Where We Work – Global Elections Lab

Projects

Rational ignorance and authoritarian elections

What’s rational ignorance? Low-information environments, constrained competition, and toothless offices sometimes discourage voters from making the costly efforts needed to cast an informed ballot. In such settings, a common heuristic device is to vote for the first option(s) on the ballot. When enough citizens take this approach, a statistically significant ballot order effect (BOE) is detectable in the aggregate. So we’re hunting for original data on authoritarian elections in contexts where BOEs are detectable in theory: ballot orders must be unrelated to party/candidate characteristics and there must be multiple ballots with differing options and orders.

Afghanistan: national legislature (2005, 2010, 2018)
Armenia: city councils (2021)
Azerbaijan: national legislature (2010, 2015, 2020, 2024)
Belarus: national legislature (2004, 2008, 2012, 2016, 2019, 2024)
Brazil: national legislature and some state legislatures (1966, 1970, 1974, 1978)
Cambodia: commune councils (2012, 2017, 2022)
China: township People’s Congresses (1998-99, 2001-02);
Hungary: national legislature (1990, 1994, 1998, 2002, 2006, 2010, 2014, 2018, 2022)
Iran: national Assembly of Experts (2024)
Kazakhstan: national legislature (2011, 2016, 2021, 2023)
Kyrgyzstan: city councils (2016, 2021)
Mexico: national legislature (1985)
Myanmar: national legislature and all state/regional legislatures (2010, 2015, 2020)
Pakistan: national legislature and some state legislatures (1988, 1990, 1993, 1997, 2002, 2008, 2013, 2018)
Poland: national legislature (1989)
Russia: national legislature (2003, 2007, 2011, 2016, 2021); regional legislatures (2003–2024)
Ukraine: national legislature (1998, 2002, 2012)
United States: state election laws (2023)
Uzbekistan: national legislature (2024)
Vietnam: national legislature (2007, 2011, 2016, 2021)
Zambia: national legislature (1964, 1991, 1996, 2001, 2006, 2011, 2016, 2021); city councils (2016)
Zimbabwe: national legislature (2005, 2008, 2013, 2018)

China’s 1979 Election Law and local People’s Congresses

Since 1954, China has had a five-tiered legislative hierarchy. Deputies to the National People’s Congress are indirectly elected by deputies to the 30-odd provincial PCs, who are in turn elected by deputies to around 300 prefectural PCs, themselves indirectly elected by approximately 3,000 county PCs. Deputies to county PCs and around 30,000 township PCs are directly elected by the voting public. But prior to 1979, county PC deputies were indirectly elected by township PC deputies. In addition to exposing would-be county deputies to the Chinese electorate, the 1979 Law also mandated that uncompetitive deng-e (等额) elections be replaced by semicompetitive cha-e (差额) elections. In the former, as was common in the Soviet Union, the number of choices on the ballot equaled the number of seats to be filled; voters had no choice. In the latter, candidates could exceed seats by 1.5x–2x.

By suddenly letting voters choose county PC deputies from a larger slate of candidates, the 1979 reforms opened the door to unexpected outcomes. How did they affect the representation of women, ethnic minorities, and non-Party members? Because the names and vote totals of winners and losers are lost to history, we’re using local gazetteers to gather legislative rosters and produce aggregate statistics about legislative composition before and after the reforms. By comparing pre-post trends in county PCs (the treated) to similar over-time trends in prefecture PCs which were relatively unaffected by the reforms (the control), we’ll be able to identify the causal effects of the 1979 Election Law. We’re focusing our efforts on six provinces — collectively home to 1/3 of the Chinese population — that are distributed across all of China’s six official administrative regions:

Hebei (North China / 华北)
Heilongjiang (Northeast China / 东北)
Shandong (East China / 华东)
Henan (South Central China / 中南)
Sichuan (Southwestern China / 西南)
Shaanxi (Northwestern China / 西北)

Network centrality for social choice

When each voter rank-orders all her options, rather than simply voting for her top choice(s), and the orderings of all voters are counted up, a preference profile is created. For a given electorate and choice set, any preference profile can be represented as a directed graph with choices as nodes and pairwise relationships as weighted or unweighted links. We explore the application of network centrality measures as a new approach to generating social rankings and an overall social choice. To that end, we utilize empirical data on ranked-choice elections originally gathered by David McCune and collaborators:

Australian state and local legislative elections
Scottish local elections
various U.S. federal, state, and local elections
Condorcet Internet Voting Service (CIVS) polls

The electoral history of NC’s county legislatures

Each of North Carolina’s 100 counties elects a board of county commissioners who hold broad authority to decide what services to provide and to set county fiscal policies accordingly. There are currently just under 600 commissioners and the modal board size is five members. Some commissioners are elected by the county as a whole, others are elected in geographic districts. Some boards are elected all in one go, while most stagger their members’ terms to ensure continuity. Electoral rules may be set by the commissions themselves, by the state legislature, or by court decree. These local legislatures constitute an excellent laboratory in which to assess the effects of various electoral arrangements. In addition, boards of commissioners have existed since the 1870s, ensuring that a complete electoral history would cover the essentially authoritarian period of Democratic party dominance known as the Solid South. We’re therefore combining multisource archival work, county-by-county requests, and webscraping to produce a complete dataset of election returns from North Carolina’s county commissions.

The Smart Vote campaign and strategic coordination in Russian elections

Five of the last nine US presidential elections were won with less than 50% of the vote. More generally, whenever more than two candidates compete in a first-past-the-post (FPTP) election, the winner may fail to obtain an outright majority as votes are scattered across candidates. In the Russian context, the hegemonic party, United Russia (UR), has frequently managed to win seats with mere pluralities, while a majority of voters opted for an opposition candidate. The late activist Alexei Navalny once proposed to curtail UR victories by spurring coordinated voting across the opposition. Called “Smart Vote,” the campaign entailed identification of a plausible winner amongst the opposition — regardless of party — in each contest and a last-minute notification of opposition voters through a dedicated platform. Although the Smart Vote-endorsed candidate may not have been an opposition voter’s first choice, coordinated voting theoretically helped avoid the voter’s last choice — United Russia.

Limited existing work evaluates the effectiveness of the Smart Vote campaign in a few city council elections and in the national legislature, the Duma. We’re working to produce an original dataset that would allow us to assess Smart Vote across Russia’s many regional legislatures using a novel research design: contrast the over-time change in the effective number of parties (ENP) in first-past-the-post elections to the change in ENP in proportional representation (PR) elections in the same regions. The former elections were the focus of the Smart Vote campaign; in the latter elections, Navalny’s prescription to opposition voters was “anyone but UR.” This means that in the mixed electoral systems common to Russian regions, FPTP elections were treated to spark coordination, while simultaneously held PR elections were left untreated.