Working Papers

Social Network Working Papers


Social Networks and the Political Salience of Ethnicity

Job Market Paper  R&R, Quarterly Journal of Political Science

Ethnic politics scholars are increasingly convinced that (a) the political salience of ethnicity and (b) the correlation between ethno-linguistic fractionalization (ELF) and poor development are driven by the dense social networks shared by co-ethnics. By this argument, social networks allow ethnic parties to leverage inbuilt networks to share information and support collective action, while ethnically fragmented communities struggle to hold politicians accountable. This paper provides the first comprehensive empirical test of the assumption underlying this argument. Using seven months of telecommunications data from 9 million mobile subscribers in Zambia – which includes records of almost 2 billion calls and SMS messages – to measure social networks across an entire country, this paper finds that electoral constituencies with high ELF also have more fragmented social networks, especially in rural areas. It also finds other potential cleavages that have not achieved political salience (namely religious identity and employment sector) are not correlated with network fragmentation, consistent with the idea ethnicity achieves salience because it offers an organizational advantage not offered by other cleavages. Finally, it also finds that both voter knowledge and public goods are negatively correlated with network fragmentation, consistent with the network-proxy hypothesis.

Social Networks and Turnout in Low Salience Elections: Evidence from Uganda

with Guy Grossman, Melina Platas, and Jonathan Rodden. Under Review. 

Citizens often mirror the behavior of their peers, but our understanding of the dynamics of this influence is limited. For example, in what settings does the choice of one person to vote cascade through a community and lead to high voter turnout? Despite substantial theoretical inroads into this question, direct empirical tests remain scarce. Using data on the social networks of 15 villages in rural Uganda, this paper develops theoretical predictions about expected cross-village variation in turnout based on the network structure of each village, and demonstrates that these predictions are tightly linked with actual turnout in low-salience local elections with limited media attention, though not in high-salience presidential elections. These results provide the first direct empirical validation of “social context” theory, and introduce a finding of importance for future empirical network research: the salience of social networks may be conditional on the information environment.

Other Working Papers


Who is my Neighbor? The Spatial Efficiency of Partisanship

with Jonathan Rodden. 

Relative to its overall statewide support, the Republican Party has been over-represented in Congressional delegations and state legislatures over the last decade in a number of U.S. states. A challenge for courts is to determine the extent to which this can be explained by intentional gerrymandering vis-a-vis an underlying inefficient distribution of Democrats in cities. We explain the problem of “spatial inefficiency” in partisan support, and measure it by borrowing from the field of plant ecology, assessing the partisanship of the nearest neighbors of each voter in each U.S. state at the spatial scales relevant for Congressional delegations and both chambers of state legislatures. We demonstrate that as a result of urban-rural partisan polarization, much of the overall Republican advantage can be explained by the spatial inefficiency of partisanship. Moreover, this provides us with a useful baseline against which to evaluate claims of partisan gerrymandering. We demonstrate that when Republicans are often able to improve significantly on their underlying geographic advantage when they control the redistricting process, while Democrats are sometimes able to ameliorate it.

Decomposing the Government-Private School Performance Differential: Village Ethnic Politics and School Sorting

Replication Materials

This paper leverages variation in sorting on academic potential caused by village caste politics to determine if private schools in rural Punjab, Pakistan out perform government schools because (a) private schools provide students with a better education, or (b) students attending private schools are more academically inclined in unobservable ways. It concludes that even the most sophisticated observational techniques — lagged Value-Added models — overstate private school quality by at least half.