Anonymity and Online Search: Measuring the Privacy Impact Of Google’s 2012 Privacy Policy Change

Abstract

Abstract One of the most vexing problems in privacy policy is identifying consumer harm from unwanted observation; because it is highly subjective and is likely to vary greatly throughout the population, it doesn’t lend itself to easy measurement. Yet, these types of situations increasingly are the focal point of privacy policy discussions, including the Supreme Court’s recent decisions regarding standing and the FTC’s recently announced commercial surveillance rulemaking. The primary approach to attempt to quantify subjective harms has been to measure consumers’ willingness to exchange personal data for money in an experimental setting. This study takes a different tack, using field data to measure actual consumer response to a real-world reduction in the anonymity of online search. In March 2012, Google began to combine user information across platforms. To the extent that Google’s policy change reduced the anonymity associated with Google search, it may have diminished incentives to search sensitive topics. Using Google Trends (GT) data and a difference-in-difference estimator with top non-sensitive search terms as the control group, the results suggest that there was a 3–7 % short-term (1–2 months) reduction in sensitive search (relative to the non-sensitive search control group), as measured by GT. I examine heterogenous treatment effects, and find that the largest measured impact is for health-related search. There is no measured difference in reaction between high- and low-privacy demand states.