Optional Randomized Response Techniques for Complex Survey Designs

Gupta, Gupta and Singh (2002) proposed an optional randomized response (ORR) technique based on SRSWR sampling using an unrealistic assumption that the probability of providing the true/randomized response for all the individuals in a population is the same. In this paper an alternative ORR has been proposed for an arbitrary sampling design. Expressions for the estimators of the population mean, variance and unbiased estimator for the variance have been provided. It is shown that the estimator, based on an ORR, is more efficient than an estimator derived from a non-ORR technique. Sufficient conditions for the superiority of the proposed estimator over the existing estimator provided by Gupta et al.'s (2002) have also been established. (© 2004 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim)