The Power of Z

This year QS has introduced several "methodological enhancements" into the THES-QS rankings. One is the use of Z-scores. Basically, this means that the mean for all universities is deducted from the raw score and the result is then divided by the standard deviation. In effect, the score represents not an absolute number but how far each university is from the average. One consequence of using Z-scores is that differences at the very top are reduced.

In principle this is not a bad idea and other rankers do it but it has produced some odd results in this case.

In the survey of academic opinion, for example, the following universities all get a maximum score of 100: Harvard, Cambridge, Oxford, Yale, Caltech, MIT, Columbia, McGill, Australian National University, Stanford, Cornell, Berkeley, Melbourne, British Columbia, National University of Singapore, Peking and Toronto.

Do THES and QS really expect us to believe that Melbourne, British Columbia and Peking are just as good at research as Harvard ? Especially since Harvard is far ahead on every one of the subject rankings?

THES has a headline about fine tuning revealing distinctions. Really?