Facebook algorithm’s active role in climate advertisement delivery

Abstract

Climate advertising on social media can shape attitudes towards climate change. Delivery algorithms, as key actors in the climate communication ecosystem, determine ad audience selection and might introduce demographic bias. Here, we present a two-part study—an observational analysis (n = 253,125) and a field experiment (M = 650)—to investigate algorithmic bias in Facebook’s climate ad dissemination. Our findings provide preliminary evidence that the algorithm’s selection of ad audiences can be explained by factors such as ad content, audience location (US states), gender and age group. Moreover, the cost-effectiveness of contrarian ads is linked with the conservative political alignment of a state, while the cost-effectiveness of advocacy ads correlates with liberal political alignment, higher population and per-capita gross domestic product; ad targeting strategies further modulate these effects. The skew in the distribution of climate ads across US states, age groups and genders reinforces existing climate attitudes.