Friday, October 04, 2013

Trust the West to Find the Best

Here's one from the Interdisciplinary Department of Huh?-Who-knew?

The awesome twosome, Times Higher Education (THE) and Thomson Reuters, informed us all that we have had this hidden gem among us all along: Panjab University is the best in Asia in citations [bold emphasis added]:

Ranked in the 226-250 bracket in The Times Higher Education World University Rankings 2013-14 announced in London, Panjab University (PU) is ranked 32 in Asia. But besides the fact that PU is the best among Indian universities - even leaving behind the IITs - it is also the best in Asia when it comes to its research being cited in journals and studies across the world.

PU's score for citation, 84.7 on a scale of 100, is higher than the University of Tokyo, Japan, which has been otherwise ranked 1 in Asia and has a world ranking of 23 as per the study. Tokyo's citation score is 69.8. The citation score was based on the frequency with which research of those from PU was used by other researchers.


  1. Vijay said...

    Dear Abi
    You can be the bestest too.
    Maybe no one else (Tokyo, JNU, Hyd, IISc etc) gave information with the diligence Panjab did :-)??:
    "An elated vice-chancellor (V-C) Arun Kumar Grover told HT, "One has to apply for such a survey, and we had supplied all the required information, based on which we were evaluated on 13 parameters."

  2. Ankur Kulkarni said...

    This looks like an artifact of "small numbers", as opposed to "large numbers" wherein the law of averages would apply. Scopus shows that PU publishes far less than all old IITs. How did its citation metric rise so much?

  3. ahannaasmi said...

    I'll put my money either on a bug in the analysis software [like counting things twice etc.], or on some kind of weighting fiasco [PU, as far as I recall, has/has had a very good group of mathematicians, and their citations might have pulled up the rest of the university.

    In any case, I think most of these rankings are highly unscientific: their results seem to go from end to end over a single year which raises serious questions about the validity and reproducibility of their methodology. All that I could find about the methodology on the THE website was how they calculated the z-score by subtracting the mean from the raw scores and dividing by the sample standard deviation and then computing a percentile (most probably wrt a normal; they didn't say). Their are just a few sentence about the crucial aspect of how the raw scores were computed in the first place, which basically comprise some name-dropping ("Thomson -Reuters"!, "Web of Science"!) and some perfunctory remarks about weighting different fields differently.

  4. Desi Babu said...

    What can I say other than "Balle Balle?!" "

    And, keeping with the age old Punjabi (or Panjabi?) tradition of naming all things great as "Lovely", have you considered renaming your institute to "Lovely Indian Institute of Science"? After all, you do have a lovely campus.

    To say the least, it might get your rankings up...;-)


    P.S: Impact factor now measures how quickly you can ram your tractor into stacked bales of hay.
    In a wheat field, somewhere between Ropar and Patiala...

  5. Richard Holmes said...

    This is explained at University Ranking Watch. Someone at Panjab University is involved in the Large Hadron Collider Project, along with 2000+ other physicists from 100+ countries. Papers from the project are widely and quickly cited, sometimes, one suspects, mainly to pad out the literature review in later papers.

    Because the HRCP publications get more citations and are cited earlier than the norm for the field,the impact score for participating institutions is greatly multiplied by TR's methods.

    Plus there is the "regional modification" which gives PU a big boost for being in a country with low citations (that would effect its position only with non-Indian universities).

    Most of the 100+ other universities received little benefit from the project because the hundreds of citations were spread out over thousands of publications. But PU has such a comparatively small number of publications, that the citations to the LHCP made an enormous difference.

    Next question. Is Tokyo Metropolitan University the best for research influence in the world except for MIT?

  6. ahannaasmi said...

    Thanks for fishing out the reason, Richard Holmes. As I suspected, the reason seems to have everything to do with the methods used to compute the raw scores. Ironically, that is the aspect of the methodology that gets the shortest attention in THE's description.