This paper examines the impact of using the Internet and social networks as sources of information on individuals’ understanding of democracy. The analysis draws on data from the sixth round of the Afrobarometer survey, conducted in 2014, across a sample of African countries, some of which are democracies and others non-democracies. This topic has received limited attention in the economic literature, despite its relevance in
understanding political preferences and behaviors. The sixth round of the Afrobarometer survey is particularly suited for this analysis, as it is the only wave to include both open-ended and closed-ended questions regarding individuals’ understanding of democracy. In this study, we prioritize open-ended questions as we consider them to better capture individuals’ conceptual grasp of democracy. However, we also use closed-ended questions to test the robustness of our results. Open-ended responses are structured and analyzed using the ChatGPT tool
to extract meaningful insights. The identification strategy leverages the interaction between lightning activity and 3G coverage. Lightning activity introduces random interruptions in Internet access, creating an exogenous source of variation that enables causal inference. Our results indicate that the bias in understanding democracy induced by reliance on the Internet and social networks as information sources is predominantly negative. This negative bias extends to individuals’ perceptions of freedom of expression and institutional corruption in political regimes. Conversely, we find a positive bias regarding perceptions of the fairness of the electoral process. These findings carry significant implications, as we document a positive association between the understanding of democracy and the preference for democratic systems. This suggests that distortions caused by Internet use, which has become a major source of information in Africa, may influence individuals’ political preferences and attitudes.