A key aspect of statistical inference is estimation of population characteristic. This paper investigates the reproducibility of estimates of population characteristics. It focuses on estimates based on data collected by survey-based randomised response methods (RRT) to obtain the truth in cases when the characteristic is sensitive. This work introduces a new approach called \(\epsilon \) -reproducibility of estimates based on data collected from quantitative RRT methods. This approach defines the \(\epsilon -\) reproducibility probability as the probability that, in the event that an experiment is repeated under similar circumstances, the estimate based on the data from the repeated experiment will not differ more \(\epsilon \) from the estimated based on the original data. To address prediction issues, the quantification approach makes use of Nonparametric Predictive Inference. The findings demonstrate that lower reported response variability for RRT approaches increases reproducibility of estimates using bootstrap and representative sample while maintaining an equivalent degree of privacy for survey respondents. A number of RRT methods are compared, including the Greenberg method, the Eichhorn and Hayre method, and the optional multiplicative method.
