A large literature treats ethnicity as a proxy for persistent group-level political attitudes, beliefs, and opinions in Africa, often using historical ethnographic classifications. This paper reassesses that assumption by quantifying how individual-level variation in political attitudes and beliefs is distributed across ethnicity and geographic context. Using Afrobarometer surveys from 41 African countries linked to the Ethnographic Atlas, we show that ethnicity explains only a small share of variation across a wide range of outcomes, while subnational place consistently explains more. Within-place specifications further show that ethnicity adds little explanatory power once local context is held constant. We find, however, that the limited explanatory content of ethnicity is concentrated among groups that remain geographically close to their historical homelands and among ethnicities with higher levels of pre-colonial political centralization. Overall, the results indicate that ethnicity is a weak organizer of mass political attitudes and primarily matters when it proxies for shared local environments.
Elections are intended to enhance the selection of politicians and motivate them to perform better. Despite this, incumbency advantages often result in the re-election of suboptimal politicians. This study investigates whether the incumbency advantage persists across multiple terms and examines how mayors might alter their behavior to secure re-election. Utilizing a regression discontinuity design (RDD) in French municipal elections without term limit. The analysis reveals that incumbency advantages are substantial in the first term, diminish in the second term, reappear in the third term, and turn negative in the fourth term. Older candidates, especially in their second terms, are less likely to be re-elected. Right-wing and non-affiliated candidates show higher re-election probabilities, primarily in their early terms. A complementary RDD study differences in budget management, corruption, or discretionary grants, no changes in behaviors are detected. Indicating that mayors do not significantly change their policies to improve re-election chances. Without term limits, politicians reveals a transparent behavior.