PDF version

Using Technology to Promote Electoral Participation in South Africa

Location: South Africa
Sample: 90,646 registered users of the VIP:Voice platform
Timeline:
2014
Outcome of Interest: 
Electoral participation
Voter behavior
Intervention Type: 
Community participation
Digital and mobile
AEA RCT Registration Number: 
Partners:

Healthy democracies require that citizens participate in political life, from turning out to vote to monitoring government performance.  In young democracies, there may be numerous individual or institutional barriers to participation. Researchers conducted a randomized evaluation of the VIP:Voice, a digital election participation platform, to measure the impact of information and communications technologies on individuals’ political participation. Via the platform, researchers interacted with nearly 100,000 citizens, recruited citizen monitors in 38 percent of the wards of South Africa and deployed 347 citizen monitors to polling places. However, around 50 percent of participants dropped out at each subsequent, and more time-intensive, form of engagement requested via the platform. 

Policy Issue 

Healthy democracies require that citizens actively participate in political life, from turning out to vote to monitoring government performance. Individuals in emerging democracies confront numerous institutional and personal obstacles to participation that can marginalize them from political processes. Low education levels, limited financial resources, or geographic remoteness may increase individuals’ exclusion from democratic processes. Activists, donors, governments, and NGOs pursue various strategies to help citizens overcome barriers to action, including rallying support for protest and mass action, educating and mobilizing voters, monitoring elections, and reporting on corruption.

In recent decades, information and communication technology (ICT) and digital media (DM) has revolutionized the work of organizations promoting political participation in emerging democracies. ICT and DM provide a relatively inexpensive means for communication and coordinating collective action by a large and dispersed user base. Despite its potential, citizen use of ICT/DM may still encounter significant barriers in developing countries. Literacy, connectivity, and the user costs of accessing ICT/DM may limit the participation of precisely those citizens who are already excluded from political activity. Can new technology be used to improve the quality of democracy by boosting citizen participation?

Context of the Evaluation 

In South Africa, the 1994 elections brought an end to apartheid, allowing for universal franchise and energizing democratic participation on the part of the non-white majority for the first time. Like voters in many other settings South African voters participate at differential rates. Since 1994, the ruling African National Congress (ANC) has received strong majorities in all elections, potentially discouraging participation since elections are seen as foregone conclusions. The 2014 national and provincial elections took place on a backdrop of rising dissatisfaction with the ANC and the incumbent president Jacob Zuma, but had the lowest voter turnout in the post-apartheid era. Election monitoring groups generally rate South Africa’s Independent Electoral Commission (IEC) highly, but because of the lack of significant electoral competition, incentives for citizens to participate actively in elections may be low.

As of the 2011 census, cell phone saturation in South Africa was almost 90 percent and has since risen to almost 100 percent, and South Africa has the fifth highest rate of internet access in Africa, making ICT and DM interventions feasible.

Details of the Intervention 

Researchers conducted a randomized evaluation of an ICT election participation platform to measure the impacts of ICT on individuals’ political participation. Working closely with Praekelt, a major South African technology firm, researchers designed a multi-channel ICT/DM platform, VIP:Voice. The platform invited users to share their opinions on the elections and to serve as citizen reporters monitoring election results. Users could interact with VIP:Voice through social media, several messaging apps, or through SMS/USSD messaging services on basic phones. Researchers implemented the VIP:Voice platform in four phases, maintaining the sample participants who took up the platform in each subsequent phase, and introduced experimental variation in each stage.

Phase one began with the launch of the platform four weeks before the election and a recruitment drive through USSD messages, social networks, GTalk and Twitter. Recruitment text messages were randomized into three formats to encourage registration: a standard message encouraging registration but paying full messaging costs to interact with the platform, an offer of free access with all messaging fees covered, and a lottery arm offering a chance to win 55 Rand ($5).

In Phase two, the platform invited registered individuals to provide their demographic data, incentivized with a lottery for all participants, and report on election-related events leading up to election day.

To improve electoral transparency, electoral law in South Africa requires polling center managers to post tally sheets. Phase three offered incentives to individuals to act as volunteer elections monitors who would report vote totals from their polling places the day after the election to evaluate whether locally posted results matched the centrally reported tallies. Researchers randomly offered incentives of 5 Rand ($0.50) or 50 Rand ($4.60) to participate as citizen observers.

Finally, phase four tested different get out the vote messages and polled voter perceptions of the election day process, incentivized with a lottery, with all 78,108 individuals who had registered on VIP:Voice. In the get out the vote experiment, researchers randomly assigned individuals in treatment groups to receive either a message on election day reminding them to vote, a reminder to vote plus a message reminding citizens that their inked finger would show others that they had voted—adding social pressure to vote. On the day after the election, participants received a text message asking whether or not they had voted and a follow-up message asking about ballot color and requesting a photograph of their inked fingers.

Results and Policy Lessons 

The total recruitment effort led 263,000 individuals to contact the platform, of whom 90,646 completed the necessary steps to register.

Recruitment through different platforms generated user groups with different gender and racial compositions. The group of individuals recruited through USSD messages was significantly more black and female than the national population. Almost two-thirds of USSD users were women and 94 percent were black as opposed to the national averages of 50 percent women and 79 percent black.. In contrast, the group recruited through MXIT, a popular social media site, was much whiter and more male than the national average. These demographic differences by platform indicate that the success of efforts to overcome political marginalization using ICT may depend on the channel used.

Through the platform, researchers recruited citizen monitors in 38 percent of the wards of South Africa and deployed 347 citizen monitors to polling places. However, each action that we asked participants to engage in led to an average attrition rate of roughly 50%, and this attrition was greatest when we asked individuals to move from digital forms of political engagement to ‘real-world’ behaviors such as monitoring their polling places.

Incentives influenced behavior of VIP:Voice users. Incentives for participation increased registration for the platform, volunteering as citizen observers, and in-person election monitoring. When offered free service or the lottery as an incentive, 1 in every 1,111messages led to a registration compared to 1 in every 1,900 without incentives. Among users who accessed VIP:Voice through USSD messages, 3.4 percent volunteered to serve as citizen observers in the absence of incentives compared to 5.4 percent of those in the incentive groups. Of those who monitored elections, the rate of successful entry of voting data via SMS almost tripled from 4.2 to 14.6 percent for those offered the larger incentive.

Ferree,Karen E., Clark C. Gibson, Danielle F. Jung, James D. Long, and Craig McIntosh. "How Technology Shapes the Crowd: Participation in the 2014 South African Election." Working Paper, April 2016.