The authors worked with Kenya’s election commission to design an experiment in which 1,674 communities were randomized into a control group or into combinations of three interventions: civic education, SMS reminders about registration opportunities, and local registration visits by election commission staff. The main outcome of interest was voter registration. While there were few measurable effects of civic education or SMS alone, local registration visits increased voter registration, particularly in poorer, more distant, and sparsely populated communities.
The study was conducted eight months prior to the 2017 Kenyan general elections. As in most other African countries, in Kenya only the electoral commission has the authority to register citizens to vote. Citizens may register at any time of the year, but this system has been known to generate inequalities in voter registration because 1) not all citizens can bear the cost to travel to a constituency election commission office; 2) not all citizens may know about registration opportunities; and 3) political interests may sponsor citizens to register in some areas. Indeed, the authors find that poverty rates, distance from the election office, and population sparsity are negatively correlated with voter registration at polling stations in the eight-month period prior to the onset of the experiment.
The authors worked with Kenya’s Independent Electoral and Boundaries Commission (IEBC) to develop interventions that might increase voter registration, particularly in communities that are poor, sparsely populated, and far from election offices. The study was conducted in seven of Kenya's 47 counties. Of the 3,828 polling stations within these counties, 1,674 were selected for inclusion in the study using a sampling strategy that blocked on a polling station area's level of poverty, population density and distance from the election offices. Blocking allowed the authors to estimate treatment effects across these different contexts.
The IEBC implemented three types of interventions at the polling station level. In the first, called localization, communities were visited for two days by IEBC staff with portable voter registration equipment to register voters from 9am until 5pm. In the second, called civic education (labelled as canvassing in the paper), IEBC staff visited communities for two days to provide basic education on the voter registration process, to discuss the importance of electoral participation, and to answer any questions. In the third intervention, called SMS, the IEBC sent two text messages over two days to already registered voters asking them to encourage their unregistered friends and family to register and identified the nearest registration opportunity.
Polling stations were randomly assigned to a combination of these three interventions and a control condition in a 2x3 factorial design as shown in Table 1.
Table 1. Number of polling stations by treatment status.
The authors find little evidence that the SMS or civic education interventions alone increased the number of registrations compared to control communities in the 1.5 month following the interventions. Localization, on the other hand, increased voter registration, measured as the total number of registered individuals divided by the number of voters during the previous election, by 2 percent. Moreover, the combination of localization and SMS led to a 2.4 percent increase in registrations relative to control communities. The study also shows that context matters. The effect of localization was six times as large in the poorest quintile of communities relative to the richest, almost one and a half times as large in distant polling stations relative to proximate polling stations, and almost six times as large in sparsely populated areas relative to densely populated areas. Finally, the authors also explored the downstream impact of the interventions on Kenya's 2017 general elections. While the authors find few effects on registration and turnout, there is some evidence that suggests that local registration increased electoral competition and vote preference diversity in down-ballot contests.
The evidence from this study suggests that status quo voter registration policies may disenfranchise certain populations, particularly those who live in poor, distant, or sparsely populated communities. The findings also suggest that in the context of fledgling democracies, interventions related to convenience and cost may be more effective in increasing electoral political participation than those related to information and civic education.