Tuesday, March 31, 2009

Open access

Yesterday, an open access proponent visited IISc and interviewed the Director and Associate Director. I was also interviewed on the applicability of open access and how it influences the research scenario in India. Unlike some academics who believe that scientists should publish in open access journals only, I have a different view as a researcher.

Unlike other professionals, research scholars write journal articles for free to advance knowledge in their field and, possibly, their careers. The entire process of writing a research paper, reviewing by other peers and editorial decisions are carried out at nearly no cost. There are four kinds of journals: (a) author pays, reader pays (b) author pays, reader does not pay (c) author does not pay, reader pays (d) author does not pay, reader also does not pay. Category (a) is followed by a few journals that are published by some societies. In this case, the author pays the page charges while the reader pays for access. Category (b) is the current open acess model wherein the author pays for publishing the article but the reader does not pay for access. Category (d) is a variation where some open access journals waive the payment for authors. These journals are generally supported by trusts. Category (c) is the most common and is adopted by all commercial publishers. Previously, when most the journals were available only in the print form, categories (b) and (d) did not make much sense and open access was not commonly used. However, with most journals going online and some journals going exclusively online, the cost of print is not an issue anymore.

Why is there such an uproar against commercial publishers and for open access? The average cost of a journal in chemistry, physics, engineering and biology are US$ 3490, 3103, 1919 and 1810, respectively, with an average increase of 5–9% in cost per year. The consumer price index (CPI) for the USA city average for all items is a widely used measurement for the general rate of inflation in USA. The Association of Research Libraries 3 USA mentions that the CPI has increased by 73% between 1986 and 2004, but the research library expendi-ture for journals has increased by 273% during the same period. These rapidly rising journal prices have thus battered the ability of researchers, libraries and institutions to access scientific publications. Global research article output is estimated to be around two million articles per year in around 18,000 journals (of which 8000 are indexed in ISI)and constitute an US$ 15 billion market. Researchers expect that the administration will subscribe to all journals that are available; librarians like to give access to all journals for the researchers,while the publishers would like to continuously increase their profit. With the current library budget of IISc at $ 2.4 million dollars per year, the efforts to contain bloating library budgets ultimately seem to rest on the director of the institute, who is unable to alter the dynamics because of resistance from every quarter. 

Thus many proponents of open access feel that scientists should publish only in open access journals. On the other hand, researchers in any prestigious institute like to publish their research in prestigious journals, which need not be open access. This is because publishing in a particular journal adds value not only to the journal but also to the publication. A publication in the top journal is likely to further the career of a scientist and he/she is not worried about whether the journal is open access or not. Therefore, a researcher, especially from South Asia/India, is likely to publish in a journal that is considered prestigious internationally, rather than just open access. Here is where an open access institutional repository will help. Almost all publishers (> 95%) of journals allow archiving a postprint of the research paper after peer review in an institutional repository. 

At the national centre of science information (NCSI) at IISc, we have the largest repository of research papers in India. Currently, the repository contains more than 11,000 full text research papers published by IISc faculty. The repository also contains the abstracts of all the 28,000 research papers published by IISc in its last hundred years. Because search engines like google index the full text of a paper that are available in these open access repositories, researchers using these search engines land up in the repositories rather than at the publisher site. Based on my analysis of the number of accesses and downloads of papers from the eprint repository of IISc, I find that people who search from developed countries see the paper in the repository but download it from the publisher site (for which they will access because the institution pays for it) while people who search from developing countries see the paper in the repository and download the postprint version of the paper. 

While agencies like NIH (and some NITs in India) have mandated that all publications should be archived in a repository, I am not so sure whether a mandate will be so effective. Instead of a top down approach, scientists should become convinced that deposition of their research work in an institutional repository will help. Why should the scientist take 10-15 minutes of his/her (or their student's) time to make the deposition in a repository? Other than the social issues aside, my analysis (not yet publishable) shows that the citations of a paper are bound to increase if it is available as open access in a repository. 

First the most prestigious journals in each field has to become open access. Then, the cost an author has to pay to publish in these journals should be reasonable. Till such time, scientists should be strongly encouraged to deposit their research papers in an open access repository. This would greatly impact the open access movement in India.

Friday, March 27, 2009

Why blog?

