RANDOMIZED ESTIMATION A PROPORTION USING RANKED SET SAMPLING AND WARNER'S PROCEDURE.

This paper is devoted to the analysis of the estimation of the proportion of a sensitive variable. The use of a randomized response (rr) procedure gives confidence to the interviewed that his privacy is protected. We consider that a simple random sampling with replacement design is used for selecting a sample. The behavior of the rr procedure, when ranked set sampling is the design used, is developed under three different ranking criteria. The usual gain in accuracy associated with the use of ranked set sampling is exhibited only by one of the designs. The behavior of the models is illustrated using data provided by a study of samples of persons infected with AIDS. [ABSTRACT FROM AUTHOR]