Friday, May 29, 2015

University Ranking Challenge: Your starter for 5,154

Phil Baty, editor of the Times Higher Education World University Rankings, has indicated that the publication of a paper from the ATLAS and CMS experiments at the CERN Large Hadron Collider project is a challenge for rankers.

The paper in question has a total of 5,154 authors, if that is the right word, with sole or primary affiliation to 344 institutions. Of those authors 104 have a secondary affiliation. One is deceased. Under THE's current methodology every institution contributing to the paper will get credit for all the citations that the paper will receive, which is very likely to run into the thousands.

For the elite universities  participating in these projects a few thousand citations will make little or no difference. But for for a small specialised institution or a large one that does little research, those citations spread out over a few hundred papers could make a big difference.

In last year's rankings places like Florida Institute of Technology, Universite Marrakesh Cadi Ayyad, Morroco, Federico Santa Maria Technical University, Chile, Bogazici University, Turkey, got implausibly high scores for citations that were were well ahead of those for the other criteria.

The paper in question does set a record for the number of contributors although the challenge is not particularly new.

At a seminar in Moscow earlier this year, Baty suggested that THE, now independent of Thomson Reuters, was considering using fractionated counting, dividing all the citations among the contributing institutions.

This would be an excellent idea and should be technically quite feasible since CWTS at Leiden University use it as their default option.

But there would be a a price to pay. The current methodology allows THE to boast that it has found a way of uncovering hitherto unnoticed pockets of excellence. It is also a selling point in THE's imperial designs of expanding into regions where there has so far been little interest in ranking, Russia, the Middle East, Africa, the BRICS. A few universities in those regions could make a splash in the rankings if they recruited, even as an adjunct, a researcher working on the LHC project.

It would be most welcome if THE does start using fractionated counting in its citation indication. Also welcome would be some other changes: not counting self-citation, reducing the weighting for the indicator, including several different methods of evaluating research impact or quality, and, especially important, getting rid of the "regional modification" that awards a bonus for being located in a low scoring country.

No comments: