Thursday, October 30, 2014

Direct PhD after BTech

A few months ago, we computed an interesting statistics. What is the median time taken to complete PhD by students of different academic backgrounds. We divided students into three buckets - those who had joined after BTech (that is, four years of college), those who had joined after MSc/MA/MCom, etc., (that is, five years of college), and those who had joined after MTech (that is, six years of college). The time taken was defined as the time from first registration to submission of thesis.

We were surprised by the results. All three categories of students had a median completion time of 5 years and six months (across all disciplines - students after 5 years of college are common in sciences, humanities, and management, and students after 4/6 years of college are common in engineering).

To put the result in perspective, we used to have a requirement of minimum 10 courses for BTech graduates, 6 course for MSc graduates and 4 courses for MTech graduates. Now, if we look at the BTech and MTech students, who both are mostly getting admitted in engineering departments, a BTech graduate is doing 3 semester course work and 4 years of research and writing thesis, while an MTech graduate is doing 1 semester course work and 5 years of research and writing thesis. So an MTech graduate is taking one extra year of research even though s/he already has experience of conducting research, has already written one thesis, and in many cases had a publication.

We don't fully understand the reasons behind this, though my own guess is that the duration is based on the expectation of students and faculty members. There is perhaps an expectation that PhD shall take 5-6 years irrespective of the academic background, and irrespective of the number of courses, etc. The hope is that it is something else. May be it is because BTech graduates are younger, and work longer hours. May be BTech graduates join at an age that they can complete their PhD, join a job and then worry about marriage and family, while MTech graduates are two years older on an average when they join the PhD program and are more likely to marry and have family during the PhD, which will provide some bit of distraction.

Whatever it may be, it is quite clear that it is in the interest of BTech students to go for PhD directly, as they save 2 years. The only issue in the past has been the long commitment of 5-6 years. It is difficult to make that commitment without understanding what you are getting into.

IIT Kanpur has made it easier for BTech students to make that commitment. It is possible to leave the PhD program in between with an MTech degree if you lose motivation or for whatever reason, you wish to leave the program in-between.

Another issue that we used to face earlier in our PhD recruitment was that many good students did not give GATE which was a requirement for getting financial assistantship. A few years ago, MHRD allowed that we could pay from government funds, a stipend to graduates of IITs even if they did not have a GATE score. Recently, the IIT Council has decided that IITs can pay stipend to graduates of ANY Centrally Funded Technical Institute, even if they do not have a GATE score. CFTIs include NITs, IIITs, and a few other institutes (and of course, include IITs).

The last problem with PhD recruitment was that students would get a job in December, while we will call them for interview in May, and by then they have already committed to join this job, the families are looking forward to a gift from the first salary, they have asked their friends to find an apartment in that city, and so on. After so much of mental commitment to join the job, it is extremely difficult for them to apply for PhD, then come for an interview, and if selected, give up that job.

We are solving this problem by seeking applications now, in October/November, and holding the interviews in December. We will make the offer within a few days of the interview for you to join in July, 2015. (In fact, if your exams get over early, feel free to join in May itself, why waste time, and why not start getting a stipend early.)

So here is the deal:

If you are from a CFTI, and a final year student with a CPI/CGPA of at least 7.5 (in case of BTech) and at least 7.0 (in case of Integrated MSc), and are going to graduate in summer 2015, please apply for PhD admission to IIT Kanpur now. Since this is the first time we are doing it, only a few departments are taking part in this round of admission. These are: Computer Science and Engineering, Material Science and Engineering, Biological Sciences and Bio-Engineering, and Chemistry. If you wish to apply to another department, send an email to them, may be they would have changed their mind too, or would change their mind by looking at your excellent application.

If you are not from a CFTI, and you already have a GATE score, feel free to apply now.
If you are not from a CFTI and are giving GATE in winter, apply to us as soon as the GATE result is out.

