Turnout in Referendums: A Formal Model with Evidence from Colombia

• In Event: A New Look at Political Participation and Turnout

Abstract

Referendums are often portrayed as exercises in direct democracy that transfer decision-making power from representative governments to the people. And yet “the people” do not seem anxious to participate in these votes: in the 271 national referendums held around the world between 1961 and 2018, almost all saw lower turnout than in candidate elections in the same countries. What explains low voter turnout in referendums?

The primary explanation scholars offer focuses on salience. On this account, referendums tend to be about low-salience issues, and consequently draw fewer voters to the polls than do candidate elections. Yet salience explains only part of the puzzle; other factors are also at play. Through a formal model, we focus on the degree of incentive-alignment within parties as key to their success in getting out the vote. In candidate elections, local-level politicians and operatives have strong incentives to get out the vote for their party for both local and national candidates. In referendums, these incentives are weaker, in part because local politicians do not garner the same kinds of spoils of victory when their side prevails, and in part because their efforts play a smaller role in the outcome of national referendums than in candidate elections.

Drawing on this theory, we identify four variables that explain the disparities between candidate elections and referendums: the types of candidates or politicians responsible for mobilization (whether office-seekers or ideologues), whether candidate elections in the system in question tend to be personalistic or programmatic, the strength of political parties, and the number of effective parties in the system. Together, these variables help determine the size of the turnout deficit in referendums.

Where politicians are ideological and accustomed to conducting programmatic campaigns, their skills and resources translate neatly to mobilizing voters for referendums. Conversely, where politicians are purely office seeking or are accustomed to running personalistic campaigns, they have neither the desire nor the appropriate skills to mobilise voters for a programmatic vote such as a referendum. In turn, where parties are strong and can credibly monitor and punish lower-level candidates and operatives, party organization can compel candidates to invest heavily in mobilizing for a referendum vote. Where they are weak, they cannot coerce their politicians to mobilize. Finally, a two-party system, in which each party defends one seat, enables greater mobilisation and higher turnout. Multi-party systems that generate short-term alliances among parties on either side of the referendum issue depress turnout by creating coordination problems among coalition partners.

Our paper employs cross-national data as well as an intensive case study to test our model. We test the generalizability of our arguments using an original dataset of more than 150 referendums in more than 50 countries. We construct an original measure of issue salience for each referendum to test the main alternative explanation: that salience explains the variation in turnout. We show that salience is an important factor, but it does not tell the whole story. Our cross-national study also allows us to test propositions related to party-system size and the candidate election-referendum turnout gap.

We further probe our model with an in-depth case study is of the Colombian referendum of 2016, in which voters were asked to ratify peace accords. To explore the dynamics of mobilization in referendums versus local elections, we interviewed more than 50 Colombian politicians, campaign managers and staffers who were involved in the referendum campaign, along with other local experts. To gain further insight into mobilization in candidate elections in Colombia, in October 2019, we also accompanied candidates for city council and department (state) governor in their campaigns for local elections. Data from the Latin American Public Opinion Project (LAPOP, 2016) supplement our qualitative evidence on the Colombian case.

Our findings suggest that mobilization by parties is central in securing participation in referendums, and internal party structures and incentives have a crucial impact on turnout. Consequently, it is misleading to think of referendums as instances of direct democracy, unmediated by party politics. Instead, they are in fact prone to many of the same hazards and pitfalls as any form or process of representative government.

Authors

• Radha Sarkar, Yale University

• Susan C. Stokes, University of Chicago

• Eli Gavin Rau, Princeton University