Although the California Institute of Technology leads the World University Rankings for the fourth year running, it fails to take top spot in any of our six subject-specific tables: five other research-intensives are the front-runners here, with Stanford University the only institution to lead more than one subject area (arts and humanities and social sciences).

Harvard University is at the head of the life sciences top 50; the Massachusetts Institute of Technology is primus inter pares for engineering and technology; the University of Oxford ¨C the only non-US institution to lead any of the tables ¨C tops the charts for clinical, pre-clinical and health-related subjects; and Princeton University steals Caltech¡¯s crown in the physical sciences.

This is Princeton¡¯s first time at the top of any subject table since 2011-12 (when it shared first place in the physical sciences with Caltech).

The university was and remains a pioneer in chemistry: it hosted the US¡¯ first undergraduate chemistry laboratory, founded by John Maclean in 1795, and is now home to the state-of-the-art Frick Chemistry Laboratory, the largest academic building on Princeton¡¯s campus today.

ÖйúAƬ

Its record in physics is similarly distinguished: although he was never a member of faculty, Albert Einstein moved to Princeton in 1933 and lived there until his death in 1955, enjoying close links with the university, whose physics department has produced more than a dozen Nobel laureates.

The subject-specific tables are, as is the case with the overall rankings, dominated by the US and the UK. The highest-ranked Asian institution is the National University of Singapore, which takes 13th place in engineering and technology.

ÖйúAƬ

³§·É¾±³Ù³ú±ð°ù±ô²¹²Ô»å¡¯²õ ETH Z¨¹rich-Swiss Federal Institute of Technology Zurich makes the top 10 in the same category (the best standing achieved outside the Anglo-American hegemon).

Different weights and measures
The subject tables employ the same range of 13 performance indicators used in the overall World University Rankings, brought together with scores provided under five categories:

Teaching: the learning environment
Research: volume, income and reputation
Citations: research influence
International outlook: staff, students and research
Industry income: innovation.

Here, the overall methodology is carefully recalibrated for each subject, with the weightings changed to suit the individual fields. In particular, those given to the research indicators have been altered to fit more closely the research culture in each subject, reflecting different publication habits: in the arts and humanities, for instance, where the range of outputs extends well beyond peer-reviewed journals, we give less weight to paper citations.

Accordingly, the weight given to ¡°citations: research influence¡± is halved from 30?per cent in the overall rankings to just 15?per cent for the arts and humanities.

More weight is given to other research indicators, including the academic reputation survey.

ÖйúAƬ

For social sciences, where there is also less faith in the strength of citations alone as an indicator of research excellence, the measure¡¯s weighting is reduced to 25?per cent.

By the same token, in those subjects where the vast majority of research outputs come through journal articles and where there are high levels of confidence in the strength of citations data, we have increased the weighting given to the research influence (up to 35?per cent for the physical and life sciences and for the clinical, pre-clinical and health tables).

ÖйúAƬ

A breakdown of the methodology for each subject is provided at the foot of the tables.

Criteria
No institution can be included in the overall World University Rankings unless it?has published a minimum of 200 research papers a year over the five years we examine.

But for the six subject tables, the threshold drops to 100 papers a year for subjects that generate a high volume of publications and 50 a year in subjects such as social sciences where the volume tends to be lower. Although we apply some editorial discretion, we generally expect an institution to have at least 10?per cent of its staff working in the relevant discipline in order to include it in the subject table.

The majority of institutions in Thomson Reuters¡¯ Global Institutional Profiles database provide detailed subject-level information.

In rare cases where such data are not supplied, institutions are either excluded or public sources are used to underpin estimates.

is editor, Times ÖйúAƬ Rankings.

ÖйúAƬ