Monday, April 30, 2007

Women in engineering: What about international comparisons?

Vivek has done a great job of collecting a lot of interesting data women in engineering for many countries (but mostly in the West). They all point to a low figure of 20 percent for the ratio of women among engineers.

A first look at these data might lead one to despair; there seems to be a grim sort of determinism at play: as development progresses and India too becomes more and more like the West, this 20 percent may be the "attractor" towards which India too will move inexorably. But, this grim view is unnecessary. There are big differences between the West and India, and the path we take need not take us to the same destination that the West has reached now. Let's look at some of these differences. In discussing these differences, I don't mean to underplay the problems women face in the West, but to point out that since the underlying causes are different, the solutions must be too.

* * *

1. In India, engineering courses (along with medicine, law and accounting where the overall numbers are small) are highly sought after; so much so that many students are willing to attend some no name, tin pot college to study computer science (and pay good money too!), instead of attending a good, reputed college to study science/arts/commerce. In the West, on the other hand, there are many, many options for an incoming freshman; not getting into engineering does not handicap a student when it comes to the job market, where there are plenty of opportunities for non-engineering majors.

This difference is reflected in the debates as well. In the US, for example, people talk about what the institutions (and their people and practices) must do to "attract" more women; alternatively, they talk about how the current practices drive women students away. In India, on the other hand, we are (at least I am!) talking about the ways in which our institutions are keeping bright and very interested women out, and why they must change their practices.

2. An 18-year old in the West is expected to choose his/her career path; others may offer advice and guidance, but the responsibility for the final choice is largely on the individual. In India, on the other hand, college related decisions have significant -- and in many cases, decisive -- inputs from the parents (and the rest of the family). [When a student rebels against the choices made for him/her, it is considered sufficiently unusual for people to talk about it!]

3. In the West too, societal pressure works to deter women from entering S&E programs in college. But this pressure is largely from peers ("math is uncool", "physics is for nerds", etc); the pressure from parents, teachers and other significant elders is subtle and covert. In India, I don't know of detailed studies, but my guess is that the pressure comes mainly from the parents and other family members. Vivek himself identified quite a few problems. In addition, Govar points to parents' concerns about overqualification that makes it difficult for them to find someone even more qualified to marry their daughters off. On a different note, Gopal points to parents' concerns about their daughters' hostel stay (since many of our top colleges have their own fully residential campuses), etc. Happily for us, many of these society- and culture-imposed problems are on the losing side. At least in cities. Increasing participation of women in computer science and electronics programs is indicative of this trend. Will other fields -- metallurgical engineering! -- to catch this trend?

4. The differences in the underlying societal causes for women's underrepresentation in engineering imply that remedies are also different. Thus, in the West, one tries to figure out how the 'peer effect' can be countered; in India, on the other hand, one looks at how parents' attitudes can be changed. However, in both places, there are problems at the institutional level too. In the US, Zuska, Janet Stemwedel and Sean Carroll devote considerable space for discussing this issue. It is this kind of discussion we are engaged in here.

5. Finally, institutions in the West (at least some of them in the US, about which I have better knowledge), are acutely aware of their inability to attract women into S&E careers. And some of them are doing something about it. Just a couple of weeks ago, I linked to this NYTimes piece about the efforts of top computer science departments to attract more women to their undergrad programs.

Moving emphasis away from programming proficiency was a key to the success of programs Dr. Blum and her colleagues at Carnegie Mellon instituted to draw more women into computer science. At one time, she said, admission to the program depended on high overall achievement and programming experience. The criteria now, she said, are high overall achievement and broad interests, diverse perspectives and whether applicants seem to have potential to be future leaders.

“In this more balanced environment, the men and women were more alike than different,” she said. “Some women are hackers and some men are hackers, and some women love applications and some men love applications.”

With the changes at Carnegie Mellon, women now make up almost 40 percent of computer science enrollees, up from 8 percent, Dr. Blum said.

In the comments section of his own post, Bill, who is from the very department mentioned in the above quote (and who is also an IIT alumnus), has this to say:

I was at CMU where they have become concerned by the low number of women entering CS (did you see the Times article just over a week ago?) and are in the middle of seeing how they can modify the situation. Their tweaks over the last five years have slowly but steadily increased the number of women applying, and also, getting admitted. Of course their entrance procedures are very different but my point is does IIT even recognise this as a problem? I, for one, see no evidence of that.

