The Hindu [India], March 30, 2004
STATISTICS AND DEMOGRAPHY
By C. Rammanohar Reddy
Suggestions that Hindus will turn into a minority are simply not validated by any projections of scholarly integrity.
FOR DECADES, the fringe elements in Indian politics have drummed up a vision of an India in which uncontrolled fertility among Muslims will reduce Hindus to a minority. Increasingly, supposedly scholarly analysis is being presented to give an intellectual veneer to this argument. An excellent example is Religious Demography of India, authored by A.P. Joshi, M.D. Srinivas and J.K. Bajaj (JSB), published last year with a foreword by no less than the Deputy Prime Minister, L.K. Advani. Critics have commented on the authors' questionable categories of `Indian Religionists' (i.e. mainly Hindus) and `Other Religionists' (Muslims and Christians), the latter by implication are `non-Indian' people. They have commented on the authors' equally questionable geographic categories of `India' and the `Indian Union' — the former covering what is India, Pakistan and Bangladesh, while the latter is the India we live in.
The core argument, however, is contained in a few pages (pp37-39). There is a very suggestive graph of two intersecting lines. One shows the declining share of `Indian Religionists' (IR) in the population of `India' from 1901 to 1991, which is projected up to 2071. The second is the rising share of the `Other Religionists' (OR) over the same period. This graph persuades the authors to observe that "if the trends of the last hundred years continue to persist in the future, then Indian Religionists shall become a minority in the near future" (p38). The lines cross between 2051 and 2061. That is when Hindus are projected to become a minority and Muslims and Christians a majority.
Leave aside the social connotations of referring to `Indian Religionists' and `Other Religionists' and the political implications of using `India' to refer to three countries. The crucial question is: How valid are projections that say Hindus will soon turn into a minority? None whatsoever. The statistical techniques the authors JSB have used to project past trends into the future are inappropriate, yield inconsistent results and are a good example of an abuse of statistics to prove a pre-determined conclusion.
Population projections are regularly made by international agencies. But as K. Navaneetham, a demographer at the Centre for Development Studies, Thiruvananthapuram, points out, these estimates are based on trends in the three determinants of population growth: fertility, mortality and migration. In contrast, Religious Demography merely projects past movements of population shares/ratios into the future. The bigger problem lies in the statistical model that has been used to generate the conclusion of a Muslim-majority `India' by 2061. A particular mathematical equation — a third order polynomial — has been estimated in order to extrapolate the 1901-1991 IR and OR ratios up to 2071.
However, there is no statistical validity in this model. Chandan Mukherjee, also at the CDS, points out that while polynomial equations can indeed serve the purpose when examining changes between two points of time or even for extrapolations into the near future, they yield major errors when used for long-term projections.
To illustrate, Dr. Mukherjee and Dr. Navaneetham use the same population share data for 1901-1991 and the same equation to extend the projections that JSB come up with a century after 2071. The share of Hindus in the total population (i.e. the IRs) does keep falling decade after decade, and that of the others keeps rising. So much so that by 2171, the IRs' share falls to minus (yes, minus) 5 per cent and that of the rest to 105 per cent! This is clearly absurd. But these results are very much part of the same model that Religious Demography uses to extend the past into the future.
In another exercise, the CDS researchers also use the same model but make projections with the data on the 1901-1991 population, not ratios. The IR population share does fall and that of the OR increases. But there are two differences. One, the decline is much more gradual, so much so that even by 2171, the IRs will still be a majority. Second, the total population of `India' projected for 2051 turns out to be much larger than the U.N.'s projections for India, Pakistan and Bangladesh for the same year: 3.8 billion versus 2.2. billion. By 2171, according to the projections generated by the polynomial, the population of the region would rise to an astronomical 16 billion. This is clearly impossible and reveals the fundamental flaws of the Religious Demography model.
In other words, the picture that the authors draw to suggest that Hindus will turn into a minority is simply not validated by any projections of scholarly integrity. The problem essentially is that techniques of the kind the authors have used are totally inappropriate for making long-term projections.
In an exhaustive critique, D. Jayaraj and S. Subramanian of the Madras Institute of Development Studies point out (Economic and Political Weekly, March 20, 2004) similar and many more fundamental errors in the JSB analysis. One absurdity is that, as a cross-check, if the same polynomial equation is used to make a projection into the past, then we will find that in 1781, Hindus accounted for 99.7 per cent of the `India' population. And if the same equation is used to project the share of Asians in the U.S. population, then we will end up saying that by 2140 Asians will be a majority in that country!
What we have here then is neither statistics nor demography. It is the use of pseudo-statistics in the interests of pseudo-demography, the objective being to feed a fear of the minorities.
|
