RANKED SET SAMPLING PROCEDURES FOR THE ESTIMATION OF THE POPULATION MEAN UNDER NON RESPONSES: A COMPARISON.

This paper analyses the use of ranked set sampling procedures for obtaining the sub-sample from the set of non- respondents. The first visit may serve for ranking accurately the sub-sampled non-respondents. The usual ranked set sample (rss) design and two variations of it : extreme-rss and median-rss are used for developing estimators of the population mean. Their expected variances and biases are obtained. A Monte Carlo experiment is developed for evaluating the behavior of the estimators. The use of rss appears as the best alternative. [ABSTRACT FROM AUTHOR]