Résumé
A 30-cluster survey method that is employed for estimating immunization coverages by the Government of India (GOI) was compared with a Purposive method, to investigate whether the likely omission of SC/ST and backward classes in the former would lead to the reporting of higher coverages. The essential difference between the two methods is in the manner in which the first household is selected in the chosen first stage sampling units (villages). With the GOI method, it is often close to the village centre, whereas with the Purposive method it is always in the periphery or in a pocket consisting of SC/ST or backward classes. A concurrent comparison of the two methods in three districts in Tamil Nadu showed no real differences in the coverage with DPT and BCG vaccines. However, the coverage was consistently higher by the GOI method in the case of the Polio vaccine (by 1.5%, 3.1% and 5.3% in the 3 districts), and the Measles vaccine (by 4.8%, 13.3% and 13.9%); the average difference was 3.3% for Polio vaccine (p = 0.08) and 7.3% for Measles vaccine (p = 0.01).