Evaluation of Randomized Response Models for the Mean and Using Data on Income and Living Conditions
Evaluation of Randomized Response Models for the Mean and Using Data on Income and Living Conditions

Summary/Abstract: The surveys on Income and Living Conditions carried out by the statistical agencies of many countries may contain sensitive questions, which can induce the respondent to give a false answer or no answer. Although the most common technique to treat this is the imputation, the randomized response techniques can also be used. In this paper we evaluate the performance of two randomized response models to estimate the population mean with real data obtained from the 2012 Spanish Survey on Income and Living Conditions. For this purpose, Monte Carlo simulation studies are carried out. Empirical results indicate that no estimator presents bias when individuals provide the true answers while the direct estimator presents a large bias when providing false answers. We also observe that one of the estimators based on randomized response models is more efficient than the other.

• Details
• Contents