In India, Dilip talks about BITS-Pilani going through an exercise in soul-searching, because women's enrollment dropped precipitously when that institution introduced an entrance exam in 2004. A similar point was raised by Tobacconist on Vivek's original post. Here's the relevant part from Dilip's post:

  • Over the years, BITS admitted an ever-increasing number of girls, still on school-leaving marks. When I visited the campus in early 2004, the sex ratio was more like 2:3. (Still f:m). I believe part of that increase is explained by hugely expanded hostel space for girls.
  • In 2004, BITS introduced their new online admissions test, BITSAT. The immediate fallout was a drastic drop in the number of girls admitted. I believe the ratio among 2004 entrants was something like 1:10, uncomfortably like in my time.
  • This drop is of serious concern to BITS authorities, who I believe are examining (among other things) if the test itself has or encourages gender bias.

Let me get back to Bill's question here: Do our top engineering institutions think of women's underrepresentation as a problem?


  1. Anonymous said...

    I do not know how many women responded to the discussion happening on your blog (Abi, may be *if you know* you can let us all know), but I would be interested in finding that out as I think somehow this is related to what is at the core of the discussion.

    "because women's enrollment dropped precipitously when that institution introduced an entrance exam in 2004."

    If I understand the arguments correctly, entrance exams *somehow* prevent women from entering some elite institutions and if it were based on more 'broader' aspects of a student's personality, there is a better chance for women to get to the institutes. Coaching classes are pretty much on the top of the evil reasons for this state of affairs, hence working backwards, if only JEE exams were *not so hard*, may be there would be no need for coaching classes, hence.. But there were no coaching classes so to speak some years back, and I believe the number of women entering IITs in 1980s is no higher than it is today, how do we understand this? May be some underlying societal stigma/pressures are the reason? if so, how different it is today?

    Should the institutes or centers of excellence be more gender friendly? I agree with it 100%. But I think what makes an Institute/university is still *people* and as long as they do not change their attitude, there may be little progress in that direction. There are many instances in the posts where *institutes* are referred to as some abstract entities, but surely they are not! They consist of people (predominantly male) making decisions. I believe it is not "A exam (like JEE)" that is biased, but the reasons are deeper than that. Pinning the gender bias in current education system on a single entrance exam is, in my opinion, too simplistic.

    As an aside:
    Coming back to what I asked at the outset, how many women read/respond to your blog? How many women *have* blogs? how many women have *time* to write blogs? To all males who regularly respond to this and other blogs: do your significant others, if they are women, also surf or have a presence in blog world?
    if not, why is it so?

    another discussion may be..


  2. Anonymous said...

    Why don't you discuss the situation at IISc also? Why are there so few women in the M.E., M.S./Ph.D. programmes in Engineering? The percentage of women taking/qualifying in GATE appears to be much lower than the percentage coming out from the undergraduate colleges. Discplines like CS, EEE, ECE, IT fare better than Mechanical Sciences.

    Some of us in IIT(s) do recognize the problem and try to address it in different ways. We try to give some consideration for women in the M.S./PhD interviews, try to take them in projects etc. But unless the number of applicants goes up nothing significant can be achieved. In the recent interview session we had in the department for the PhD/MS programme out of 65 applicants for 20 positions only two were women. It is an even more serious problem for mechanical sciences.

    The percentage of women in IITs has gone up marginally from the late 1990s than before the coaching class mania and I say that the JEE papers were tougher in the mid-late 1980s than in the last 10 years.


  3. Ritwik said...


    Point 3 about peer pressure explains the relative unpopularity of science and engineering among all US students, but it doesn't explain anything about the gender difference. If math is uncool and physics is for nerds, does it mean that more American boys think it is ok to be "uncool" or "nerdy" than American girls?

    About point 5, even if the enrollment of women in CMU's CS program increased from 8% to 40% due to a re-tweaking of criteria, are we to conclude that that the new criteria are necessarily better. "high overall achievement and programming experience" seems like a natural consideration for a CS program. " high overall achievement and broad interests, diverse perspectives and whether applicants seem to have potential to be future leaders " can easily be the admission criteria of the Kennedy School of Government.

    Anyway, all that aside, my views on the entire "women in engineering" debate are rather politically incorrect. I speak from various readings and my own experience.

    In my school, in 11th and 12th, the science section consisted of 50 people. The gender ratios were similiar in the science and commerce sections. It seems foolish to generalise from a sample space of 50, but the peculiar nature of an elite school in Gujarat made the group of 50 wonderfully controlled in all other respects. Gujarat is one of the more equal socities in terms of gender relations. The students in my school all came from affluent and educated families where it could be reasonably presumed that the social outlook on a girl's education was not very different from a boy's. Medicine as a career is considered more respectable than engineering, and the engineering colleges in the state admit students on the basis of 12th std board marks, which means that no one can afford to ignore the boards. People are not that enthusiastic about the IITs and none of us went to any coaching classes as there were no coaching classes worth going to. None of us re-took the examination either. With all these controls, it seems reasonable to presume that if there was a difference in the relative performance of girls and boys in the JEE (and related entrance examinations) it would reveal something either about the test or about the test-takers.