For more information on direct PhD after BTech for CFTI final year students, please visit this link.

For admission to Computer Science and Engineering Department, please visit this link.

Saturday, October 25, 2014

Review of Academic Programs Review

As we all know, the Institute has had a tradition of reviewing all under-graduate programs and associated academic processes every 10 years. In 2008, the Institute set up an Academic Programs Review Committee, which was to review, for the first time, both UG and PG programs. The new programs for under-graduate students have been implemented from July, 2011. In this blog we look at the changes brought in and review them.

The review's biggest boast has been that it allows students unprecedented flexibility. We now have minor programs (3 regular courses typically), double major (about 10 courses), and dual-degree (the PG degree is not just in the same discipline as the UG degree but can be in any other discipline). We also have a new program called BTech in Engineering Science, which is available in two variants, both sort of inter-disciplinary programs. We have a large number of Open Electives which can be used to tailor one's degree based on one's interest and can also be used to take up courses that will lead up to minor, double major and dual-degree.

What is the reality. Unfortunately, the reality is rather grim.

It was envisaged by the authors of the report that the courses constituting minor program would be open to a large number of students (at least 165+ students will be allowed in each such course). Even if someone did not want to complete a minor, these courses would have enough capacity that students would be able to take these courses for their open electives. Unfortunately, the committee did not look at the earlier variants of minor, and why they failed. In an institute where teaching a larger class counts for nothing, why would anyone want to teach a larger class (except when it is forced like the core courses). Teaching larger class gives less time to faculty for research (the only metric for all promotions, awards, chairs, and lately even administrative positions). Teaching large classes invariably leads to poorer teaching evaluations. (If we look at the statistics of student reaction survey, the average of larger classes is less than the average of smaller classes.) From the department perspective, a larger class means many more TAs and tutors, who could have been deployed for core department courses. So a large class of outside the department students impacts its own core classes. And then saying that every minor course should allow 165 students made no sense, since it meant that EE department who would have about 150 of their own students only had to allow 15 non-EE students, while the Physics would have to allow 135 of other students, besides just 30 of their own students.

So the departments protested, and made sure that only a small number of students are able to take a minor program in their discipline. Ideally, there should have been some kind of incentive mechanism to the department to teach as many non-department students as possible. Since the numbers are very small, it was no longer possible to just let the students take minor courses and apply for the minor at the time of graduation. We now expect the departments to have a system of "admission" to the minor program. The OARS will mention who are enrolled in minor so that the students can be allowed to register for the minor courses. This has made administration of minor a difficult affair. Also, minors are only within a department. Earlier, it was envisaged that there will be minors consisting of courses from more than 1 department, but that has not happened. Also, 3 courses for a minor is just too few to really appreciate that discipline, and has ensured that we don't have multi-disciplinary minors. But given that we all want to minimize our teaching load, almost all minors are of 3 courses only.

The double major is even worse. In the first two batches, the students started doing double major only in the 6th semester, which is very late. Unless the student had enrolled in a minor and has done some basic courses of the second major, it is almost impossible to complete all those 10 courses within two extra semesters, even after using the open elective slots for some of them. And mind you, we are only talking about a secondary major and not a second degree. I think if we plan things well, it should be possible to do two under-graduate degrees in 5 years. But what could have been the most popular level of flexibility, is not even on offer. The rules of double major are so complex that anyone who reads them will never even apply. And the CPI limit of 8.0 for entry into double major makes no sense. We allow students with as low as 6.0 CPI (in special cases) to go for dual-degree. So what are we saying. That someone with a CPI of 6.0 or 6.5 is quite capable of doing PG level courses of the other department. However, such a student is not capable of doing even the basic UG level courses of the other department. No wonder we have only 14 students out of about 820 in the 2012 batch taking this option. (9 of them for Computer Science.)

