Honeypots of the world

One of the markers of a world-class university is its international profile. Indeed, the two are virtually synonymous. A world-class university is one that has such an outstanding knowledge and learning environment that it is the place to which you would want to go to study as a student, to begin your research as a postgraduate and to work as a researcher.

Thomson Reuters has carried out several pieces of work confirming that the identification of an institution as an intellectual honeypot is one of the key criteria that people use to judge excellence across all disciplines. In 2006, we reported to the UK's Economic and Social Research Council on the sort of indicators that the social science community associates with top-rank departments. "Internationally attractive" was one of the most cited indices.

The problem is that "internationally attractive" is difficult to quantify. The expert in the field can quickly quote the top 10 places they would love to work, but they aren't so obvious to the less informed. So what can we use as a proxy indicator when calculating the Times Higher Education World University Rankings?

For a start, we can look at the flow of visitors that an institution receives, especially from overseas. The places that get endless requests from people hoping to study alongside their research teams or who want to work with leading experts in their field for the summer are clearly leading the pack. But how do we evaluate "visitors"? Should we judge by length of stay, their eminence, the distance travelled or the funds required?

International students and their number as a share of an institution's taught population are a bit simpler to define. That said, some locations are clearly more attractive and more of a natural pool than others, so we have to be careful about geographical, linguistic and political factors that can skew this measure.

Collaboration is more robust. Scholars collaborate in order to get access to scarce resources - facilities, know-how or ideas. When they collaborate they normally share credit and, on the whole, academics are reticent to do so without reason. That shared credit shows up as co-authorship and we have excellent databases with all the authors' names - and their addresses.

Collaboration as a proportion of output is growing. The amount of work carrying the names and addresses of multiple universities is increasing. In most countries, international collaboration is growing too, and the coalition government has asked for this to be a key indicator in our annual report to the Department for Business, Innovation and Skills on the UK research base's international performance. This makes it clear that a lot of institutions want to collaborate with partners in the US and Western Europe, with the most diverse foreign links being forged by universities that also perform highly according to other criteria.

The value of collaboration data is that they work well as indicators. Our publication database is comprehensive geographically, temporally and in terms of subject. Research collaboration requires real resources, as even a joint paper by correspondence requires time and the sharing of ideas. For these reasons, the presence of publication links is self-validating.

Collaboration is also associated with excellent outcomes. Internationally collaborative papers tend to be relatively frequently cited. This may be partly because of their exposure in more than one country, but it is also because people collaborate in order to achieve goals they cannot meet alone. The US is a big place and Harvard University is a great institution, but its papers with UK partners are cited more frequently than those it produces when it goes it alone.

All this means that we can readily build up a picture of who is working with whom, how this varies by region and what the trends look like. Additionally, we can associate this reasonably well with main subject areas.

Of course, there are caveats. The measure certainly won't cover the subjects that rely more heavily on books, let alone non-print media, for research outputs. But it works pretty well, and we propose to use it as a new indicator for the THE World University Rankings in time for this year.