    The cliches held true. The girls outperformed the boys in the boards. Their passing percentage was better. They were more numerous among the toppers too. Not a single one of them, however, even cleared the screening. Abi would conclude that this shows something about the test. I, however, think that this says something about the test-takers.

    Since we enterd class 11, I noticed some revealing phenomena. Fewer girls than boys, even among the really smart ones, truly understood the mole concept or ionic equilibria. Fewer of them were comfortable with normal reactions, free body diagrams or pseudo-forces. Fewer of them understod the implications of probability theory. And through those two years as well as 4 years of engineering (not at an IIT), I can safely say that I haven't met a single girl who truly understood rotational motion or permutations and combinations. (Wait, I met three who did, but they were math olympiaders. And 3 out of 30) Not that too many boys understood the above things either, but the percentages were much higher. In general, I've noticed that when it comes to certain topics that I believe involve a rather intuitive understanding of certain physical and mathematical concepts - girls bettered boys when the questions were of average difficulty and predictable (Boards), but fell way behind when questions involved picking up a concept and applying it to a hitherto unseen type of problem (JEE).

    Study after study has shown that men on an average perform slighlty better on mathematical and spatial tests, while women have better verbal ability and learn to speak faster and at similiar ages in their growth as children know far more words than men. Study after study has also shown that the fat tail phenomenon of the Bell's curve is fatter in the case of men. Yet, I see the politically correct lobby launch spears into anyone who claims that men may be more mathematically talented than woman, yet no one feels the need to question the validity of studies that show differences in favour of women in verbal abilities.

    The 1:10 ratio at IITs surely doesn't reflect true natural differences. The reasons for a ratio as skewed as that are largely societal, and I tend to go with Vivek and Falstaff on this. I have gathered though, that even when the societal differences are removed, the ratio will never approach 1:1. The JEE is indeed biased Abi, but it is biased against people who are not very comfortable with rotational motion, P & C and suchlike. That a higher proportion of these people are women, is in my belief, just one of the facts of nature.

    As an aside, if you feel that the IITs need to re-think their selection policies to remove the inherent bias against women, shouldn't you also feel that St Stephens needs to re-think its admission policies to ensure that the sex ratio in the English Lit class is not as lop-sided as it is (F>M), or that the NIFTs need to re-design their entrance in such a way that they do not end up with classes that have less than 10% men?

    I know that you will find my views disgustingly chauvinistic. I do hope that you'll respond, though.

  4. Anonymous said...

    Zen babu:

    1. The differences might be larger intra-gender than inter-gender.

    2. There might be a difference but how large is the effect size?

    3. The small differences inter-gender(at the extreme ends or tails) may not be what drives success in many fields (including post Class 12 engineering and math).

    Therefore, while it is tempting to ascribe this difference in mathematical and spatial ability (even give the different distributions and effect sizes) to boys and girls' relative success in engineering exams (or at least some part of it), it may not be entirely valid.

    Second, what is your definition of "understanding" probability or rotational motion? Self-report? Doing well on a test that tests these concepts? If self-report, confidence alone would explain the results: the girls you might have met might have been able to "understand" rotational motion or permutations and combinations or probability as well as the guys but were less confident in reporting their "understanding" relative to the boys. See my comment somehwere on the differences in confidence (at the SAME level of math performance) between girls and boys. (I've actually found this anecdotally to be the case and have even personally experienced it, but alas, I would die a million statistical deaths before admitting it into any testimony on gender differences. However the published research results should be a useful starting point).



  5. Ritwik said...


    I read your copmment after posting mine, and it made perfect sense to me - the difference in confidence at the same level of performance opened an aspect of the issue that I'd previously never explored.


    1. I do understand the concept of intra-gender differences and inter-gender differences. A small inter-gender difference coupled with the fat tail phenomenon probably implies that at the "very high" level of skill, the inter-gender difference may become more pronounced. On the SAT math section, men outscore women by 30 points on an avg. This is a small inter-gender difference as compared to the larger intra-gender one. The number of women who score more than 700, however, as a percentage of all candidates who score more than 700 is significantly lower than the number of women as a percentage of test-takers. This is roughly what I meant.

    2. As for success in any field, forget abt inter-gender differences, very often IQ itself doesn't necessarily drive success. Haven't we all seen less intelligent men and women becoming more successful than more intelligent ones simply because they persevered, etc. Here, however, we are not considering "success in field" - we're simply comparing different people's abilities to maximise performance in an entrance test, under the assumption that a high level of mathematical/analytical skill is a necessary (though not sufficient) condition to do well in the JEE or comparable exams.

