Publications
Publications
- September 2018
- Organizational Behavior and Human Decision Processes
When and Why Randomized Response Techniques (Fail to) Elicit the Truth
By: Leslie K. John, George Loewenstein, Alessandro Acquisti and Joachim Vosgerau
Abstract
By adding random noise to individual responses, randomized response techniques (RRTs) are intended to enhance privacy protection and encourage honest disclosure of sensitive information. Empirical findings on their success in doing so are, however, mixed. In nine experiments, we show that the noise introduced by RRTs can make respondents concerned that innocuous responses will be interpreted as admissions, and, as a result, yield prevalence estimates that are lower than direct questioning (Studies 1–4, 5A, & 6), less accurate than direct questioning (Studies 1, 3, 4B, & 5A), and even nonsensical (i.e., negative, Studies 3–6). Studies 2A and 2B show that the paradox is eliminated when the target behavior is socially desirable, even when it is merely framed as such. Study 3 shows the paradox is driven by respondents’ concerns over response misinterpretation. A simple modification designed to reduce concerns over response misinterpretation reduces the problem (Studies 4 & 5), particularly when such concerns are heightened (Studies 5 & 6).
Keywords
Truth-telling; Lying; Privacy; Information Disclosure; Survey Research; Surveys; Attitudes; Behavior
Citation
John, Leslie K., George Loewenstein, Alessandro Acquisti, and Joachim Vosgerau. "When and Why Randomized Response Techniques (Fail to) Elicit the Truth." Organizational Behavior and Human Decision Processes 148 (September 2018): 101–123.