In the weeks leading up to the November 2012 elections in Sierra Leone, villagers gathered to watch a 45-minute debate among candidates for parliament. The debaters were not there in person, but rather in a video projected on the outside of the local polling center. For Katherine Casey, an assistant professor of political economy at Stanford Graduate School of Business, who was one of about 250 people watching in one village, the attentive crowd offered hope for Sierra Leone's restored democracy — and may even provide lessons that can be applied elsewhere.
The lesson, she suggests, is that providing voters with information about candidates can help create the conditions that encourage the election of better leaders, the dampening of ethnic rivalries, and, as a result, a strengthened democracy. Sierra Leone has now held several peaceful elections, a far cry from the intimidation and violence that racked the country of 6 million between 1991 and 2002, when a civil war took 50,000 lives.
The candidate debate nights were part of an experiment devised by Casey and colleagues to test what sort of information affects how citizens vote. In previous research, they had found clues that better information plays a significant role. In national elections in 2007-2008, 86% of Sierra Leone's voters voted for the party historically associated with their ethnic group, but in local elections only 75% did so. Exit polls pointed to a possible explanation: Voters knew twice as much about local candidates as national candidates, and that additional knowledge about individuals, as opposed to just party labels, may have played a role in their willingness to cross over ethnic lines on their ballots.
The ability of information to change voters' choice matters for two reasons, Casey believes. If political parties cannot rely on people voting for traditional affiliations, such as an ethnic or religious group, they may be more likely to field candidates who pay closer attention to what policies constituents want. And, armed with more knowledge, voters may be better able to hold politicians accountable, which makes it more difficult for those elected to divert public resources.
But getting information to voters in rural villages, with limited access to radio, television, and newspapers, is no easy task. Moreover, in a setting where 70% of voters have no formal schooling, it is critical to deliver information in a way that is accessible and engaging. One of Casey's ongoing projects, with partial funding from the Stanford Institute for Innovation in Developing Economies (SEED), is to tease apart what sort of information makes a difference. To find out, the researchers teamed up with Search for Common Ground, a nonprofit organization, to organize moderated debates among candidates for 14 of the most closely contested national parliament seats in 2012. Videos of the relevant debate were shown in 112 villages. Other voters were asked to listen to the radio report, and still others were shown a "get-to-know-you" video of candidates talking about themselves.
The results showed a dramatic effect of the debate video. In villages where voters had not seen the video debates, exit polls indicated "only 28% of voters could name the two parliamentary candidates in the election they had just voted in," Casey says, "and only 3% knew roughly the amount of money [$11,000] that members of parliament are given every year" to spend in their constituency.
Among those who saw the video debates, many more voters knew the candidates' names and other details about them, such as who was better educated or had held office before. The number who knew the amount of money the elected candidate could spend quadrupled.
Furthermore, Casey said, those who had seen the debates were more likely to vote for a candidate whose policy stance was the same as the voter's policy preference. "Across the board, we had large positive impacts on people being able to tell you where each of the candidates stood on important issues such as health care and gender equity."
The result was a change in the vote totals. "The candidates who won the debates, according to various ways we had measured the winners, got almost a five percentage point increase in votes," Casey said. "Also, the candidates responded to these debates by stepping up their campaigning in the villages that saw debates. They gave more gifts, like T-shirts and posters, and made more in-person visits."
In another 40 villages, the researchers measured the impact of a radio journalistic account of the debates and of the personality videos to see what kinds of information were influential. Some people were asked to listen to a journalist summarizing all the policy stances and professional qualifications covered in the debate. "It was just straight facts and excluded the candidates' charisma," Casey said. "Others watched the get-to-know-you video, where the candidates talked about their families and their hobbies, but nothing about their professional qualifications or policy stances. It was pure personality."
In exit polls, the researchers learned that the 5-minute personality video allowed voters to pick up on which candidate was more educated and which one had public office experience, but they did not know more about the policy stances of the candidates. The radio report increased voters' knowledge of politics, candidates, and policy positions, although somewhat less than watching the debates.
"Only witnessing the debates actually changed voting choices," Casey said. "More people voted for the candidate who performed the best, and they moved into policy alignment with their preferred candidate."
For Casey, voting experiments are part of a larger investigation into the most effective ways to provide assistance to developing countries. She is part of a generation of economic and political development scholars who advocate randomized field experiments to guide decision making. Just as medical practice dramatically improved after the adoption of research protocols that randomly assigned some patients to experimental treatment and others to standard or no-treatment control groups, Casey and her peers believe donor interventions in extremely poor countries will be more effective if rigorously evaluated.
In another experiment in Sierra Leone, she and her colleagues used randomization and control groups to look at the value of infrastructure loans and grants given at the very local level of villages with about 300 residents, or 50 households, instead of at the level of districts serving about 10,000 residents. The idea behind decentralization, Casey says, "is that people at the local level may have a lot of information that the central government doesn't have about what they need and how to accomplish it."
Four years after the World Bank and national government distributed grants, the researchers found that in those villages receiving the grants, there were more public goods, such as schools, latrines, and traditional birthing centers, and that the quality of those goods was better than those of the control villages, Casey says. "This was physical assessment of things like what the roof is made out of, what the walls are made out of, and the overall condition of repair. They had more facilities of better quality and also more petty traders and goods available to purchase, which are indicators of economic well-being." The study did not, however, show any impact on a complementary goal of enhancing the role of traditionally excluded groups — such as women and younger men — in village decision making, underscoring the difficulty in changing local power structures and governance institutions.