Sukhatme vs Mahalanobis

There was a time when India was at the forefront of Statistical, in particular Sampling, theory and practice. Yet, it followed very foolish policies and preferred to rely on begging bowl diplomacy rather than develop its agricultural sector properly.

The following extract from an article on Prof. Sukhatme sheds light on the bureaucratic in-fighting which led to so deplorable an outcome.

The main effect of Prof. Sukhatme’s first project in ICAR, on the evaluation of the performance of the Etah Goat Breeding Project headed by Dr. Slater, as already mentioned earlier, was that the complexion of ICAR began to change gradually. Dr. Slater had left India to go back to his original post in UK and Vice-
Chairman of the ICAR did not want further scrutiny of other projects by Prof. Sukhatme from statistical angle lest other managers of the projects might leave. Prof. Sukhatme started organizing training courses and writing text-books. With World War II in full swing, the nature of statistical work further changed. 

Thus, Statisticians were only listened to if they said what the big nobs wanted them to say.

With Japanese close to Burma, the Government wanted to be kept informed continuously about the food position in the country, particularly because 1942 Bengal famine had claimed a lot of lives. Prof. Sukhatme had a difficult task to handle in rushing from one place to another and making enquiries into the stocks of food as well as initiating sample surveys to determine probable size of the crop in the field and writing his reports. This meant further expansion of his activities on war footing for which he was given all facilities including travel by air. Eventually this resulted in popularizing the use of random
sampling technique for estimating crop areas and yields. He could successfully demonstrate to people how with the use of this technique one can estimate the size of the crop in different provinces within a very short time of the harvest. What he changed was only the fields to be located for crop-cutting using random sampling methods, other details remaining as before. That was the only practical way to provide scientific basis to his inference. In the midst of this work, one fine evening in 1941 he was informed that his designation was changed from Statistician to Statistical Adviser and his salary scale was upgraded. 

However, with this success came the confrontation with Professor PC Mahalanobis. He viewed that Sukhatme’s approach was too costly, the only way of reducing the cost being the use of random sampling method evolved in UK and tried by him in states of Bengal and Bihar. Prof. Sukhatme disagreed and claimed that unless we have a field organization which can watch the crop as it grows, a small plot is bound to lead to biased estimates as per his results enumerated above. 

In fact the truth was that Prof. Mahalanobis wanted that that all statistical work in India should be centralized in his Indian Statistical
Institute at Calcutta regardless of whether it concerned agricultural, medical or other applied areas of research. This was neither acceptable to Prof. Sukhatme nor to the ICAR.
So rapid was the development of training courses in statistics for research workers, statistical services to aid them in their work as well as the spread of random sampling methods to estimate crop areas and
yields in the provinces in addition to international recognition that the Statistical Wing of the ICAR had virtually become an institute by itself. Prof. Sukhatme, therefore desired that, in keeping with the
international stature, the Statistical Branch should be upgraded to an Agricultural Statistics Research Institute. The Ministry of Agriculture was entirely agreeable to this proposal but was powerless in the face
of opposition from the above.

The then Minister of Agriculture, Shri K.M. Munshi sensed that so long as Prof. Sukhatme headed it, it was most unlikely to be approved. Prof. Sukhatme was advised, therefore, to leave India to accept a pending offer of the post of Chief, Statistics Branch, FAO, Rome earlier made to him by the late Mr. Norris E. Dodd, the then Director General of FAO. Prof. Sukhatme reluctantly took the decision to leave India, a sacrifice which led to the establishment of the Indian Agricultural Statistics Research Institute (IASRI) in due course of time. His departure was widely noticed in the press and the
general feeling expressed was that the departure of an agricultural statistician of his eminence was an irreparable loss to the progress of development of agricultural statistics in the country.

(PDF) Prof. Pandurang Vasudeo Sukhatme- A Statistician - Par Excellence. Available from: https://www.researchgate.net/publication/326549758_Prof_Pandurang_Vasudeo_Sukhatme-_A_Statistician_-_Par_Excellence [accessed Dec 23 2018].

By far the most important contribution of Dr. Sukhatme, while he was in the ICAR and in the wake of Bengal famine of 1943, was the rapid improvement which he brought about in the estimates of yield of
food crops (particularly wheat and paddy) in the country, which at that time were based on the old official procedure of annawari estimation and were, consequently, defective, unreliable and subject to unknown
margins of error. Dr. Sukhatme evolved the technique of stratified multistage random sampling for the conduct of surveys in agriculture in which random cuts of specified size are located in the randomly
selected fields growing the particular crop. During 1943 to 1946, he conducted crop-cutting sample surveys on wheat in Punjab, Sindh, Uttar Pradesh and North-West Frontier Province (NWFP). In this
experimentation, rectangular cut of size 33.0 ft × 16.5 ft was adopted as against smaller circular cuts of 2 ft to 7 ft radius adopted by Professor PC Mahalanobis, who experimented with jute and rice in Bihar and West Bengal. The ICAR finally adopted Sukhatme’s method for estimation of average yield of major crops like rice, jowar, wheat, etc. It subsequently led to a national programme of Crop Estimation Surveys( CES) and by 1970s India achieved an operational capability to conduct about one and a half lakh crop cutting experiments covering 30 major crops. By 1980s, this capability had been expanded to conduct about 3 lakh crop cutting experiments covering 63 crops all over the country


I.2 Plot Size in Crop Estimation SurveysThe plot size in crop estimation surveys was a bone of contention between Sukhatme and Mahalanobis, their successors and the institutions founded by them – the Statistical Wing of the ICAR at New Delhi and Indian Statistical Institute at Calcutta.
While recommending the use of large plots, Sukhatme argued that he had taken cognizance of the fact that the departmental staff entrusted with the crop cutting work was familiar with the demarcation of large
plots and that the use of small plots would be a serious deviation from their normal routine, which would inevitably result in biases in their crop cutting work, all the more so as crops in this country are generally
unevenly sown. The fierce methodological debates involving shape and size of crop-cuts for crop estimation surveys during 1950s and 1960s provided with extensive data and results to assess the magnitude of over-estimation in the methodology of crop-cuts. It was amply demonstrated that small plots significantly overestimate the yield, with the degree of over-estimation becoming smaller with larger
plots.