On Quantitative Randomized Response Models with Repeated Trials

Abstract

In survey sampling, it is sometimes needed to obtain data on sensitive variables. However, its reliability becomes a question mark, owing to the potential of high rate of non response or dishonest responses from the respondents. Randomized response technique proves to be a protective shield against such problems, encountered by the interviewer. A wide range of response models based on this technique have been proposed so far. One of its variants to obtain reliable data on quantitative variable, includes Scrambled Randomized Response models (SRRM). A lot of literature and diversity of models, can be found on SRRMs. In this paper, a new technique of repeated trials has been introduced for the first time, in obtaining responses on a quantitative sensitive variable under study. However, repeated trials has been done by employing POONAM randomization device, proposed by Singh(2010), for the ease of computations and simplification. Three models have been designed under the proposed scheme and comparisons are made with POONAM model. All of the three models are also compared among each other to evaluate the most suitable one. It is shown that the proposed methodology performs better, and provide a new direction in estimating mean of a sensitive quantitative variable, which can further be modified for more efficient estimates.