    As far as my assertions about "understanding"or "not understanding" some concepts go, I have only anecdotal evidence (my own understanding of who among my peers understood what) - no self reporting - which was later corroborated in an exam that I believe genuinely tests these concepts (JEE). I know that the sample size was too small, however for reasons that I have mentioned above, I also think that the sample was wonderfully controlled in all respects.

    In any case, I too support a statistical, non-anecdotal approach to understanding these things. The trouble is, in the light to two opposing statistical evidences, one is more likely to believe that evidence which strengthens the results from one's anecdotal experience.

    Thanks for replying, though. I'm far from settled in my views on this and the study that you quoted was illuminating.

  6. Anonymous said...

    Zen: that remark on "understanding" was related to your comment, not aimed at whether you undrestood the concepts or not. Any offense not intended and all that.

    Any cites on the inter-gender differences? Would love to read more on that. I think the lowe rpercentage on SAT is more likely attributable to the confidence hypothesis than innate intelligence, but I don't know enough about it to stte that with certainty.



  7. Abi said...

    TR: Your entire response has been made into a post!

    Vani: I really haven't kept track of the comments by women, but quite a few have come by to comment. But you are right, I too think that comments by men have outnumbered those by women (barring anonymous commenters, of course). Please tell us: why do you think this is necessary?

    As for the core argument, an elite institution need not even look at "broader aspects of a student's personality" to result in a larger number of women. It is enough if they took into account their high school marks!

    Now, coaching centres: The need for coaching existed even during the eighties. I myself took Brilliant Tutorial's classroom coaching for six months. In addition, Brilliant also offered postal coaching. But it had a far smaller market share than the biggest coaching institution of that era: Bombay-based Agarwals Centre. Very few of my fellow-batchmates wrote JEE without the help of any coaching (I would put the figure at something like 20 out of 250). We were in awe of these souls!

    The far smaller number of coaching centres during the eighties might have been due to the small size of the target market: IIRC, just about 50,000 students took the exam in 1981, and coaching during Class XI was unheard of. These days, the market is much, much bigger because (a) the number of JEE takers has gone upto 250,000 or more, and (b) coaching has percolated down to the ninth standard!

    Yes, Vani, people in institutions do play a role in shaping policies, and the fact that they are predominantly male may have something to do with under-representation of women in engineering (and science too, particularly at higher levels). Except that I do not know of overt acts of discrimination.

    This does not mean that there are indirect ways in which discrimination operates. I believe entrance exams (particularly JEE) has this bias. In asking for this bias to be removed, I'm certainly not saying the other problems in society are not important. They are. It's just that I prefer to concentrate on what things I know.

    Gopal: As a matter of policy, I normally don't comment on IISc-related issues, except when there's a public announcement. I will leave IISc to the outsiders to blog about!

    Having said that, entrance exam bias may yet exist at graduate level too. It's worth looking into.

    One more general comment: since you are in an IIT, I'm sure you know that great graduate students are so hard to come by that any institution would just love to hire demonstrably bright students -- both men and women. This is particularly so in engineering. We are always looking for bright students: if you have anyone that you would recommend highly, get in touch with us!

    Zen Babu: Yes, "math is nerdy" is a sentiment that affects girls far more than guys: "the difference seems to be the fierce social pressure on girls not to be nerds."

    The Carnegie Mellon experience and their special efforts to attract more women are "better" because it's what they wanted. [At the end of the article, the professor says something to the effect that they are still among the top universities. ] From my point of view, it's "better" because that's the kind of effort that I want to see happen here.

    You said "girls ... fell way behind when questions involved picking up a concept and applying it to a hitherto unseen type of problem (JEE)". Now, that's not how I would characterize JEE. If that exam was just about applying concepts to a new type of problem, we wouldn't be having this extended discussion. JEE is about very, very high level problems -- problems that you could not imagine doing if you had no external help. Some people say things are changing at JEE, but I haven't seen any evidence of that change in the final results.

    N! has already addressed the issue of what the tests really test for, and how it brings -- ostensibly inadvertently -- a set of cultural issues to the situation of test-taking (stereotype, anxiety, expectations, etc.). In your response, you said that your views are far from settled. So, from my side, I will offer some links: a bunch of posts on Cosmic Variance, and a couple of posts by Janet Stemwedel. If you wouldn't mind some blunt talk, try Zuska's blog.

    As for test scores, they are meant as a predictor for success in college. But each college also has to evaluate their predictive capability for themselves, since success in college requires many other ingredients as well: diligence, emotional stability, etc. Take a look at what is here:

    "One of my faculty colleagues ... did a study of whether admissions performance measures -primarily the math SAT--actually predicted the academic performance of students, not just as freshman but throughout their undergraduate careers. He did this differentially for men and women and got some surprising and very important results. He found that women outperform their predictions. That is, that women perform better as students than their math SAT scores would predict. The effective predictive gap is about 30 points."