Youth engagement in democratic processes is declining across Africa, with many losing faith in the power of voting (Berinsky et al., 2016; Resnick and Casale, 2014). This disengagement, however, varies by ethnic group, but is simultaneously orthogonal to ethnic voting. After matching societies in the Ethnographic Atlas to ethnic categories in Afrobarometer, the surveys' election behavior questions reveal that ethnic groups socially organized along members' age (age-sets) experience relatively lower reported voter turnout in elections with low youth participation, than those organized by kinship lineages networks. I argue that age-set customary institutions foster strong peer networks and cohort-based social pressures, making age-set youth more likely to receive political information from, and be influenced to vote by, their fellow youth. Six rounds of Afrobarometer surveys in thirty-three Sub-Saharan countries show that age is a stronger predictor of longitudinal shifts in respondents' attitudes in age-set organized ethnic groups, suggesting stronger peer-to-peer information networks and norm adoption. Combined, age-set social network structure and age-cohort conformity appear to exacerbate the generational downward trend in voter turnout for the ~130 million Sub-Saharan Africans from ethnicities organized along age-sets.

From the moment social scientists began systematically documenting conflict events, they also began to develop methods for conflict prediction. The two leading approaches—using humans or models to make geopolitical forecasts—are enhanced when combined, as humans and machine models exhibit divergent biases (Benjamin et al., 2023). However, human forecasts by subject domain experts, while necessary for a combined approach, have fallen short. We take a seemingly simplistic approach that has, to the authors’ knowledge, not yet been tested in the conflict literature: to employ locals–instead of conflict experts–as conflict forecasters. We argue that local residents who share information networks with locally recruited rebels and possess knowledge of context-specific factors, such as soldiers' paydays and village celebrations, may be best positioned to predict the likelihood of future attacks. We test local forecasting accuracy in the context of Cabo Delgado, Mozambique where an insurgency has attacked villages and government forces since 2017. For over seven months in 2022-2023, we surveyed 4,420 residents in Cabo Delgado, Mozambique: half were randomly selected to predicted the likelihood of attacks in their community over the next two weeks, while the other half reported on recent attacks. We find that conflict predictions based on local forecasts surpassed our baseline estimates, based on ACLED and past reported conflict rates. We then review features of accurate predictions such as village location, attack type (e.g., beheading, kidnapping), and forecasters' demographic background, social networks, and information environments.

Theories explaining collective mobilization under conditions of ethnic group inequality presuppose that ethnic group boundaries are impermeable, and that members of powerful groups protect their resources by denying access to members from subordinate groups. However, inter-ethnic marriage is rising across Africa and now characterizes one-fifth of all marriages on the continent. The frequency of these unions break existing assumptions, meaning that scholars can no longer assume that clusters from kinship networks perfectly match ethnic group labels. This is the first paper to test these assumptions empirically and find shifts in between-group inequality, measured via the Ethnic Power Relations dataset, are correlated with a reduction in inter-ethnic marriage. I find additional evidence that ethnic group status impacts the likelihood of inter-ethnic marriage after accounting for groups’ kinship norms (matrilineal/patrilineal descent). I observe that inter-ethnic marriages are more likely to occur between unequal groups if descent is passed from the parent from the more powerful group (mother/matrilineal, father/patrilineal).