On Procedures of Rotation Sampling

Abstract

The reliability of any repeated occasion survey is reduces due to marked increase in non-response, which occurs due to increased burden on some of the respondents. To reduce this defect, in the present investigation a relation scheme has been developed to select sample on successive occasions in multi-stage sampling designs. For these three different type of sampling plans have been developed to estimate one more characters at the same time. Using these sampling designs an attempt has been made to obtain the minimum variance linear unbiased estimate of: 1. The population mean vector of characters at the most recent occasion. 2. The change in population mean vector of characters from one occasion to another. 3. an overall estimate of the population mean vector of characters overall occasions for a dynamic population, and 4. The modified population mean vector of characters of past occasions using knowledge of estimate of population mean vector of characters for following occasions, in a stage sampling design. Optimum sample size in different sampling plans for a given cost (after developing the appropriate cost function) has also been worked out. Further an attempt has also been made to estimate the missing value of the characters of non-responding units. using the information on previous occasions and recent occasions , under some restrictions. For illustration a study of I.A.D.P. data, collected for Bench mark surveys in Aligarh district from 1962-63 to 1965-66 has been made.