Saturday, 5 October 2013

On University Rankings, Again.

This is an expanded version of an article that appeared  in the Indian Express on October 5 2013. As you can see, I have used some of the material from my earlier blog post on this subject.

Rankings: Why is India Nowhere?
Gautam Barua
QS and THE have  recently released their 2013 global University rankings.  In both the rankings, institutions from India are nowhere in the picture. In fact, their ranks have fallen compared to last year. Are institutions from India that bad? Have institutions from India been “lazy” in providing the right data? To show that there are problems with the  rankings, I analyse the parameters of IIT Guwahati (since these are available to me, but the results can be easily generalized to other Indian Institutes; IIT Delhi  and Panjab Univ. data are also shown). The first table is for QS and the second for THE.

Academic Reputation (AR)
Employer Reputation (ER)
Faculty: Student Ratio (FS)
Citations / faculty (CF)
Intl faculty (IF)
Intl students (IS)
Total Score


Max marks
IITD (2012)
IITD (2013)
IITG (2012)
IITG (2013)

Industry Income
International Outlook

All scores are relative, with the top ranked institution in each category getting a score of 100.
1. The FS ratio of IITG is the best among IITs (according to the information tabled in the lok Sabha and as reported by the media) but it is showing a decline and is much worse than IITD’s in the QS table. Clearly there is an error here.  IITs have not done well  in the faculty to student ratio, and it is well known that there are many vacant faculty positions. But even here, the methodology is not clear. MIT states in its web site that it has a faculty to student ratio of 1:8. It also states that the total number of students is 10,894 students. The number of faculty is given as 1018, whereas 1362 are required to meet the above ratio. Then the number of “senior lecturers, lecturers, and Professors emeriti” are given as 540 (most of whom are on contract for temporary periods. Possibly 1362 is reached after deleting the “Professors Emeriti”.  But how are scores calculated?  Consider this: MIT with a 1:8 ratio gets a score of 99.9, while IIT Guwahati with a ratio of 1:13.3 (as per QS) gets only 35.5. I am not able to figure out how this score was arrived at. In any case, is this a fair marking system? Shouldn’t an ideal ratio get full marks and other ratios be given lesser marks in slabs?

2.  IITs are not allowed to take international students at the B.Tech level  unless they sit for JEE. Very few do.  There is scope for increasing the number of foreign PhD students. But even here there is a restriction, as Govt. Assistantships   can be given only to Indian citizens. Without aid, it is difficult to attract good international PhD students. Hiring international faculty on a regular basis is  not allowed. They can be hired on contract upto five years, but only if the salary is  at least $25000 annually (only Profs  are allowed effectively). The moot question remains: is the internationalization of campuses an important parameter for excellence? The Western countries are clearly at an advantage here with Indians, Chinese, and others going to their Univs in droves..
3      50% of the weightage is based on “reputation” (AR: 40% and ER: 10%) in QS and 33% in THE (not shown above). IIT Guwahati got a score of 0 for Academic reputation and a score of 1 in Research reputation in THE. How did they arrive at these scores? It is a wonder that top students of the country competing to attend an institution with a ZERO academic reputation! These scores are  getting reflected in the scores for Teaching and Research in the table above. These organisations are now aggressively marketing their products through which institutions can enhance their “reputation”. Thus we have been invited to advertise in their “QS Top University Guide 2013” (with discounts if we opt to advertise in more than one language) and in other publications, to attend seminars and conferences (with registration fees of course), and so on. Can we rely on reputations primarily to decide ranks?    Academics all over the world are asked their opinion of the top institutions globally and in their country. The chances of getting an IIT’s name included by a US professor are quite slim.  Alumni in academics may help the IITs as they know about them, but the number of alumni in academics in the US is a small fraction of the number of alumni in the US. This is a legacy of the past. Young institutions like IIT Guwahati without sufficient alumni, are at a particular disadvantage. The number of respondents is proportional to the number of institutes available for selection in that country. So the responses are heavily weighted in favour of responses from the developed countries. Respondents are not asked to give their inputs for each of the listed universities (it may be impractical to do so, as there are a large number of them). Instead, each respondent is asked to give a list of 5-10 Universities he or she thinks are globally well known, and well known in their country. This method perpetuates the existing ranks.
4. Now, consider the categories CF and Citations. The total number of citations in the last 5 years is divided by the number of faculty in the last year by QS. IIT Guwahati had 323 faculty in 2013 but only 220 in 2009. So its numbers clearly cannot be compared with institutions like Cambridge and Oxford where the faculty numbers are almost constant. Further, since a five year average is taken, one or two “star” papers can make a huge difference to the numbers. For example, a review paper “The Hallmarks of Cancer” authored by two professors from UC San Francisco and MIT has about 10,000 citations. This paper alone will have boosted the CF figure of both these institutions significantly. THE clearly uses some other method for citations. It probably does not remove self citations.  The high scores of Panjab and Guwahati vis-√†-vis IIT Delhi could be explained by this.  Panjab University’s High Energy Physics group (and to a lesser extent IITG’s) is part of global experiments at CERN, Belle, Fermi Labs, and papers from that project have very high citations. Thus a small of group of international collaborations are providing a high score.  Isn’t the median number of citations per faculty a better measure than the average (there are other issues, for example, citations in the Sciences are usually much more than in Engineering)?

So, what can we conclude from all of the above? Surely it should be clear, ranking of universities is not a simple task. We have only scratched the surface as have QS and THE. There are so many other aspects of an educational institution that they have not even touched upon. Many of these aspects are qualitative in nature, and it is very difficult to quantify them. This is not to say that Indian Universities do not need to improve their rankings. They do, and to begin with, we will have to provide data to these organizations in the format they expect. Interactions are already on. But if Society wants Indian Institutions to get appreciably higher QS  and THE rankings, then they must allow the Institutions to do the following: a) spend heavily to aggressively market the Institute among academia and Corporations in the US and Europe, b) substantially increase the number of foreign students (Govt. must allow UG admissions and allow Govt. assistantships to foreigners, and remove ceilings on incomes for foreign faculty)  c) hire a large number of temporary “teachers” to boost the FS number (which counts the number of “academic staff”, and which apparently is done by many US universities), d) create a network among Indian Institutions to encourage citations of papers of other Indian Institutions (scratch each others’ backs). Finally, of course, all Institutions must strive to  improve the quality  and quantity of research, teaching,  industry interaction, etc.


  1. Excellent analysis of the recent rating controversy by Prof. Barua. But at one point he says that international professors need to be paid a minimum dollar salary and later on he says that the ceiling on salary to international professors has to go. Is it me or is there an inconsistency?

  2. skeptic wonders "Is it me or is there an inconsistency?" I don't see what is the inconsistency. At present, all faculty members, including foreign ones, have the same pay scales. Visa norms require the minimum pay to be USD 25000 per year and at the current common pay scales, this is met only at the full Professor level - so, any foreign Assistant or Associate professors hired will not be eligible to get a work visa. If we wan't to get more foreign faculty, we have to allow different pay scales. The maximum pay any faculty member can draw at any IIT is today Rs. 1.56 lakhs + 23K for HRA. The minimum pay for an entry level asst prof is Rs.78,280 + HRA 11,400 and for asso prof it is Rs.1.05 lakhs + HRA 15690. Including HRA, the asso prof salary is 23.5K USD per annum, not eligible for work visa. Prof. Barua's point is that unless the salary structure for foreign faculty is different, it is unrealistic to expect any significant number of them at any of the IIXs.

  3. It is a very good analysis. Ultimately, rankings are meant for undergraduate students all over the world to choose their institution to do graduate school. There is very less likelihood that this is how it will be used in India.

    You can include IISc data in your table also

    Teaching 36
    Research 37
    Citations 59
    Industry income 36
    International outlook 17


    Giridhar Madras

  4. Skeptic: Actually, I did not mean what IITMsriram said, but what you inferred: I meant that the "floor" and not the "ceiling" should be removed. I dont think it will be possible to have a different salary structure for foreign faculty.

  5. Old thread, but MHA has issued a circular on Dec 20 that for faculty in IIT, NIT and central universities the minimum eligible salary for visa will be USD 14000 instead of the common USD 25000, so the floor has been lowered (to the equivalent of entry level asst prof).