Bayesian evaluation of privacy-preserving risk communication for user android app preferences

The purpose of this paper is to propose practical and usable interactions that will allow more informed, risk-aware comparisons for individuals during app selections. The authors include an explicit argument for the role of human decision-making during app selection and close with a discussion of the strengths of a Bayesian approach to evaluating privacy and security interventions.

The authors focused on the risk communication in mobile marketplace’s realm, examining how risk indicators can help people choose more secure and privacy-preserving apps. Combining canonical findings in risk perception with previous work in usable security, the authors designed indicators for each app to enable decisions that prioritize risk avoidance. Specifically, the authors performed a natural experiment with N = 60 participants, where they asked them to select applications on Android tablets with accurate real-time marketplace data.

In the aggregate, the authors found that app selections changed to be more risk-averse in the presence of a user-centered multi-level warning system using visual indicators that enabled a click-thru to the more detailed risk and permissions information.

Privacy research in the laboratory is often in conflict with privacy decision-making in the marketplace, resulting in a privacy paradox. To better understand this, the authors implemented a research design based on clinical experimental approaches, testing the interaction in a noisy, confounded field environment.