Archive for rankings

Why Universities should ignore League Tables

Posted in Bad Statistics, Education with tags , , , , , on January 12, 2017 by telescoper

Very busy day today but I couldn’t resist a quick post to draw attention to a new report by an independent think tank called the Higher Education Policy Institute  (PDF available here; high-level summary there). It says a lot of things that I’ve discussed on this blog already and I agree strongly with most of the conclusions. The report is focused on the international league tables, but much of what it says (in terms of methodological criticism) also applies to the national tables. Unfortunately, I doubt if this will make much difference to the behaviour of the bean-counters who have now taken control of higher education, for whom strategies intended to ‘game’ position in these, largely bogus, tables seem to be the main focus of their policy rather than the pursuit of teaching and scholarship, which is what should universities actually be for.

Here is the introduction to high-level summary:

Rankings of global universities, such as the THE World University Rankings, the QS World University Rankings and the Academic Ranking of World Universities claim to identify the ‘best’ universities in the world and then list them in rank order. They are enormously influential, as universities and even governments alter their policies to improve their position.

The new research shows the league tables are based almost exclusively on research-related criteria and the data they use are unreliable and sometimes worse. As a result, it is unwise and undesirable to give the league tables so much weight.

Later on we find some recommendations:

The report considers the inputs for the various international league tables and discusses their overall weaknesses before considering some improvements that could be made. These include:

  • ranking bodies should audit and validate data provided by universities;
  • league table criteria should move beyond research-related measures;
  • surveys of reputation should be dropped, given their methodological flaws;
  • league table results should be published in more complex ways than simple numerical rankings; and
  • universities and governments should not exaggerate the importance of rankings when determining priorities.

No doubt the purveyors of these ranking – I’ll refrain from calling them “rankers” – will mount a spirited defence of their business, but I agree with the view expressed in this report that as they stand these league tables are at best meaningless and at worst damaging.

Advertisements

An Open Letter to the Times Higher World University Rankers

Posted in Education, The Universe and Stuff with tags , , , , , , , , on October 5, 2015 by telescoper

Dear Rankers,

Having perused your latest set of league tables along with the published methodology, a couple of things puzzle me.

First, I note that you have made significant changes to your methodology for combining metrics this year. How, then, can you justify making statements such as

US continues to lose its grip as institutions in Europe up their game

when it appears that any changes could well be explained not by changes in performance, as gauged by the metrics you use,  but in the way they are combined?

I assume, as intelligent and responsible people, that you did the obvious test for this effect, i.e. to construct a parallel set of league tables, with this year’s input data but last year’s methodology, which would make it easy to isolate changes in methodology from changes in the performance indicators.  Your failure to publish such a set, to illustrate how seriously your readers should take statements such as that quoted above, must then simply have been an oversight. Had you deliberately witheld evidence of the unreliability of your conclusions you would have left yourselves open to an accusation of gross dishonesty, which I am sure would be unfair.

Happily, however, there is a very easy way to allay the fears of the global university community that the world rankings are being manipulated: all you need to do is publish a set of league tables using the 2014 methodology and the 2015 data. Any difference between this table and the one you published would then simply be an artefact and the new ranking can be ignored. I’m sure you are as anxious as anyone else to prove that the changes this year are not simply artificially-induced “churn”, and I look forward to seeing the results of this straightforward calculation published in the Times Higher as soon as possible.

Second, I notice that one of the changes to your methodology is explained thus

This year we have removed the very small number of papers (649) with more than 1,000 authors from the citations indicator.

You are presumably aware that this primarily affects papers relating to experimental particle physics, which is mostly conducted through large international collaborations (chiefly, but not exclusively, based at CERN). This change at a stroke renders such fundamental scientific breakthroughs as the discovery of the Higgs Boson completely worthless. This is a strange thing to do because this is exactly the type of research that inspires  prospective students to study physics, as well as being direct measures in themselves of the global standing of a University.

My current institution, the University of Sussex, is heavily involved in experiments at CERN. For example, Dr Iacopo Vivarelli has just been appointed coordinator of all supersymmetry searches using the ATLAS experiment on the Large Hadron Collider. This involvement demonstrates the international standing of our excellent Experimental Particle Physics group, but if evidence of supersymmetry is found at the LHC your methodology will simply ignore it. A similar fate will also befall any experiment that requires large international collaborations: searches for dark matter, dark energy, and gravitational waves to name but three, all exciting and inspiring scientific adventures that you regard as unworthy of any recognition at all but which draw students in large numbers into participating departments.

Your decision to downgrade collaborative research to zero is not only strange but also extremely dangerous, for it tells university managers that participating in world-leading collaborative research will jeopardise their rankings. How can you justify such a deliberate and premeditated attack on collaborative science? Surely it is exactly the sort of thing you should be rewarding? Physics departments not participating in such research are the ones that should be downgraded!

Your answer might be that excluding “superpapers” only damages the rankings of smaller universities because might owe a larger fraction of their total citation count to collaborative work. Well, so what if this is true? It’s not a reason for excluding them. Perhaps small universities are better anyway, especially when they emphasize small group teaching and provide opportunities for students to engage in learning that’s led by cutting-edge research. Or perhaps you have decided otherwise and have changed your methodology to confirm your prejudice…

I look forward to seeing your answers to the above questions through the comments box or elsewhere – though you have ignored my several attempts to raise these questions via social media. I also look forward to seeing you correct your error of omission by demonstrating – by the means described above – what  changes in league table positions are by your design rather than any change in performance. If it turns out that the former is the case, as I think it will, at least your own journal provides you with a platform from which you can apologize to the global academic community for wasting their time.

Yours sincerely,

Telescoper

IQ in different academic fields – Interesting? Quite!

Posted in Bad Statistics with tags , , , on May 26, 2013 by telescoper

You all know how much I detest league tables, especially those that are based on entirely arbitrary criteria but nevertheless promote a feeling of smug self-satisfaction for those who lucky enough to find themselves at the top. So when my attention was drawn to a blog post that shows (or purports to show) the variation of average IQ across different academic disciplines I decided to post the corresponding ranking with the usual health warning that IQ tests only measure a subject’s ability to do IQ tests. This isn’t even based on IQ test results per se, but on a conversion between the Graduate Record Examination (GRE) results and IQ which may be questionable. Moreover, the differences are really rather small and (as usual) no estimate of sampling uncertainty is provided.

Does this list mean that physicists are smarter than anyone else? You might say that. I couldn’t possibly comment…

  • 130.0 Physics
  • 129.0 Mathematics
  • 128.5 Computer Science
  • 128.0 Economics
  • 127.5 Chemical engineering
  • 127.0 Material science
  • 126.0 Electrical engineering
  • 125.5 Mechanical engineering
  • 125.0 Philosophy
  • 124.0 Chemistry
  • 123.0 Earth sciences
  • 122.0 Industrial engineering
  • 122.0 Civil engineering
  • 121.5 Biology
  • 120.1 English/literature
  • 120.0 Religion/theology
  • 119.8 Political science
  • 119.7 History
  • 118.0 Art history
  • 117.7 Anthropology/archeology
  • 116.5 Architecture
  • 116.0 Business
  • 115.0 Sociology
  • 114.0 Psychology
  • 114.0 Medicine
  • 112.0 Communication
  • 109.0 Education
  • 106.0 Public administration