Monday, February 18, 2008

Prof P. C. Mahalanobis (June 29, 1893 — June 28, 1972)

By Kuldip Dhiman

HE played in the streets of Calcutta with physicists Satyendra Nath Bose and Meghnad Saha, and he, too, would have gone to become a great physicist if chance had not played its part.

When he was professor of physics at the Presidency College, Calcutta, Prasanta Chandra Mahalanobis was approached by Sir Brajendra Nath Seal, then university Professor of Philosophy, who asked the young physicist to help him with enquiring into the examination system by using statistical methods. Professor Mahalanobis enjoyed the work so much that he gave up physics and fell in love with facts, figures, graphs and charts.

Before he arrived on the scene, statistics was practically an unknown profession in India. None of the Indian universities offered statistics as a subject, and we did not even have a body of professional statisticians. He realised that progress can only be made if we have the right facts and figures first, only then it is possible to plan anything.

In order to train new statisticians, the Professor, as he was referred to by his students, founded the Indian Statistical Institute (ISI) at Calcutta in 1931. He also started publishing Sankhya: The Indian Journal of Statistics in 1933 with the help of his associates R. C. Bose, S. N. Roy, and C. R. Rao. Mahalanobis nurtured the journal and edited it all his life as his child. In the inaugural issue of Sankhya he wrote: "We shall try to keep before us this comprehensive idea of the scope of statistics. We are convinced that statistics represents a fundamental method of analysis of data in the mass which is applicable to any science of observation, and we feel that it is desirable to emphasise this essential unity in the methodology of statistics."

One of his first exercises in using statistics for field-applications was to employ the method of random sampling to estimate acreage and yield of crops in a large region. The new technique was applied for the first time on the jute crop in the province of Bengal in 1937. Post-graduate courses in statistics were opened for the first time in India in the Calcutta University in1941. Under his influence the Indian Science Congress started a separate section for statistics in 1945.

At the invitation of his old friend, Nehru, who had just occupied the chair of Prime Minister, Mahalanobis returned to India. He was made the Honourary Statistical Adviser to the Cabinet. He guided the Central Statistical Unit, and later the Central Statistical Organisation. The National Sample Survey was started in 1950 thanks mainly to his efforts.

Mahalanobis was the author of India's First Five-year Plan which was based on the two-sector growth model, and is called the 'Mahalanobis Model'. It showed that an increase in the allocation of investment in capital goods vis-a-vis consumer goods results in a correspondingly higher rate of savings ultimately accelerating the GNP growth. Not many approve of his model now, because it did not explore the possibilities of growth through trade expansion that proved to be the force behind the rise of miracle economies. The model also ignored the wage goods constraint and did not emphasise land reform and investment in agriculture. In the end, it was aborted within two years of its launch an India ran into foreign exchange crisis in 1957-58.

The Professor's errors in planning notwithstanding, his made three notable contributions: The formulation of the concept of optimum design survey in terms of the variance and the cost functions. He showed how pilot surveys could be profitably used to estimate the parameters of the variance and cost functions. And he proposed various techniques for assessment and control of errors, which were inevitably, creeping into the data.

The innovations that Mahalanobis made in the field of statistics were recognised the world over by professionals. "What at first most strongly attracted my admiration," observes R. A. Fisher, famous British statistician. "was that the Professor's work was not imitative . . . Imitative books are as common as dirt; and the work of striking originality is as rare now, as it was when I first read of the surveys of the jute crop in Bengal in which the Professor was trying out his new ideas."

While the American statistician and economist, H. Hotelling feels: "No technique of random sample has, so far as I can find, been developed in the United States or elsewhere, which can compare in accuracy with that described by Professor Mahalanobis."

No comments: