Monday, August 29, 2011

Japanese Universities Send a Strong Request

A very interesting document from the top 11 Japanese research universities has appeared. They are unhappy with the citations indicator in last year's Times Higher Education -- Thomson Reuters World University Rankings.

"The purpose of analyzing academic research data, particularly publication and citation trends is to provide diverse objective information on universities and other academic institutions that can be used by researchers and institutions for various evaluations and the setting of objectives. The 2010 Thomson Reuters / THE World University Rankings, however, do not give sufficient consideration to the unique characteristics of universities in different countries or the differing research needs and demands from society based on country, culture and academic field. As a result, those rankings are likely to lead to an unbalanced misleading and misuse of the citation index.

RU11 strongly requests, therefore, that Thomson Reuters / THE endeavors to contribute to academic society by providing objective and impartial data, rather than imposing a simplistic and trivialized form of university assessment."

It is a tactical mistake to go on about uniqueness. This is an excuse that has been used too often by institutions whose flaws have been revealed by international rankings.

Still, they do have a point. They go on to show that when the position of Asian universities according to the citations indicator in the THE-TR rankings is compared with the citations per paper indicator in the 2010 QS Asian university rankings, citations per paper over an 11 year period from TR's Essential Science Indicators and citations per paper/citations per faculty in the 2010 QS World university rankings (I assume they mean citations per faculty here since the QS World University Rankings do not have a citations per paper indicator) leading Japanese universities do badly while Chinese, Korean and other Asian universities do very well.

They complain that the THE--TR rankings emphasise "home run papers" and research that produces immediate results and that regional modification (normalisation) discriminates against Japanese universities.

This no doubt is a large part of the story but I suspect that the distortions of the 2010 THE--TR indicator are also because differences in the practice of self citation and intra--university citation, because TR's methodology actually favors those who publish relatively few papers and because of its bias towards low--cited disciplines.

The document continues:

"1. The ranking of citations based on either citations per author (or faculty) or citations per paper represent two fundamentally different ways of thinking with regards to academic institutions: are the institutions to be viewed as an aggregation of their researchers, or as an aggregation of the papers they have produced?  We believe that the correct approach is to base the citations ranking on citations per faculty as has been the practice in the past.

2. We request a revision of the method used for regional modification. 

3. We request the disclosure of the raw numerical data used to calculate the citation impact score for the various research fields at each university."

I suspect that TR and THE would reply that their methodology identifies pockets of excellence (which for some reason cannot be found anywhere in the Japanese RU 11), that the RU 11 are just poor losers and that they are right and QS is wrong.

This question might be resolved by looking at other measures of citations such as those produced by HEEACT, Scimago and ARWU.

It could be that this complaint if was sent to TR was the reason for TR and THE announcing that they were changing the regional weighting process this year. If that turns out to be the case and TR is perceived as changing its methodology to suit powerful vested interests then we can expect many academic eyebrows to be raised.

If the RU 11 are still unhappy then THE and TR might see a repeat of the demise of the Asiaweek rankings brought on in part because of a mass abstention by Japanese and other universities.

Saturday, August 27, 2011

The THE Citations Indicator

The Research Impact indicator in last year's Times Higher Education - Thomson Reuters World University Rankings led to much condemnation and not a little derision. Alexandria University was fourth in the world for research impact, with Bilkent, Turkey, Hong Kong Baptist University and several other relatively obscure institutions achieving remarkably high scores.

The villain here was Thomson Reuters' field and year normalisation system by which citations were compared with world benchmarks for field and year. This meant that a large number of citations within year of publication  to a paper classified as being in a low cited field could have a disproportionate  effect, which might be further enhanced if the university was in a region where citations were low.

Now THE have announced that this year there will be three changes. These are:

  • raising the threshold for inclusion in the citations indicator from 50 publications per year to 200
  • Extending the period for counting citations from five to six years
  • Changing regional normalisation so that it takes account of subject variations within regions as well as the overall level  of citations.
Here are some things which Thomson Reuters apparently will not do:

  • reducing the weighting given to citations
  • not counting sell-citations, citations within institutions or citations within journals
  • using a variety of indications to assess research impact, such as h-index, total citations, citations per paper
  • using a variety of databases

So, everybody will have to wait until September to see what will happen.

Sunday, August 21, 2011

Value for Money

Recently, the Higher Education Funding Council of England published data indicating the percentage of UK students at English universities with grades AAB at A level. Oxford and Cambridge were at the top with 99% and Wolverhampton, Staffordshire and Hertfordshire at the bottom with 2%.

Now Hertfordshire statisticians have produced graphs comparing performance on four British league tables with tuition fees. Hertfordshire offers best value for tuition money in its band. Oxford, Cambridge, LSE, Derby and London Metropolitan do well in theirs. Liverpool John Moores, East London and Bedfordshire are among the worst.

It should be noted that at the moment the differences between tuition levels are relatively small so this table may not mean very much.

Saturday, August 20, 2011

Perhaps They know Something You Don't

The Pew Research Center has issued a report showing that women are more likely than men to see the value of a college education. Men, says the report, are laggards. The implication is that women are more perceptive than men.

At a time when women surpass men by record numbers in college enrollment and completion, they also have a more positive view than men about the value higher education provides, according to a nationwide Pew Research Center survey. Half of all women who have graduated from a four-year college give the U.S. higher education system excellent or good marks for the value it provides given the money spent by students and their families; only 37% of male graduates agree. In addition, women who have graduated from college are more likely than men to say their education helped them to grow both personally and intellectually.

An article in reviewing the report refers to another study from the Brookings Institute that finds that college is in fact an excellent investment .

Why then, are men apparently so uninformed about the benefits of higher education? The Pew report provides part of the answer when it discloses that men are much more likely than women to pay for college by themselves. A good investment, it seems, is even better when it is paid for by somebody else.

Also, let us compare the career prospects of men and women with average degrees in the humanities or social sciences. Even without affirmative action, men who are bored by diversity training, professional development, all sorts of sensitisation and other rituals of the feminised corporation and bureaucracy are unlike to get very far if anywhere.

And perhaps men are more likely to grow by themselves.