This paper presents a new randomized response model that combines Kim and Warde's
(2004) stratified Warner's randomized response technique using optimal allocation with the
unrelated question randomized response model. The empirical studies performed show that,
for the prior information given, the new model is more efficient in terms of variance (in the
case of completely truthful reporting) and mean square error (in the case of less than
completely truthful reporting) than its component models.

This paper applies stratified random sampling using Neyman allocation to Mangat et al.
unrelated question randomized response (RR) strategy for both completely truthful reporting
and less than completely truthful reporting. It is shown that, for the prior information given,
our new model is more efficient in terms of variance (in the case of completely truthful
reporting) and mean square error (in terms of less than completely truthful reporting) than
Kim and Elam's model. Numerical illustrations and graphs are also given in support of the …