Monday, July 28, 2014

Beyond CSAT - aim to select the optimum group of civil servants and not the "best" individuals

I watched a discussion on Lok Sabha Tv for an hour about the aptitude test of UPSC which has been in the news recently - its need and its fairness. Neither the discussion nor anything I have seen in the paper talks about the fact that if there is a single test, any single outcome may seem unfair even if the test is not.

As the ratio of the number of candidates to the number of people selected increases to the levels in India, a thought experiment can tell us that if the test were repeated, examiners interchanged, the list of selected candidates can vary substantially.

A very large number of candidates not selected could have done better than the selected candidates on another day or in another mental state.

The scenario becomes far more complex if one tries to think of what would be an excellent group of candidates for the civil service. After all no one would select a soccer team by any method where all the selected players  could turn out to be goal keepers!

The concern about the mix of the outcome is very real. The discussion had references to states  complaining about the number of candidates selected from their state because of the changes in the exam system.

Suppose there was a  formula for the optimum selection of the group with factors like
  • Test result
    • overall
    • individual subject
  • hetergeneity
    • male/female
    • economic background
    • mother tongue
    • place of residence
    • social background
  • Affirmative action 
    • Replacement of reservations
  • New blood
    • people whose parents/relations have not been in any government service
Factors to be considered could be specified. The multiple objective functions could be specified. The number of objectives could be quite large. It will make no difference to the implementation of the algorithm. The source code of the algorithm could be published.

These could be tested against the past examinations to create the lists of who would have been selected had this process been in use.

I am certain if the outcome is fair to all stakeholders, the concern over unfairness of the exam and bias of examiners will decrease. But more significantly, we can have a group of administrators in whom the society as a whole has confidence and which as a group is likely to deliver better outcomes for the governance of the country.

No comments:

Post a Comment