Navigating protection and presence: Trade-offs around data suppression for small Pacific populations

Abstract

Introduction: Datasets, their analytics and their interpretation are key decision support tools for Pacific Island communities, with the potential to shape public policy, healthcare, and social interventions in the Pacific ‘Blue Continent’. However, in the case of numerically small island populations, privacy concerns have motivated widespread use of data suppression. While suppression safeguards privacy, it also risks erasing the visibility of these populations, leading to ‘statistical invisibility’ that obscures the social, health, and economic challenges. This study critically reviews the practice of data suppression, emphasizing its rationale in privacy protection, but also highlighting the impacts on resource allocation, advocacy, and equitable policy-making for Pacific populations.

Methods: We explored the rationale behind data suppression, and its legal and regulatory context. Using case studies including the U.S. Census Bureau, Centers for Disease Control and Behavioral Risk Factor Surveillance System, we assess the impact of suppression thresholds and privacy-preserving methods on Pacific Island communities. We present a novel analysis of data suppression impacts on ICD code suppression across different levels of geographical units in the Pacific to illustrate disproportionate impacts. We review alternative privacy-preserving methods, including data smoothing, statistical masking, and synthetic data generation, that could mitigate the effects of suppression without compromising individual privacy.

Finding and Conclusions: We recommend inclusive and transparent data practices needed to prevent data suppression compounding systemic marginalization of small Pacific populations. By critically evaluating current practices and proposing alternative strategies grounded in ‘Critical Data Theory’ and Pacific knowledge epistemology, this paper aims to inform policies that balance protection of individual privacy with the accurate representation of small, geographically dispersed populations.