|
|
 |
| Methodology |
| |
The hallmark of CSDS surveys has been large cross-section samples representative of the universe of study. Most of the CSDS surveys are designed by employing the multi-stage stratified random sampling technique, which ensures representation of the social, cultural and political diversity of the population within the sample.
The first stage of sampling involves the selection of Assembly Constituencies. In national surveys they are stratified according to the states in which they lie, and the unit of sampling is the state in all surveys. Assembly Constituencies are sampled through the Probability Proportionate to Size (PPS) sampling method. Past election results and demographic characteristics of the sampled constituencies are rigorously matched with the overall results of previous elections in the state provided by the Election Commission of India, and demographics of the state listed by the Census of India. This is to ensure that the sampled Assembly Constituencies are representative of the state. The second stage is to sample Polling Stations within each sampled constituency. Polling Stations are again sampled by employing the PPS method. The PPS method ensures that locations with larger population are sampled, and the underlying assumption is that locations with larger population will be more representative of the diversity within the universe.
After sampling the Polling Stations, the final stage is that of sampling respondents, who are selected from the Electoral Rolls provided by the Election Commission. Respondents are sampled by the Systematic Random Sampling (SRS) method, which is based on a fixed interval ratio between two respondents. A Sampled Respondent’s List for each Polling Station is prepared, which is a comprehensive list of selected respondents with their complete name, address, age and gender. Field investigators who are trained to conduct face-to-face interviews in a rigorous training workshop go and interview the listed respondents. In most CSDS surveys substitution is not allowed, and in some exceptional cases where it is allowed, a strict criteria of substituting the respondent by another individual in the same household, of the same age group and gender is followed.
The only exception to the discussed methodology is when few target group surveys are conducted. However, even in target group surveys the first two stages of sampling is the same as cross-section surveys, and at the stage of sampling respondents, first a comprehensive list of potential respondents is prepared by selecting all the voters of the target group listed in the Electoral Rolls. Then the Samples Respondent’s List is prepared by using the SRS method. The State of Nation Survey of January 2008, which was an exclusive survey of Indian women, is an example of such a target group survey. In some surveys, such as the Indo-Pak Poll of August 2007, which was an all-urban survey, a purposive sample was drawn. However, even in such surveys strict protocols of randomisation are followed at the final stage of sampling respondents.
In all CSDS surveys interviews are conducted face-to-face using a standard-structured questionnaire. Another achievement has been that all interviews are conducted in local dialects, and the National Election Study – 2004 was conducted in as many as 22 Indian languages. After the data is collected, rigorous contingency checks are run, after which it is analysed using SPSS (Statistical Software for Social Sciences).
|
|
|
|
|
|
 |
|
|
