A Comment on Infant Mortality Rate in India

IntroductionThis review uses unit-level record (individual recoded data file) from the National Family Health Survey (NFHS)-III (2005-06), NFHS-II (1998-99), and NFHS-I (1992-93) conducted in India. 'NFHS-III collected information from a nationally representative sample of 109,041 households, 124,385 women age 15-49 and, 74,369 men age 15-54. The NFHS-III sample covered 99 percent of India's population living in all 29 states' (1: xxix). 'The NFHS-II survey covered a representative sample of more than 90,000 eligible women age 15-49 from 26 states that comprise more than 99 percent of India's population' (2: xiii). The NFHS-I survey covered a representative sample of 89,777 ever-married women age 13-49 from 24 states and the National Capital Territory of Delhi, which comprise 99 percent of the total population of India (3: xix). It is worth to note that NFHS-II (1998-99), the second round of the series, is regarded as a 'storehouse of demographic and health data in India' (4).To compare with NFHS data, we use data from the Sample Registration System (SRS) published by the Registrar General of India (RGI), Government of India (5-7). With a view to generate reliable and continuous data on these indicators, RGI initiated the scheme of sample registration of births and deaths in India popularly known as SRS in 1964-65 on a pilot basis and on full scale since 1969-70. The SRS since then has been providing data on a regular basis. The SRS is a large scale demographic survey for providing reliable annual estimates. The sample size of these data are based on the preceding census frame and covered over 6,000,000 people living in more than 903,300 households. The revision of SRS sampling frame is undertaken in every ten years based on the results of latest census. While changing the sample, modifications in the sampling design, wider representation of population, overcoming the limitations in the existing scheme, meeting the additional requirements are taken into account. To fulfil its objective of monitoring the changes in vital indicators, the SRS sampling units are retained for about 10 years, making it a panel household survey. The first replacement (of sampling units) was carried out in 1977-78, then in 1983-85, in 1993-95 and the last being in 2004. Whereas the replacement of samples in earlier years was undertaken in phases spread over 2-3 years, the replacement in 2004 was done at one go within a year. In 1992, the overall sample at the national level comprised 6022 (4176 rural and 1846 urban) sample units, each comprising nearly 150 households and about a 1,000 population. The overall sample at the national level rose to 6671 (4436 rural and 2235 urban) and 7597 (4433 rural and 3164 urban) sample units in SRS-1998 and SRS-2005 data.Difference in sampling techniquesNFHSA uniform sample design adopted in each state is a systematic, stratified sample of households, with twostages in rural areas and three-stages in urban areas. The rural and urban samples within states were drawn separately and, to the extent possible, sample allocation was proportional to the size of the ruralurban populations. In each state, the rural sample was selected in two-stage stratified random sampling, with the selection of Primary Sampling Units (PSU), which are villages, with probability proportional to population size (PPS) at the first stage, followed by the random selection of households within each PSU at the second stage. Villages were stratified prior to selection on the basis of several variables. The first level of stratification was geographic, with districts being subdivided into contiguous regions according to their geographical characteristics. Within each of these regions, villages were further stratified using some of the following variables: village size, distance from the nearest town, proportion of non-agricultural worker, proportion of the population belonging to scheduled castes/ scheduled tribes, and female literacy. …