[Paper Review] How good is good enough for COVID19 apps? The influence of benefits, accuracy, and privacy on willingness to adopt

Gabriel Kaptchuk, Daniel G. Goldstein+3 more|arXiv (Cornell University)|May 9, 2020

COVID-19 Digital Contact Tracing14 references57 citations

TL;DR

The paper surveys Americans to quantify how accuracy and privacy concerns, plus public and individual health benefits, influence willingness to install COVID-19 contact tracing apps, and models thresholds for adoption.

ABSTRACT

A growing number of contact tracing apps are being developed to complement manual contact tracing. A key question is whether users will be willing to adopt these contact tracing apps. In this work, we survey over 4,500 Americans to evaluate (1) the effect of both accuracy and privacy concerns on reported willingness to install COVID19 contact tracing apps and (2) how different groups of users weight accuracy vs. privacy. Drawing on our findings from these first two research questions, we (3) quantitatively model how the amount of public health benefit (reduction in infection rate), amount of individual benefit (true-positive detection of exposures to COVID), and degree of privacy risk in a hypothetical contact tracing app may influence American's willingness to install. Our work takes a descriptive ethics approach toward offering implications for the development of policy and app designs related to COVID19.

Motivation & Objective

• Evaluate how accuracy and privacy concerns affect willingness to install COVID-19 contact tracing apps.
• Identify which groups weigh accuracy vs. privacy more heavily in adoption decisions.
• Model how varying levels of public health benefit, individual benefit, and privacy risk influence willingness to install a hypothetical app.

Proposed method

• Conducted large-scale surveys with representative US demographics (n=4,615 total across studies) to assess willingness to install apps under different accuracy and privacy scenarios.
• Use vignette and randomized experimental questions to vary accuracy (false positives/negatives) and privacy leakage to different entities.
• Apply mixed effects binomial logistic regression to model willingness to install as a function of accuracy/privacy factors and respondent demographics.
• In RQ3, present a scenario-based survey to quantify how public health benefit and personal benefit, plus implicit/explicit privacy risk, influence adoption willingness.
• Compare implicit privacy perceptions with explicit risk statements to validate privacy effects.

Experimental results

Research questions

• RQ1 RQ1: Do both accuracy and privacy influence willingness to install a COVID app?
• RQ2 RQ2: Do different types of people weight accuracy or privacy more heavily?
• RQ3 RQ3: How much public health benefit, accuracy, and/or privacy risk is necessary for people to adopt the app?

Key findings

• Between 70-80% of Americans say they would install an app that is perfectly private and/or perfectly accurate, higher than the 50-60% for unspecified privacy or accuracy.
• False negatives have a significantly stronger negative influence on willingness to install than false positives or privacy leaks.
• Willingness to install correlates with the app’s public health benefit and personal health benefit, with a majority accepting at least a 50% improvement in either dimension.
• People who know someone who died from COVID-19 are over 5 times more likely to be willing to install an app with errors in accuracy.
• Respondents are, on average, more comfortable with false positives than false negatives, and with certain privacy leaks (e.g., non-profit) than others (e.g., employer).
• Younger individuals and women show lower willingness to install an app with privacy risks; higher internet skill increases willingness when facing accuracy or privacy errors.
• Implicit risk perceptions of privacy leaks align with explicit risk statements in predicting willingness to adopt.