Wednesday, June 18, 2008

Another article on Indian education

From Gautam Desiraju (University of Hyderabad), who previously wrote this this piece in Nature, comes another excellent article, "Science Education and Research in India", in the current (June 14) issue of EPW. Unfortunately EPW's website is rather weird and I can't provide a direct link; but The PDF is here (thanks km!), and it is on the front page at the moment and will be freely readable for the next four weeks.

I agree with nearly everything he says. Only with regard to the cost of education I have a disagreement. (Based on the rest of his article, and my previous conversations with him, the disagreement may not be major.) He seems to suggest that, in contrast to subsidising everyone heavily as at present, only meritorious students should be subsidised and others asked to pay full fees. I think the concept of "merit" is overvalued, and financial aid should be entirely need-based (see MIT's policies for example). In the "nature vs nurture" debate, I come down on the side of nurture, but I also believe that nurture does not end at a cram-school in Kota. It should be the job of an elite institution to create merit, not just demand it at the entry level.

This also relates to a current debate on Abi's blog on the IIT-JEE rankings. The reason the IIT-JEE test is so savagely difficult is that the IITs want to rank each individual candidate on entry. I think that is an extremely unhealthy practice. But the reason they want to do it is that they want to assign courses and campuses to students at the entry level: a "meritorious" student picks, say, computer science at IIT Kanpur even before ever entering the campus of that institute. It would be far better to give all students the same broad-based education (covering engineering, science and humanities) for two years, and ask them to specialise in the third year onwards; that way, some bright and motivated students, who have seen the atmosphere at several departments first-hand and done hands-on work there, will learn that there are interesting and productive areas beyond the "hot topics" favoured by the media, reducing the competitive pressure to get into the "hot" departments (and departments can effectively compete to attract the better students); and, to the extent that comparison of student merits is required at all, it will be based on their performance over 2 years at the institution itself, not on a 3-hour examination.

The "merit" mania and the mushrooming of cram-schools has poisoned the childhoods and teenage years of a few million Indian schoolchildren by now. It needs to be tackled, and the IITs should take the lead, since they are the major cause of the problem.


Anonymous said...

Overall, I agree with what you have written.

> The reason the IIT-JEE test is so savagely difficult is that the IITs
> want to rank each individual candidate on entry.

I think you are wrong on two counts here.

First of all, I do not think the IIT-JEE is "savagely difficult". the
problem is that the so-called "Advanced" level (or 12th Standard) for
the sciences in most School Boards is excessively knowledge-based and
low on analytical thinking. Since the IIT-JEE is geared towards the
latter rather than the former it is only natural that school children
who have been used to learning things by rote even for the sciences
find the test hard.

Secondly, the faculty at the IIT's (who design and conduct the test)
are largely in favour of the system you suggest (i.e. delay branch
selection for two years). So, I don't think that they are looking to
"rank each individual candidate on entry". The problem is that they
need to find *some* mechanism to select only 5000-odd students out of
the 3 Lakh or so students who apply.

The difficulty with testing analytical ability is that "everything
has a formula". A smart enough person sitting down with 10 years of
questions can discover certain "patterns" by which questions are set.
These are turned into "formulas" which the coaching institutes then
drill into the students, mostly by rote learning. Blindly carrying
out an algorithm to solve a series of questions will always be more
efficient (in terms of number of questions answered) than thinking
about the questions; providing that the algorithm actually works.

This led to a long war of attrition between the coaching classes and
those setting the question paper with the students caught in the
middle. I have been told that with the "new JEE pattern" of a year or
so ago, an alternate approach has been evolved by the examiners. Let's
hope it works out.

Rahul Siddharthan said...

Kapil - this semi-anonymous guy (who may or may not be well-informed) disagrees with your statement that "the faculty at the IIT's... are largely in favour of the system you suggest (i.e. delay branch
selection for two years)." He says "earlier most courses were similar for 2 years, no longer today, since specialization is the order of the day. And letting students choose specializations will only result in mayhem..."

Even if he is wrong and you are right, why haven't the IITs done this? It would be a win all around, it seems to me. Students feel less pressured to get a "top rank", institutes can evaluate students over 2 years before letting them pick a stream (and try and tempt them into a more diverse set of streams), students can escape parental pressure and choose unfashionable fields.

Rahul Siddharthan said...

ps - One vested interest I can think of is that the lesser/newer IITs may worry that all the top ranks will be grabbed by the older/better IITs. With the current system, at least they can hope for computer science students from the top 200 who fail to get a seat in the "elite" IITs. But if that is the concern, I think they're wrong on many levels...

Anonymous said...

Why haven't the IIT's adopted the idea of delayed branch selections?

The original reason I heard was indeed that the newer IIT's were
worried that the students with intermediate ranks would want to hang
around in the better known IIT's in the hope of a "suitably hot"
branch selection.

More recently, I heard that most of the IIT faculty senates
have endorsed this delayed branch selection approach (ref?).

So why have they not adopted it? I would suspect interference from
the "hot-shots" in the IIT Council.

Anonymous said...

Ultimately, any admission system that tries its level best to get
the "elite" students is bound to fall. Primarily, because this notion
of "elite" students is itself flawed. Secondly, because the education
system will get skewed by the demands of these "elite" students and
their parents. (For example, IIT's are increasingly introducing
courses on "Management" which have little to do with science or

The "correct" approach (IMO) seems to be to decide at the time of
setting the questions what constitutes an adequate score for a
candidate to be eligible for admission.

When that still leaves the problem of too many students who make
the qualifying grade, then the actual score in the test can be used
to offer a seat on a deadline-for-acceptance basis. It should be
understood by all concerned that the latter is only marginally better
than a lottery for most students (who are at near centre of the bell

km said...

Rahul: Great link and here's the PDF you (probably) could not link to: PDF of the report.

Rahul Siddharthan said...

km - thanks! Updated post.

Anonymous said...

How about subsidizing meritorious students who are poor and asking all else to bear the full cost ? If you dont look at "merit" how else will you choose whom to admit ? Are you suggesting that everyone who wants a seat should get it ? Now everyone will want a CompSci seat. What do we do ? Close all other departments ? And what will all these CompSci graduates do after they leave college ?

I am all for free basic education. But it simply does not work economically to provide free degree education.

In UK, almost anyone who wants a degree seat can get it, albeit not in a good college. The results is an oversupply of unmotivated and untalented students, who are simply there because it is free. The state can very well do without this burden. It is a waste of money which could be better spent elsewhere.

Rahul Siddharthan said...

pc -- I agree admission should be on "merit" but I don't agree with a one-point criterion (a single entrance exam) for it. Take a look at how the top US universities do it: SAT scores are a criterion but not the only one. They're more interested in what sort of interesting school projects you have done, what evidence of independent thinking and creativity you have shown, and so on.

That apart, what I was talking about above was not admission but assignment of specialisations at the entry level. Even in India, newer institutes (eg IISERs) are moving away from this model and towards a more broad-based early undergraduate education. It's high time IITs did the same.

km said...

Rahul: The SAT score isn't the only criterion. Race is just as important :)) (Just ask Princeton or Harvard.)

Anonymous said...

BE/BTech is a basic course. At this stage students already know which branch they want. Of course they decide this based on current fashion but who are we to tell them. It would be unpopular among students if they had to choose the branch after joining; I dont think there would be many takers. What is more reasonable is to allow students to change their branch after say first year if they want to. There will be limits on this depending on seat availability. Many institutes like NIT already have this practice.

Making it easy for students to move from one college to another would also help. All this puts extra administrative burden. In a system where there is plenty of choice, it would not be a problem.

I agree that current methods of selection based on a single exam is not the ideal one. But we live in a society where everybody is assumed to be a cheat; hence you have elaborate rules and regulations. Professors are not trusted to do their jobs impartially and honestly; hence you need a centralized exam. I dont like this state of affairs; but under the prevailing conditions of shortage I dont see any other option. Shortage always leads to corruption and that makes innovative approaches difficult to get accepted. First address the problem of shortage and then you can play with all sorts of ideas.

Anonymous said...

reg. EPW article and S&T education in general -
Yes, that excellent scientists, engineers, teachers are required and the education system has to nurture these stars of tomorrow. But has anyone given a thought to that S&T education which produces good technicians, scientists, engineers who will man the lower rungs and labour aspects of the scientific enterprise ? For. e.g. it is true we need the top class "single" PI theoretician/experimentalist but what about large scale experimental science (or for the matter of fact even industry )that require many folks who do the nuts and bolts of technology and tools. I would dare to say that the Indian education system fares poorly even in this regard.