The case of Engineering Science is really pathetic. In the first three years (2011, 2012 and 2013 batches) in which it has been offered, only ONE student has moved to BTech in Engineering Science, that too in 2011 batch. None in 2012 or 2013. There is no champion for this program. A whole lot of faculty members had suggested that an inter-disciplinary program at the under-graduate level would be a big hit with the students. But none of them are really keen on motivating students to do this program. There is no commitment to offer courses, to help in placement, and other such things. The sad situation of this program has ended one of an interesting experiment in admission process. Had this program been a success, it would have been possible to start several more programs where the admission is after one year through a branch change process. For example, the proposed under-graduate program of the Earth Science department could have admitted students after 1st year, and this would have been a great step forward in terms of deciding the programs for the students after one year. But alas, the program has died even before it was young. We should ensure that this program is not on the rule book at all.

An alternative Engineering Science program could have been an exit option for academically weak students who could have been given this degree after completing a bunch of engineering and science courses instead of terminating their programs, something similar to the UG Diploma of IIT Delhi.

The dual degree programs where the PG degree is in a different discipline than the UG degree are in a mess. (The others too have problems, but at least the problems there is that the committee has not tried to solve any of the dual-degree problems. They have not created many more problems.) What is the graduation requirement for these programs, there is no philosophy for deciding the graduation requirements of such programs. In case of PG degree being in sciences or economics, we do not even know what degree will be given. Will it be "Master of Science." If yes, is the requirement of 2 year MSc anywhere close to requirement of this Master of Science degree. If it is a different degree as many faculty members have claimed and the graduation requirements are significantly less, then is it really comparable to a master's degree. Will students be able to join PhD program in any place outside IIT Kanpur.

An obvious way to specify the graduation requirements would have been to say that a dual-degree would be given if the student completes the graduation requirements for both the degrees and certain credits (like open electives) could be counted towards both the degrees. Or some specified credits could be waived for dual-degree programs. But we don't do obvious things. We have this fascination with templates. Every program has a template and every template has to be such that a dual degree can be completed in 5 years, at least theoretically. If this is becoming difficult, let us reduce the courses even further. Five years is sacrosanct, not credits. But we will claim to be running a credit based system.

If one asks anyone related to administration of existing dual-degree programs, one complaint that you would hear from everyone is that having two roll numbers was an absolute brain dead idea, which is single most pain point for everyone involved - students, DUGC/SUGC, DOAA office, OARS, and so on. And guess what do we have for the new dual-degree programs - yes, you guessed it right, we will continue to have two roll numbers. Not only that, the implementation committee has mandated that the PG roll number should encode the department and the program. Talk about progress. And of course, the PG roll number can be allocated only after 7 semesters are over and they have passed most of the UG credits. But they are supposed to do PG credits even earlier when they did not have PG roll number. And during the pre-registration of 8th semester, they can't have PG roll number, but they have to register for PG credits. The APRC has made sure that as a Dean of Academic Affairs, I am always busy in solving non-issues.

The last component of "designing your own degree" was a large number of electives. In particular, Open Electives, which you were supposed to take from across all the departments depending upon your interest. But as we discussed above, most faculty members are not keen on teaching larger classes than absolutely necessary. So most will reject non-department students on flimsy grounds. Or better still, won't even offer a course which does not have multiple department courses as pre-requisites. So most students end up doing more department courses under the OE slots - no broadening of horizons envisaged by APRC. The department which allows a large number of non-department students in its courses as Open Elective is IME. A large number of students do MBA courses. And we all keep wondering why students are not interested in engineering. If we don't even offer interesting engineering electives and they only find interesting electives from MBA, what are they going to do in future.

The new curriculum has 5 HSS courses as opposed to 4 earlier. However, the HSS faculty continues to be overloaded with teaching duties. Number of students taught per faculty is the highest for HSS department, and as a result the flexibility of getting HSS courses of one's choice has further reduced.

