A rich man’s world: biases of big data in a development context

Big data is becoming increasingly prominent within the 2030 Sustainable Development Agenda as new avenues for data generation and analysis open up. Analysing mobile phone records to inform development policy is one of the emerging options. Yet, because mobile phone records are the digital footprints of mobile phones used by a specific part of the population, they are unlikely to represent broader populations or the specific sub-groups that might be targeted by development policies and interventions. This article describes case studies from rural India and rural China to illustrate the variations among mobile phone users who would ultimately produce the digital signatures for big data analyses. Compared to the average population, phone users are more likely to be young, educated, wealthy men, whose behaviours, lives, and constraints are probably different from people who do not use mobile phones. This potential bias does not render big data unusable, but it will require the continued use of complementary conventional data in order to anchor it in local realities and to avoid systematic analytical biases. Big data can be an additional tool to inform development policy, but it cannot dictate policy on its own