Faculty Selections: Preference for PhDs from top institutions?

 Recently, Prof. V Ramgopal Rao, Vice Chancellor of BITS Pilani, commented on LinkedIn about the practice of giving higher weight to PhDs from certain institutions in shortlisting criteria. In the specific example, the institution has given 25 marks to PhDs from IITs and top 500 ranked universities in the world, 20 marks to NIT PhDs, 15 to CFTI PhDs, 10 to state government institutions and 5 to others.

While his specific comment was use of source of funding to judge the quality of PhDs, the larger question is whether any criteria that gives preference to the students from certain set of institutions is a good criteria.

Let me admit that the specific criteria was very poorly designed even if such a criteria is needed. For example, a PhD from a university which is ranked in 501-1000 in the world will be considered amongst the worst category in this list. But assuming that someone can do a better categorization of universities, would it then make sense. Almost everyone who commented on Prof. Rao's post (barring a few exceptions) seem to agree that it wouldn't make sense even then.

One of the rare rebuttals had this to say:

"Prof. V Ramgopal Rao It's seemingly easy to successfully play to the gallery. I will be happy to see at least one "excellent faculty candidate from a non-descript institution" at BITS.

As a matter of unwritten rule, top pedigree, foreign post doc and at least one Q1 paper always mattered in IIT faculty candidate shortlisting process. On the other hand, barring a few, almost all NITs till recently nurtured Godfather's candidates or a deep regional bias in the faculty recruitment process. Course correction has just started with the plethora of applications as IITS are churning out Ph.Ds in hundreds of not in thousands and some directors are determined to shed nepotism, parochialism and regional culture.

This is precisely because if you don't disregard the non-descript institutions, you run into thousands of applications in search of fire in their belly at the cost of tilting the faculty quality balance in your own institution. Truth is better told louder (by e.g NIT Hamirpur) than practised in silence by all directors of prestigious institutions.Does maintaining public transparency amount to trivialising the recruitment process ?"

Interesting debate, and it made me think. Of course, almost everyone use pedigree as a signal. In fact, a lot of people have argued that the value of the university is more in signaling than anything else, and if somehow we are able to evaluate a large number of people quickly, the top universities of the world will face tremendous challenges in attracting good students. After all, the next best institutions aren't that bad but cheaper, easier to get in and all that. (And it is happening slowly with AI evaluating millions of CVs on a portal - which, by the way, has its own problems, but that is for another time.)

But I still disagree with the specific rebuttal. It almost gets personal. I think there is a difference between factoring in pedigree as one of the many indicators of quality informally, and formally saying that we have a strong preference for pedigree and it will impact shortlisting. For example, at JK Lakshmipat University (JKLU), we always say in our faculty ads for engineering faculty, "Preference for IIT PhDs." To me, that line is a signal to potential job seekers that my quality requirement is at least an average PhD from an average IIT. If you think that you are equal to that, please do apply. (And in the last two years as VC of JKLU, we have made several offers of faculty positions to non-IIT PhDs, though the majority offers are to IIT PhDs.) There is no doubt in my mind that the best PhDs from NITs are better than worst PhDs from IITs, and hence a blanket ban or reducing the order in the shortlist without checking any other parameter of the quality is problematic.

But the rebuttal raises an important question. What is to be done if there are a large number of applications. And this is where, Prof. Rao is silent. Indeed, I feel frustrated in reading a lot of posts from Prof. Rao because he will raise questions but either no answers or the answer is a single word, "leadership." He does not explain what leadership should do. Raising questions is a good service to the nation, but I think someone with his pedigree should also analyse the situation in greater detail and suggest solutions.

The way I look at the issue is this. What is the problem being solved through this mechanism by the specific institution. The problem today is that the government in 6th pay commission and later in 7th pay commission raised the faculty salaries tremendously. This did lead to improvements in faculty selections in some government institutions (though poorly funded government institutions stopped hiring regular faculty and started getting dependent on ad hoc faculty). But these salaries are beyond the capacity of an average Indian student to pay for. Therefore, the private institutions salaries are much less (in most cases the fees is capped so even if the private institutions were to try increasing the salary, they can't). Now, the private sector employs 80% of the faculty. So whenever a government institution like an NIT advertises a job, a very large number of applications pour in since the salaries and perks are so good and along with that they are funded well for research (again, something that private institutions are unable to do) and there is stability in job. What should be the process to deal with such a deluge of applications and select the best faculty.

There are two processes really. One process is to do the shortlisting, and the other process is to do the selection. Now, if there is a deluge of applications and it is understood that faculty is the most important resource for an educational institution, the first thing that should happen is that the institution should be willing to have a larger number getting shortlisted and spend time and resources to deeply consider that large number of shortlisted candidates. If an NIT is using a shortlisting criteria to decide interaction with only 5 individuals for a post, they are not serving themselves well nor are they being fair to the applicants.

