasing-power parity, or PPP.

Reputation data undergo a related adjustment. Where disciplines or countries were over- or under-represented in the survey, THE’s data team weighted the responses to fully reflect the global distribution of scholars. This is a form of discipline and country weighting applied to reputation responses, rather than a simple count of every response without regard to representation.

One current detail makes the distinction between collection and scoring especially clear: one THE metric, study abroad, currently has zero weight, but THE intends to include it in the future, subject to review and consultation. A metric can therefore be part of the data system without contributing to the current composite.

These adjustments are the reason a single raw number cannot be carried unchanged from one ranking to another. Each publisher applies its own definitions, transformations, reference distributions and weights before producing a total. Normalisation is part of score construction, not a cosmetic label added after the score has been calculated.

Adjusted values become a weighted composite

For THE’s World University Rankings 2026, the pillar contributions include Teaching at 29.5%, Research environment at 29% and Research quality at 30%. The methodology also names International outlook and Industry as pillars. THE’s internal Teaching breakdown is Teaching reputation 15%, Staff-to-student ratio 4.5%, Doctorate-to-bachelor’s ratio 2%, Doctorates-awarded-to-academic-staff ratio 5.5% and Institutional income 2.5%. These are the component contributions that make up the overall Teaching allocation, rather than weights independently chosen by the institution. Survey, staffing, doctoral and income measures therefore reach the composite through publisher-defined transformations and contributions.

QS builds its composite through the lens–indicator–metric hierarchy. Research and Discovery carries 50%, divided into Academic Reputation at 30% and Citations per Faculty at 20%. Employability and Outcomes carries 20%, divided into Employer Reputation at 15% and Employment Outcomes at 5%. Learning Experience carries 10%, represented by Faculty Student Ratio at 10%. Global Engagement carries 15%, divided into International Faculty Ratio at 5%, International Research Network at 5%, International Student Diversity at 0% and International Student Ratio at 5%. Sustainability carries 5%. The hierarchy, the metric definitions and the annual review of weightings are all part of the score construction.

ARWU uses a top-relative, percentile-style scale. Its indicator weights are Alumni 10%, Award 20%, HiCi 20%, N&S 20%, PUB 20% and PCP 10%. For each indicator, the highest scoring institution is assigned a score of 100, and other institutions are calculated as a percentage of the top score. ARWU also examines the distribution of each indicator for any significant distorting effect and uses standard statistical techniques to adjust the indicator if necessary.

Its construction rules also affect the values that enter the composite. If a person obtains more than one degree from an institution, the institution is considered once only. If a winner is affiliated with more than one institution, each institution is assigned the reciprocal of the number of institutions. If the number of academic staff for institutions of a country cannot be obtained, the average number of academic staff for the world’s top 1,000 universities is used for PCP. More than 2,500 universities are actually ranked, with the best 1,000 published. In each system, the final position follows a publisher’s rules for combining adjusted inputs rather than displaying one submitted number on its own.

Publication is not the end

THE says all concerns raised about data accuracy or calculation errors are investigated by its data team. For an error made by THE, the policy says it will update the relevant online university ranking and any related rankings, and publish a footnote to that effect.

A third-party data problem can require a broader response. Where many, most or all universities are affected, such as when a set of input data from a third-party supplier is found to contain systemic issues, THE may recalculate the entire ranking and republish it with a footnote. The important distinction is between a correction to one calculation and a change affecting the ranking as a whole.

THE also specifies how tied positions are displayed. Its example says that if an institution is ranked 30th and should be ranked 25th, the corrected sequence reads: “24th, 25th, 25th, 26th, 27th, 28th, 29th, 31st.” A correction can therefore alter the positions around a tie rather than simply replace one label in isolation.

For data manipulation, THE reserves the right to exclude an institution from future rankings, remove it from historical rankings and publish a note to that effect. These sanctions concern the institution’s presence in the ranking record, while a calculation correction changes a published result. In either case, a published ranking is a revisable output of the production chain.

How to verify one link at a time

A reader can verify the chain by starting with the exact ranking edition and its publisher. For THE WUR 2026, check the indicator count, five-pillar structure, survey description and bibliometric source description. For QS, check the definitions of lens, indicator and metric, followed by the published category and metric weights. For ARWU 2025, check the indicator weights, the rule that the top indicator score is 100, the distribution-adjustment rule and the treatment of affiliations and missing staff numbers.

Next trace provenance. A THE institutional value should be identified as self-reported and linked to the required submission and validation language. A bibliometric value should be identified as Elsevier and Scopus supply. A reputation value should be identified as an Academic Reputation Survey or QS Academic Survey input. THE’s stated survey period, response-combination rule and weighting for over- or under-represented disciplines and countries allow the survey-processing claim to be checked against the methodology.

Then check transformation and weighting in the publisher’s order: subject mix, PPP and survey-response weighting for the relevant THE measures; the top-score and distribution treatment for ARWU; and the exact weight assigned to each metric. Finally, compare the published position with THE’s ranking page and corrections policy. A footnote records a published correction, while the policy explains when a systemic third-party issue may lead to full recalculation. Concerns about accuracy or calculation can be sent to profilerankings@timeshighereducation.com. This route lets a reader check the rule, source, adjustment, weight and final position separately instead of treating a ranking label as a self-explanatory fact.