Having all admissions to 4-year program is a good idea as it does not force students to decide about the post-graduate degree right in the 12th class when they have no idea of even under-graduate program. However, it would have been better if this was accompanied by departments trying to attract students to the dual-degree programs. To the best of my knowledge, there have been no attempts made by various departments to talk to the students and convince them to do the second degree. Now that the placements are quite good for the dual-degree students, I am sure we will continue to see interest in such programs, but one poor placement year, and you would lose all students from the program. And these dual-degree students could have been an excellent resource for TAs and Tutors in various core courses, further improving our teaching.

The other problem with 4-year programs is that we have made no attempt to sensitize others in the country about our BS programs in sciences and economics. This would seriously create problems for the graduates of these programs if they do not migrate to dual-degree or double major.

Another kind of flexibility was introduced through modular courses, courses that would last only half the semester. However, there are hardly any courses in this category. Modular courses would have been great for a faculty member to voluntarily teach a new topic to a few students, may be just 2 hours a week for 7 weeks, but given that teaching is no longer given the same credit during the promotion, awards, etc., this has not happened. We could have brought in visiting faculty, who would find it easier to come in for 2 months rather than 4 months, and offer modular courses, but that has not happened. One reason is that even 2 months is too long for some of the visitors, and of course, we have not really looked at visiting faculty option seriously. The modular courses fit into summer term so perfectly, but there is always a confusion about who is eligible to do what in the summer.

One shortcoming of the modular courses has been that it requires you to teach over 2 months. I could get a course approved for 1 hour a week for 7 weeks as a modular course, that is, the smallest modular course could be just 7 lecture hours. Now, why can't we have a visitor for a week, and deliver 7 lecture hours during the week. It would be so much more interesting. We could permit some such credits on pass/fail mode so that the visitor is not burdened with the issue of grading, etc. Also, we need to make sure that such courses can be floated with short notice and that such courses can be added to the registration of the student even in the middle of the semester. But nothing of this sort is allowed.

As a result of all this, there is very little flexibility that a student can exercise and enjoy during his/her program. The concept of "designing your own degree" has remained on paper.

There are several other issues too with the new program.

If we look at the examinations, we have moved from two mid-semester exams to a single mid-semester exam. This has destroyed whatever little of the continuous evaluation we were left with. The number of courses which have quizzes in the first half and the second half of the semester is minuscule. The number of courses which have the end-sem weight of 50 percent or even higher has increased tremendously, and with that the stress levels of the students have gone up too. The number of requests for not having two exams on the same day or one exam in the afternoon and the next exam in the morning of the following day has increased tremendously. There was really no reason to migrate from 2 exams to 1 exam during the semester.

The APEC rules (for warning, probation and termination) made absolutely no sense, and have already been changed multiple times, and now we have different rules for different batches. Hopefully, the latest rules will provide stability.

The students getting above 8.5 CPI will be awarded distinction. Is it really needed. Don't the companies and all other stake holders already know that 8.5 is higher than 8.0. I am sure everyone looking at our transcripts know this much of mathematics to compare two numbers, and no purpose is served by saying that a student will be awarded "Distinction" beyond a certain CPI. It would have been so much better to offer such a distinction (or rather an honors degree) on completion of additional credits at a good academic standing, something what many other IITs do.

One of the most serious problem in the new program structure is that the graduation requirements do not mention a minimum CPI. The Committee was able to convince Senate that this is what a credit system is - you pass certain credits and you get a degree. The fact of the matter is that no good institute (except now, IIT Bombay) will give a degree on passing all courses. Most universities would want you to not just pass courses but also maintain a C average (6.0). In our case, the minimum of 5.0 had worked very well in the last 50 years, and there was really no need to change. This clause alone has caused serious degradation of academic standards as now many students are ok with getting D grades in the courses. When students are happy with a lower grade, in a relative grading system, every one can get the same grade with lower effort, and that is not good for the system. If we design a system which allows students to get away with doing very little, we should not be surprised that students do get away with very little. Instead of criticizing the students for gaming the system, we should look inwards and see if we deliberately designed the system to encourage gaming.

