Data Donations for Digital Contact Tracing: Short- and Long-Term Effects of Monetary Incentives

Abstract

Data donations promise to unlock the social benefits of personal data. Most recently, contact-tracing apps have been developed to collect contact and health data from individuals to fight the COVID-19 pandemic. In comparison to commercial apps, the adoption of contact-tracing apps involves a unique cost-benefit calculus. In particular, the pro-social motives to engage in data donations and the need for long-term but mostly passive usage render digital contact tracing a novel IS adoption setting. Because the effectiveness of contact-tracing apps hinges on widespread adoption and continuous data collection, we use a randomized controlled online experiment to evaluate the effectiveness of different monetary incentive mechanisms at promoting verified installations of the German Corona-Warn-App and short- and long-term data donations. We find that monetary incentives are effective in the short term, with no evidence of a crowding-out of pro-social motives: Monetary incentives significantly increase app installations and short-term data donations, tripling the number of data donors after 14 days compared to a no-compensation treatment. However, the positive stimulus of monetary incentives vanishes in the long term: After eight months, adopters in treatments with extrinsic monetary incentives are significantly more likely to have stopped donating data than intrinsically motivated adopters who do not receive monetary incentives. Consequently, long-term data donation rates are not significantly higher in treatments with monetary incentives. This suggests that one-time payments ineffectively promote long-term data donations and that positive effects from app sampling and user inertia are limited, especially if opportunity costs of app usage are experienced to be high, personal benefits are perceived to be low and the formation of user habits is constrained by passive app usage. Finally, we present experimental evidence that analyses based on hypothetical scenarios without verified actions are prone to overestimating the pro-social behavior of individuals in the context of digital contact tracing.

Victoria Fast, Daniel Schnurr

SSRN