In a very detailed blog, Arunn discusses Indian scientists and science blogging. So, why do scientists blog? I believe in the Colgate's statement, "It is not a bad idea to get in the habit of writing down one’s thoughts.It saves one having to bother anyone else with them." Ben Franklin said, "Either write something worth reading or do something worth writing." Because I have not done anything for the latter, I attempted to do the former.

Seriously, I started blogging because several people asked me to write about factual data on publications, citations, higher education (numbers of graduates etc), pay scales etc.
I am fascinated by statistics. I can sit for hours looking at numbers. I used to teach it as well but that is another story. The key is that the factual data is always missing even in the era of information overload.

For example, one newspaper report mentions that an assistant professor at IIT gets Rs. 12,000 while a IIT B.Tech graduate gets Rs. 1 lakh per month while another newspaper report says that IIT faculty can expect an 200% hike in salary with the basic salary going up from 12,000 to 37,000. Another report published today that discusses the Goverdhan Mehta report says that the basic has gone up from 20,000 to 48,000. These reports are factually wrong. The current salary for an assistant professor is 12,000 basic + 124% of dearness pay and dearness allowance. These are being merged into the new scale and thus the salary, which is already at 28,000 is going up to 37,000 and is thus an increase of only around 30%.

Even though we are in science and supposed to be objective, many scientists have no idea about the number of publications or graduates we produce or what comparable universities in US produce. Lots of ideas are based on perceptions and examples are given based on exceptions. It is surprising for many scientists to know that we produce less than 1000 doctorates in engineering and 50% of them are from IIT/IISc (Many IITs now produce more doctorates per year than IISc). Many refuse to believe that the number of engineering doctorates awarded to Indians in India is lesser than the number of engineering doctorates awarded to Indians in US universities alone.

Some one in the administration told me we (in India) produce 30 lakh engineering graduates and 5000 research papers per year.
A B.Tech graduate who works in the National Knowledge Commission told me that the number of engineering graduates produced by India every year is around 30,000. I was even more surprised when this person had written a booklet on how to improve engineering education in India ! The actual numbers are 6 lakh engineering graduates and 30,000 research papers per year.

When I expressed shock at these kind of statements to the director, he encouraged me to write a proposal to the government for a project to gather actual numbers for IISc, IITs, India etc. The government was most forthcoming to fund this. While some of the documents I write are indeed funded by the government, I felt the reports would be better disseminated if this was available freely (on a blog!) because there is nothing secretive about any of this. Thus, most of the articles on this blog are on higher education.

The more obvious reason I blog is because I write most of the time, whether it is proposals, projects, papers or reports. Having started reading books in pdf format, more than 75% of my waking hours are spent on reading or writing. So, why not write in a blog, whenever one finds time?

New pay scales for IIT/IISc

The new pay scales for IIT/IISc based on the sixth pay commission were proposed by the committee headed by the former director of IISc, Goverdhan Mehta. Here is the report that discusses the new pay scales for IIT, IISc, NIT, IISER and IIM. Please note that these are only the recommendations and they have to be approved by the government. Sometimes the government approves the recommendations (like the UGC Chhada committee) in toto for the salaries of assistant professor, associate professor and professor.

The basic pay for an assistant professor, associate professor and professor will be Rs. 38,000, 53,000 and 59,000, respectively. Institute professor will receive Rs. 75,000. For those familiar with the Sixth pay commission terminologies, an assistant professor will be in pay band 3 for the first three years and then moved to pay band 4 (PB-4). This is due to an excellent suggestion that was originally mooted by the director of IISc. 

For those unfamiliar with the terms in the report, IIT, IISc, IIM follow the four tier system (Lecturer, Assistant professor, Associate Professor, Professor) while NITs follow the three tier system (Lecturer, Assistant professor and Professor). An associate professor from the four tier system draws the same salary as that of a professor in the three tier system. 

There is also a higher scale than professor in IIT/IISc/IIM, which was termed as professor of eminence, superscale etc. Now it is called Institute professor. MHRD had previously regulated that only 10% of the faculty can belong to this cadre, but Mehta's committee recommends that only 3% be given this scale.

The increments proposed by the committee are that 25% of faculty will get 4% increment every year while the rest will get 3% increment every year. Though many faculty in IIT/IISc were expecting better allowances than that of UGC, all other allowances are the same as that of proposed by UGC. The only exception is that Rs. 4 lakhs will be given every three years as professional development fund that can be used for contigencies, travel and research expenses. Thus all allowances mentioned in my previous post (which were based on UGC) stand unaltered.

Thursday, March 26, 2009

Graduate studies

My colleague in IIT-M, Niket Kaisare, talks about GATE, GRE and the flight to US and makes an interesting point,

We need to get the word out that doing a Masters (and even a PhD) at IIT/IISc is a viable option. Just to give an example, working with me for MS allowed one of my students to secure admission in a top-20 US university, which he otherwise would not have gotten based on his undergraduate degree alone. More of “Flight to US” will be dealt with in a follow-up post.
In fact, after my undergraduation, I could not get admission to a top 20 school in USA (I did get a few in 30-50) and, therefore, I joined IIT-M and then did my doctorate in a top 20 school in US. However, the situation has changed in that many middle class and upper middle class students who get admission in the top 20-30 schools in USA without financial aid still leave the country after getting loans. This was not common when I studied.

I am sure that all students joining the program in IIT/IISc know that getting financial aid in a top ten school requires a degree from IIT/IISc. Of the 32 master's students that I have graduated, 11 of them were interested in going abroad and all of them went to the top 20 schools in USA with financial aid. All of them published research papers with me and used it as a basis to get into the top schools in the US. Thus, IISc (or the IITs) has not been facing any problems in getting students for M.E/M.Tech programs. Our cut off is all India rank of 25 for admission in the M.E. progam in chemical engineering.

The problem is why these students do not continue to do a doctorate in India. Institute like IISc benefit greatly from a doctoral student than from a master's student. My colleagues often ask me whether I advise a master's student to pursue a doctorate in IISc itself and not go abroad. I do not.  I think facilities are now comparable and the quality of *some* faculty in IISc are comparable to that of the US. I, however, tell them the advantages of doing a doctorate in IISc (faster completion because they are already familiar with the experiments etc., closer interaction) and USA (more money, more independence from families (especially for girls)) and then ask them to decide.

GATE and CGPA

My colleague in IIT-M, Niket Kaisare, talks about GATE, GRE and the flight to US. He says, "If someone is incapable of securing more than a bare minimum threshold (in GATE), s/he is likely to struggle in the grad school." This was my point in my earlier post regarding GATE. If someone is unable to score 15% in GATE (50th percentile in chemical engineering paper), should he/she be interviewed for the Ph.D program at all? He remarks, "In fact, in the students I have encountered, there seems no correlation between GATE score and performance in the courses." I do not what was the data set and how good was the correlation. I had taken the CGPA of the masters students in the department for ten years and tried to correlate it against GATE percentile. The graph below is based on the CGPA of masters students during the period 2001-2006 (six year period) with their GATE percentile.



There was no correlation between CGPA and with students scoring above 95 percentile in GATE. However, when I divided the students into three groups: percentile less than 85, percentile between 85 and 95 and percentile above 95, the CGPA had a strong correlation with GATE percentile. For example, there was no student who had a CGPA of more than 5.5 but with a GATE percentile of less than 84.5. There was only one student who had a CGPA of more than 6.5 but with a GATE percentile of less than 94.5. Also, there was only one student who secured a CGPA of less than 5.5 but who had a GATE percentile of more than 95.

Sunday, March 22, 2009

Pay scales in IISc

There has been a lot of discussion in the comments section of my previous post on the pay actually obtained by professors in IIT/IISc. There has been confusion because the website of IISc and some other places advertise only the basic pay.

However, the scale without mentioning the allowances and other benefits is misleading. Please note the new sixth pay commission pay has not yet been fixed for faculty in IIT/IISc but it can be assured that it will not be less than that of UGC. Based on the document provided by IIT-M for pay fixation and by UGC, here are the details that a professor in IISc actually earns,

Basic + Grade Pay : 57,000 (52,000 for associate professor; 38,000 for assistant professor)
Dearness allowance (DA) as on 1 March 2009 (22%) : 11,000
House Rent Allowance, HRA (30% of B+GP+DA): 22,000
Travelling Allowance (TA): 3,200 + 15% of DA
Academic Allowance : 1,500
Telephone : 1,500

All the above are provided without submission of bills. HRA will be paid to employees who do not stay in quarters provided by IISc/IIT. It does not matter whether the employee is actually staying in an own house or paying rent. The travelling allowance will be paid to all employees even if they stay in the quarters.

