Adaptive strategies in unrelated question randomized response techniques for sensitive data collection

Résumé fourni par la source

When surveys involve sensitive issues-such as gambling habits, drug addiction, alcoholism, tax evasion, induced abortions, drunk driving, criminal history, or homosexuality non-response and response biases frequently occur, as respondents may be unwilling to disclose truthful information. To address this issue, Warner introduced the Randomized Response Technique (RRT), which protects respondent privacy while facilitating accurate data collection. Building on this idea, the present study proposes a novel unrelated question randomized response model designed to reduce both response and non-response biases associated with stigmatized attributes. The model provides an unbiased estimator for the population proportion possessing the sensitive characteristic. Furthermore, a cost function is incorporated into the analysis to evaluate the trade-offs between efficiency and resource utilization, offering a practical decision-making tool for survey planners. Theoretical properties of the model are thoroughly investigated, and a small-scale simulation study is conducted to validate its performance. Numerical illustrations and R-based simulation codes are also provided to support the applicability of the proposed method.

Ce résumé expose les affirmations des auteurs. BNTIC ne l’interprète pas comme une validation indépendante des résultats.