Monday, April 30, 2007

Some interesting developments to watch for ...

Umesh Kumar Singh reports in ToI [Thanks to Chitta for the pointer through this comment]:

The Central Information Commission has directed IIT Kharagpur, which conducts the joint entrance exam for the engineering courses, to furnish all the details regarding cut-off marks and the procedure to reach to the cut-offs, the model answer sheet and so on by May 15 to a candidate, Eklavya's parents.

Note the candidate's name!

* * *

The IITs (and other institutions, too!) have always hoarded the huge wealth of information that's available with them regarding the students' background, their performance, number of previous attempts, and so on. The stuff they have is an ideal playground for freakonomics-types to show off their skills. [I have ranted about this before.] But even with the meager information put out in the public domain by the IITs, we can look forward to some data that are likely to throw light on some of the questions we have been grappling with in the past week. Here's the scenario:

While JEE allowed three attempts until last year, it allows only two attempts from this year on. Thus, one should expect a somewhat better rate of success for the first timers. Moreover, it will also allow one to test Falstaff's model [fn 1], in which men's advantage accrues to them because they keep taking the exam again (and again), whereas women essentially stop with just one attempt [Is there any empirical evidence for this? We should ask the IITs. Sigh!]. If the model's view of the situation is right, it would imply that the fraction of women in the list of successful candidates will be greater as well.

We will wait and watch. The next few months promise to be quite interesting ...

* * *

In Falstaff's model, the agents are identified only as X and Y. I have 'translated' the terminology to suit my purpose here.


  1. Anonymous said...

    RTI is cool! I didn't like the fact that JEE was like a black box; it definitely needs to be more transparent.

    Abi, I don't quite like the idea of setting a cap on the number of attempts a person can make. In the present scenario, if someone wants more first timers, one could say that in case of a tie, the candidate with the fewer number of attempts gets preference.

    I don't see any problem why someone shouldn't be allowed to take the entrance exam 10 times. If you want to discourage coaching, do something else. Restricting the number of attempts is simply going to increase the pressure in the minds of people who are really desperate to get into IITs.

  2. Anonymous said...


    I posted this on Vivek's blog as well, you might be interested in this, if you haven't already read it.

    Since I took it off google, not sure of the credibility of this article but it cites some interesting research. Some parts that I found aprticularly relevant:

    Steven Spencer and Claude Steele (1994) suggest that when female
    students are frustrated by the difficulty of math problems, they
    associate this frustration with the belief that they as women are not supposed to be able to do math. This leads to anxiety, which impairs performance. To test this hypothesis. Spencer and Steele (1994) performed a number of experiments utilizing males and females who were highly skilled at mathematics and highly motivated to perform well. As was predicted, females scored as well as males on a test of moderate
    difficulty, but underperformed relative to males when the test was more difficult. This is in keeping with earlier findings.

    This same difficult test was given to another group of students
    with one minor change in the procedure: some students were told that the test was gender-fair (i.e. females performed as well as males on the test), while others were told that the test differentiated between males
    and females. When females believed that they could do as well on the test as males, they did so. There were no significant differences between males' and females' performance in this condition. Females who expected
    the test to be difficult for females showed the usual pattern of
    underperforming relative to males.

    Some other interesting points are:

    1. Women are less confident than men about their mathematical ability than men even at the same level of performance.

    2. Parents attribute men's success to innate ability and women's to hard work.

    3. Taking up hard math courses (beyond the minimum required for graduation) is correlated more with perceived ability than with actual performance. (My question: could this apply for difficult exams as well?)

    4. Women have lower confidence in their perceived ability to performn on a math test.

    5. Women are less willing to approach new material.

    To the extent that the JEE is the most difficult math exam and to the extent that it tests new material, to the extent that the communication might be that the JEE currently might be more about "innate ability" rather than "hard work" and to the extent that women might be less confident abotu their chances of success even at the same math performance as men, any test like the JEE might be biased.

    I suggested to Vivek that perhaps the IITs should do an intervention like the one mentioned above and see if there are any differences.