Overall, the new program launched in 2011 has caused more problems than it has solved, so much so, that the right thing to do will be to roll back and then think of a fresh set of small incremental changes rather than the wholesale changes that have been introduced without a real understanding of what will work and what will not.

Thursday, October 23, 2014

Dropping Courses Late in the Semester

 In 2007, I took a proposal to Senate of IIT Kanpur. This was to allow students (only UG students at that time, PG happened a year later) to drop a course for any reason whatsoever till the middle of the semester. (Up to two weeks after the first mid-sem. We had two mid-semester exams in those days.) The idea was that students could not judge how much load a course would be in the first week of the semester, and hence need time to take a decision on drop. Also, situations change. Someone might have some medical issues. Someone might get too involved in extra-curricular activities. Someone might have a stress due to a relationship issue or a family issue. We must support students through all this, and the least we could do was to allow a course drop to the student. It was clear in the proposal that the instructors would have very limited right to say no to the drop, and they could refuse only if dropping the course would cause a problem in the conduct of the course. A specific example was given: If there are group projects in the course, and one member of one group wants to drop out, it could affect grading of that group and it may not be possible to reorganize groups and projects. The DUGC role in all this was to counsel the student, for example, if the course is pre-requisite to future courses, this could be pointed to by the DUGC Convener. So the word used in the proposal and in the Senate minutes was "endorsement" and not "permission." Instructor and DUGC Convener were only to "endorse" the application.

Students were obviously happy. Any flexibility is good from their perspective. And the Academic Program Review Committee proposed to keep this flexibility even after the two mid-sems were merged into one mid-semester exam. And since the exam papers were now 2-hours instead of 1-hour and hence grading would take more time, the deadline for drop was extended from 2-weeks after 1st mid-sem to 3-weeks after the only mid-sem exam.

The number of students who have been seeking to drop courses has been on the rise throughout these 7 years. And this increase has got the faculty worried. We hear all sorts of horror stories - courses which have lost 50 percent of their students, DUGC Conveners who have signed several hundred approvals for dropping of courses, all students in certain departments dropping courses and completing their degrees late, and so on. Whenever any such claim is made, I try to check the data in DOAA office, and of course, what I see is not the same as what was claimed.

After all, the total number of drops after the first week of the semester is roughly 4 percent of total registration. Out of which a substantial contribution is of those who have decided to leave the Institute or take a semester drop. I am sure faculty has no objection to such drops. Then there are students who are weak and want to reduce their load from the normal load. This too should be ok. The problematic case is when student took overload initially and then later drops the course. This is being seen as gaming the system. I am told that faculty members feel demoralized when someone of this type drops the course, since they take it as a reflection of their teaching abilities. Some faculty members also see this as an additional teaching burden, which has gone waste.

Really??? Such drops form between 1 and 2 percent of total course registration. Surely, in some courses this number could be 5 to 10 percent. (Of course, in some courses with just one student, it will be 100 percent, if that student decides to drop. But let us talk about courses with at least 20-30 students.) Should a faculty member feel demoralized because 1-2 percent students have decided to drop the course. Should we really think of this additional 1-2 percent extra workload (in which we only graded one exam, the end-sem is not done yet) as such a serious overhead that we need to penalize the students.

We are hearing of all sorts of penalties. There may be a drop fee. The drop may be mentioned on the students' transcripts. And so on.

Why do faculty members think like this. In my opinion, it is because the credit based system is still not part of our DNA. We would like to control and dictate what students should do each semester. (The Senate/SUGC have created templates of all programs and combinations of programs - and some faculty members would want them to be mandatory for the students to follow.) When we talk about flexibility, it is only that a faculty member as an academic administrator should be able to accept/reject applications of students on a case-to-case basis. But students should be able to take their own decisions - this is too much flexibility.