<p>In recent years, survey statisticians have taken a keen interest in exploring efficient estimators for the population variance in randomized response surveys. The existing sensitive variable-based variance estimators traditionally employ one-time measurements on sampled units using randomized response techniques. A serious issue with one-time measurements is that the interviewing process often results in measurement errors which may badly influence the estimates of parameters. This paper presents an optimal Exponential Weighted Moving Average (EWMA) variance estimator under randomized response surveys on sensitive variables. Additionally, a new randomized response survey technique has also been developed. The results suggest that the proposed estimator achieves better efficiency over its competitors. Further, the improvement of the proposed technique over the competitor techniques has been shown using different measures of performance evaluation.</p>