The total of the above comes to around 90,000 for a professor and 60,000 for an assistant professor. 10% of the basic + GP is contributed by the institute and 10% is deducted from the salary for future retirement benefits. Thus, the pretax income is around 85000 and 55000 for a professor and assistant professor, respectively. The Dearness allowance increases based on inflation and increases by 6-7 percent every year. In addition, there is an increment of 3 percent every year.

In addition, they are benefits that are provided only if it is claimed. This includes children education allowance, subsidy for single girl child, Leave travel concession for the whole family every year,medical benefits to the family, travel money for attending conferences etc. If one counts all this, the cost to the institution (CTC) is around Rs. 15 lakhs per year for each professor. In addition, one can get fellowships that vary from Rs 60,000 to Rs 6 lakhs per year from DST/DBT, depending on the fellowships. In addition, if one is a fellow of two academies or has won the Bhatnagar award, one would get Rs. 1.8 lakh per year till retirement. Roughly 30% of the teaching faculty in IISc get these.

The amount obtained by consultancy also varies from faculty to faculty with some faculty getting close to Rs 20 lakhs per year. For consultancy, the faculty receives 60% of the amount while the institute takes 40% as overheads. This may be revised in the sixth pay commission.

Further, each faculty is given vacation for three months every year where he/she can travel/work anywhere and earn whatever money.

It is best to stick with CTC for comparison. One becomes a professor after 10-12 years after joining with a postdoc. This translates to 12-14 years after obtaining a Ph.D. How does this compare to pay scales in the rest of the government and industry?

Currently, professors are in the S-29 scale of 18,400- . After 10 years as professor, one can become a professor of eminence at the S-31 scale of 22,400-. Above that is the director scale (S-33) of 26,000 (fixed), which is equivalent to the secretary scale. This is the highest scale in IIT/IISc. Above this, there is only scale in the government, which is the cabinet secretary scale (S-34) of 30,000. This is the scale that a director general of CSIR, DST/DBT secretary (all technical positions) gets and is the highest in the government. Thus, a regular professor in IIT/IISc is three scales below the highest scale offered in the government.

In IITs (unlike the US), the salary for a professor in humanities or sciences or engineering is the same. This makes comparison of pay scales to industry difficult because the pay differential between these branches of study is considerable in industry. Based on this, I presume the CTC for a doctorate after 14 years of service would be around Rs. 25-30 lakhs per annum.

There is one more component that I should mention. The retirement age for all purposes is now 70. However, if I retire (hopefully !) from IISc in 2019 after 20 years of service at the age of 50, I will get pension for the rest of my life at 50% of the basic that I get at 2019. The pension will increase with increase in DA. Of course, this will not be applicable if one works in a private industry.

For the current employees of IIT/IISc: Arrears and the pay fixation in the new pay scale can be calculated from this document made by Dr. Ramesh of IIT-M. I know him when both of us were involved in GATE administration and software development. I can vouch for the thoroughness and correctness of his software !

Wednesday, March 18, 2009

Distribution of Marks in GATE

My last post discussed the actual marks obtained in GATE 2009 and the corresponding percentiles for several branches of engineering. My colleague, Abi, collected the data for metallurgy and posted it. Here it is (percentile-marks)

94.44-68.67
90.66-63
84.72-56.33
81.57-53.33
74.87-46

These marks are generally higher at every percentile level compared to chemical engineering, mechanical engineering, electronics and civil engineering. Let us look at another major branch, electrical engineering,

88-20
92-22
94-25
95-26
96-27
98-34
99-38
99.8-48
99.9-55

The difference in the distribution of marks is obvious. EE is much more closely clustered than Chemical and mechanical, which are closely clustered than metallurgy. A mistake in one 2 mark question will set you back by 2.66 marks from the candidate who answered it correctly. Thus a large distribution of marks is desirable. Setting a paper like EE where scoring 22 marks lands you in the 92nd percentile is not good for separating the good from the excellent students.

Both my colleagues who set the EE paper and MT paper, respectively, are good friends of mine but let me go ahead and congratulate the colleague who set the MT paper and achieved a good distribution of marks.

