Thumbnail Image

Theses

Browse

20051
Reset filters
Subject: Agricultural Statistics and Informatics × Author: Gayatri, R.K. × End date: 2009 × Start date: 2004 × Type: Thesis × Thesis Type: M.Sc ×
Reset filters

Search Results

Now showing 1 - 1 of 1
  • Thumbnail Image
    ThesisItemOpen Access
    Studies of Roychoudhury method in unequal probability sampling
    (G.B. Pant University of Agriculture and Technology, Pantnagar - 263145 (Uttarakhand), 2005-05) Gayatri, R.K.; Amdekar, S.J.
    The major objective of a sample survey is to make inferences about some characteristics of a population. Mainly, one is interested in estimating the population mean or population total of some characteristic called as study variable. Unequal probability sampling is one of the basic methods of sample selection. When the selection probability is based on the auxiliary variable it is commonly called as probability proportional to size (PPS) sampling. Estimator based on PPS sampling is expected to be better than simple random sampling, When there is proportionality between auxiliary and study variable. Roychoudhury (1957) gave a method in which the estimator has no sampling error even when the intercept of regression line is away from origin. Amdekar (2003) has generalized the Roychoudhury method. In the present study the performances of Roychoudhury and generalized Roychoudhury estimator are investigated empirically by considering two superpopulation models one involving normal distribution and other involving gamma distribution. From these distributions samples were drawn and by considering each sample as a population variances of various estimators are worked out. It is observed that in case of populations having normal distribution with increase in relative intercept the efficiency of Roychoudhury and generalized Roychoudhury generally increases and these estimators are better than PPSWR, ratio and regression estimators. Further, for the populations having moderate departure from symmetry generalized Roychoudhury estimator is better than other estimators and is less efficient when the distribution becomes more skewed. Some investigations were also carried out for estimators based on sample size two and it was observed that the weighted estimator has smaller variance than all the estimators included in the study.