Martin Ince dissects the data to identify centres of science-based excellence
Cambridge University is regarded as the best for science by academic observers. Its closest UK rival, Oxford University, is in second place, but falls well short in terms of its weighted score.
Behind these two, the big US institutions make an appearance with six entries, while Tokyo University and Imperial College London complete the top ten.
But the table's citations column tells a different story. The figures cover highly cited papers in whole-organism biology, microbiology, agriculture, environment and ecology, mathematics, materials, chemistry, physics, space science and the geosciences. To do well in these tables, institutions had to be strong in several of these areas.
On the basis of the impact of research published in the world's leading science universities, US institutions hold sway, occupying the first 16 positions. ETH Zurich appears at 17, with the UK's best, Durham University, Cambridge and Oxford, in at 18, 19 and 22.
The top citations-impact institution, Harvard University, garners more than 20 references per paper compared with 12.7 for Cambridge.
The US also wins another numbers game, by having 24 of the top 100 universities. But 26 nations are represented in this group, a substantial number when one recalls that the top 200 in our World University Rankings for all subjects involved only 29 countries. It shows that excellent science is widespread and cannot be defined on the basis of narrow measures such as whether a university employs Nobel laureates.
This table also confirms the World University Rankings findings about the excellence of specialist institutions. While the top four - Cambridge, Oxford, Harvard and Berkeley - are all involved in the full range of academic subjects, the top 20 also includes the Massachusetts Institute of Technology, the California Institute of Technology, Imperial College London, ETH Zurich and Paris VI University, all of which are science centres of some renown. The same players perform well in engineering and information technology.
A comparison between peer-review results and the citations impact of the universities listed shows that several highly rated institutions are not generating enough publications that appear in the Thomson Scientific Essential Science Indicators database.
This is especially true of the five Chinese universities listed. Part of the reason may be that much of the welter of research listed as coming from the Chinese Academy of Sciences is performed at these universities. World awareness of their excellence may also be running ahead of their international publishing record.
The analysis of highly cited non-university engineering and information technology institutions is not feasible for science. Only one organisation from outside higher education appears in the elite, the Max Planck Society, which has 12.5 citations for every paper it publishes in the sciences outside biomedicine.