Monday, March 16, 2009

Marks and GATE

Subsequent to my last post, I received a few emails inquiring about the actual marks obtained by the candidates and the corresponding GATE percentile. The GATE 2009 results were announced yesterday and the following are the percentile with the corresponding marks in chemical engineering.

75-22;
85-31;
90-36;
95-45;
97-49;
99-66;
99.9-73;
99.99-79.

Suppose you think this is an anomaly, let us look at major branches,

Civil engineering

72-22
80-26
88-31
92-35
95-41
98-49
99.9-60

Computer Science
80-21
85-24
87-25
91-29
98-44
99.9-62
99.99-73

Electronics and Communication

70-19
86-27
89-29
93-35
95-37
97-42
98-45
99-50
99.9-74

Mechanical Engineering

72-18
85-26
90-30
95-38
96-41
98-49
99.8-69
99.9-76

The cutoff for the M.E program in chemical engineering in IISc is higher than 99 percentile i.e., the candidate should have scored more than 66 out of 100 in the GATE exam. However, if the candidate has scored less than 1/3 of this (i.e.,22), (s)he is eligible for the Ph.D program !

Thursday, March 12, 2009

Admission without GATE

As I written earlier, GATE has been officially removed as a requirement for applying for admission to the Ph.D program in engineering. This was done because the number of doctorates in engineering from IISc is roughly around 60, while that from IIT-B is nearly 120. There is a hope that this modification will lead to a surge of doctoral admissions in IISc.Thus the admission form reads,

"Candidates with BE/ B Tech/ M Sc or equivalent degree who may not have qualified in any of the above mentioned National Entrance Tests will also be considered for the Ph D program in Engineering."

The same was considered for admission to Ph.D in the science departments but that was shot down by the faculty of science. Therefore, candidates applying for Ph.D in the science departments have to still qualify in one of the national exams.

All my friends are all gaga about this. I was one of the very few people who protested against this move and even wrote a letter that was discussed in the Senate meeting. Though the advertisement reads that it is for Ph.D in engineering, candidates admitted to the program can quit after a master's degree after receiving scholarship for two years. Therefore, it is against the MHRD regulations of giving scholarships to only those master students who have passed atleast one national level exam.

Secondly, many of my colleagues that GATE is a difficult hoop while GRE is not and many bright students do not write GATE. I do not think GATE is more tedious than GRE but it requires preparation on the subject. You can not throw out an exam because it requires preparation !

Instead of removing GATE, I have been requesting that GATE be modified i.e., it should include section A, which is on english comprehension, english writing, analytical ability etc. (similar to GRE) and section B on the actual subject in engineering/science.

Sometime ago, Arunn Narasimhan wrote on GATE and how they may prevent good students from getting into graduate programs in India. Abi, in quoting him, says that "our institutions must take a step back from using entrance exams as the sole first-level filter, and try to incorporate other metrics in the selection process: performance in their university exams, consistency in academic performance, recommendation letters, etc." While I agree that entrance exams like GATE should not a sole filter, it should be a filter. The reason that we have entrance exams and do not trust marks in the university exams because they vary widely depending on the type and grade of universities. A student with a CGPA of 7 from IIT-M may be better than a student with a CGPA of 8.5 from one of the private deemed universities. There is simply no way one can compare these candidates unless you have an exam, which should act as a filter. The differences in the marks obtained in GATE by candidates who are in the 99.9 and 99 percentile may not be significant and thus admission should not be given or denied to one on this. However, this candidate would have scored much higher than a candidate who is in the 50 percentile.

Thus though I believe that scoring marks in either GATE or university exams do not translate directly to research skills, a minimum cutoff in both is necessary. Therefore, I had suggested that the admission be based on writing GATE but the cutoff for an interview call be reduced to say, 50 percentile. Qualification in GATE requires a percentile of more than 75 but now that GATE gives the actual marks as well as the percentile for all candidates who write the exam, we can call candidates who have not qualified but who atleast prepared and wrote the exam reasonably well.

In most cases, bright students are not opting to Ph.D in IIT/IISc because of the monetary benefits that is provided by the US universities. One is not sure to see a reversal of trend just because GATE is removed. Maybe I am wrong and hope that I am. After all, I also wish that a student who did not write GATE, got admission in MIT but actually wanted to join IISc now joins IISc in preference to MIT.