The methodology and sample design

THE nationwide National Science Survey-2004 (NScS) was undertaken "to generate a statistically appropriate database and reliable estimates" of the various parameters related to the issues covered in the first India Science Report (ISR); namely, science and engineering education, human resource in science and technology and public response to S&T. This was a household survey but the sampled population included only individuals above 10 years of age.

The survey adopted a three-stage stratified random sampling design. The first two stages of stratification and selection were based on 1991 and 2001 data for the listing of individuals. The last stage of the sampling frame for drawing the representative sample was developed specifically for the project. The survey design was such that the sampling covered a wide cross-section of individuals (households) across the country's geographical spread and both urban and rural populations. The rural sample was selected from a representative number of districts, while the urban sample was selected from a range of big cities and small towns (with population varying from below 5,000 to over 10 million).

All the listed individuals (of over 10 years of age) were classified into 18 strata based on age, sex and educational levels. From each stratum the desired number of individuals were then selected at random. The NScS based on this sample resulted in dual sets of data. A large data set of 3,47,000 individuals was a natural outcome of the listing operation. Besides particulars of identity such as age and sex, this data set included exhaustive information on levels of education and occupation. The second smaller data set, covering about 30,000 individuals, with detailed information on their response to S&T together with demographic particulars, arose as a sub-sample from the larger sample.

The rural sample was selected from the 550 districts in 32 States/Union Territories and included 1,15,074 individuals of which 9,952 were chosen for eliciting detailed information. For the urban population, all cities with population of over a million were chosen with a probability of one and the remaining were stratified into seven groups based on their population size and from each stratum a sample of towns was selected independently. In all, 213 towns/cities were selected. A total of 2,31,841 individuals were listed of which 20,303 were selected from the 1,128 blocks of the Urban Frame Survey of the National Sample Survey Organisation.

For detailed questioning, a sample of 6,722 students and 1,681 teachers was drawn in the following manner. In each sample four students (two from Classes VI to X and one science and one non-science student from Classes XI and XII) and one schoolteacher were selected. Among the sampled students, 50 per cent were from Classes XI and XII and 36 per cent in Classes IX and X and the remaining 14 per cent belonged to Classes VI to VIII. Thirty-three per cent of the total surveyed students were girls.