22 October 2010

World Statistics Day

The first World Statistics Day, 20 October 2010 (20.10.2010) went by without any formal activity on campus at the Indira Gandhi Institute of Development Research (IGIDR), Mumbai. However, for the Research Methodology course there was a lecture on Measuring Government Output, which was apt for the occasion. This was preceded by a lecture on Price Indices in National Accounting the previous day (19 October 2010).

From a research methodology perspective, the two lectures also have some additional learning. Let me elaborate on two points.

The first is to go down to details. While looking into a problem, start asking questions and while answering some of the questions you will end up many more questions leading to a search for some more answers. The process of raising questions and answering is rewarding as it will take you a step closer towards unraveling 'curiosity', which is an unending journey.

This reminds of a folklore from the Indian Statistical Institute. Post-partition, post-independence in New Delhi a sizable proportion of Muslim population were staying inside the Red Fort. Food was provided to people inside, but there was a feeling in the establishment that large amounts were siphoned.

If I recall correctly, Pandit Nehru called up Professor Mahalanobis so that the Indian Statistical Institute could help them out. There was no way that they could do a survey because anybody going inside would have been killed and likewise someone coming out would have met the same fate.

The statisticians went about the problem in the following manner. They took expert advise to arrive at the minimum amount of salt an individual will consume per day. This figure was divided by the total amount of salt requisitioned (as a proxy for having been consumed) to arrived at an estimate of the total population. Similar expert advise helped them arrive at the maximum amount of rice and other items an individual will consume per day and and divided this by the total amount of that item requisitioned to get estimates of the population. Note that the minimum salt per day will give overestimates of population whereas maximum amount for rice and other items will give underestimates of population. Despite this, the population estimates through rice and other items were many times higher than that estimated by salt giving us an extent of the corruption.

The second is to inculcate self-doubt. This is easier said then done. But, then these are some of the first principles. It is only when we start doubting that we raise questions. When we put ourselves to scrutiny, we may be able to appreciate another point of view. As an aside, it reminds me of the proverbial statement, "if you point a finger at someone then do keep in mind that three of the fingers are pointing towards you."

Having said this, I would like to state that one should be careful in dealing with statistics. It has the potential to mislead. You must have heard about the story of how three persons repeat a lie and make a fourth person believe that the lie is indeed the truth. No wonder, it is often said that "there are three kind of lies: lies, damned lies and statistics." It is for this that everyone in the business of statistics should work with honesty and integrity in their observation and analysis.

This takes me to Tycho Brahe, who till his death believed that the Sun went around the Earth, but despite this belief he did not tamper with his observations and it is for this that Kepler could come up the three laws of planetary motion. Brahe's observation also paved the path for least-squares curve fitting with the most popular being the linear regression analysis in statistics. Long live statistics!

No comments:

Post a Comment