On the other hand, shortlisting a large number of candidates will also not lead to good selections. If we shortlist 50 candidates for a post and interact with each of them for 10 minutes each, that selection will eventually lead to some bean counter counting the number of papers and other parameters.

To criticize shortlisting is to miss the point that a large shortlist will compromise the selection process badly and while it may seem "fair," it actually will be much worse.  For example, if research output is one of the most important criteria then I can either shortlist 20 candidates based on number of papers, citations, ranking of those journals/conferences and so on, and then read a few papers, get a few experts to comment on them, get these 20 to give research seminars and interact with them, and all that. Alternatively, I can shortlist 50 candidates, and then I don't have time to do any of the above. I interact for 10 minutes each and I have no option but to only decide selection based on the CV. Taking a call based on this 10 minute interaction will usually not result in good selection.

So shortlisting is important and it is important to shortlist a reasonable number - not too small and not too large a group. (I am using 20 as just an indicative number, it is not a magic number.)

Now, ideally, the shortlisting too can be subjective like selection often is. But there is a crucial difference between the two processes when you have such a deluge of applications. As an NIT, you can request an external expert to come for 2 days and interact with 20 candidates over 2 days (and even more, if needed). It will be very difficult for an NIT to seek help from an external expert in reading 100s of CVs and then record their comments on each of them and then help in shortlisting. Remember, you need multiple experts who should have a deep insight into their area and have some understanding of the discipline at large so that they can work with other experts to compare candidates across sub-disciplines. Frankly, not a very practical approach. And hence it becomes necessary for them to use numbers as a proxy to quality.

Also remember that regulators like UGC and the politicians have been strongly encouraging objective criteria. UGC has been sharing a methodology to give marks for every SCI paper, or every FDP, or every year of experience and so on. So you have a large number of applications, you don't have strong internal resources for evaluating them, and you have the government telling you that you must use objective criteria. What are your options now?

So the only thing you can do is to choose those numbers carefully. Anything that you will choose will be criticized. This is because your choice of parameters can only be dictated by statistics (or perception of it) and individuals cannot be evaluated based on statistics. For example, statistically, an average PhD from an IIT could be in some sense better than an average PhD from an NIT, and hence we may be tempted to assign higher marks for an IIT PhD over an NIT PhD, but we all know that the best PhDs of NITs are certainly better than worst PhDs of IITs (they could even be similar to the best PhDs of IITs). Similarly, someone with 10 SCI papers is likely to be a better researcher than someone with 5 SCI papers, but there are far too many exceptions to this.

So any objective criteria will lead to missing out of talent. But remember not doing shortlist or doing a very liberal shortlisting will compromise your selection process to the extent that you will miss out on talent there. So the question to me is the following:

Assigning 5 extra marks to an IIT PhD compared with an NIT PhD will lead to some talent missed. At the same time, assigning 1 mark for each SCI paper will lead to some talent being missed. The solution is to have subjective evaluation but I do not have the capacity. What should I do. And frankly, there is no easy answer. Whenever anyone has asked me for an answer in the past (when I was not VC), my solution has been that I can personally help you with the subjective process in Computer Science if you do not have the capacity to do it on your own. If enough faculty members of top institutions are willing to help the next tier institutions, this comparison of two bad policies will not be needed. That is when the leadership of NITs will come into play because they will have to defend the subjectivity. A large number of applicants will point out that their number of SCI papers was larger than those shortlisted. They will put in RTI queries, they will go to court, the government will ask questions, and the leadership will have to stay firm. But if the leadership knows that they don't have the capacity and it is not easy to get several experts in each area to do this semester after semester, I wouldn't blame them for using objective criteria.

Of course, one still has to see what objective criteria is likely to retain most of the talent in the shortlisted pool. Ideally, one should decide the shortlisting criteria after one has seen the applications. I would like to do the experiment of tweaking the criteria and seeing that with each tweak, what applications get missed out and what comes in and then study those applications to see what works better for the institute. This is where I would blame the leadership. Most institutional leaders would love to announce the shortlisting criteria upfront and keep it same for all disciplines which is frankly, stupid. It is because in financial matters, tenders, it is assumed that if you do not reveal the criteria upfront, you will manipulate it later to benefit someone and CAG/CVC have taken exception to this. Yes, the same thing can happen in faculty selection also, but a leader should be confident that s/he will not let it happen in the final selection where external experts will play an important role.

Finally, should we have 5 extra marks for IIT PhDs over NIT PhDs. I don't think the answer is as straightforward as Prof. Rao makes it out to be. It is very nuanced as I have explained above. I wouldn't want to do it myself, but would I criticize others who do it, may be not.