Voting and peer effects


Published 27.03.18
Photo credit:
Commonwealth Secretariat

A study in Mozambique shows how peer effects can increase information and interest in politics but at the same time reduce voter participation

Why do people vote? The individual rationality of voter turnout in political elections has often been questioned – unless a person casts the deciding vote, voting has no effect on the outcome (e.g. Feddersen 2004). If no one votes, however, the electoral outcome is unlikely to reflect the preferences of the electorate. As a consequence, voting is often seen as a civic duty. Although some countries (e.g. Belgium, Brazil, and Peru) make voting a legal obligation, most do not. The level of electoral participation is therefore expected to depend on the social norms that are in place regarding voting at the time, as well as on the probability voters attribute to being the one casting the deciding vote. Influence by peers may affect both.

While the number of political elections has increased worldwide since the mid-1980s, voter turnout has experienced a rapid decline (World Bank 2017: 228). This has been particularly noticeable in sub-Saharan Africa where, according to the 2010-15 Gallup World Poll, less than 45% of the population has confidence in the integrity of elections. Mozambique is a good illustration of this distressing phenomenon. At the first elections held in 1994, turnout was 88%. Since then, Frente de Libertação de Moçambique (FRELIMO) and its sponsored presidential candidates have won all national elections, increasing their vote share over time. Ten years later, in 2004, turnout had fallen to 36%.

A campaign to increase voter turnout in Mozambique

In an attempt to reverse this trend, a randomised controlled trial was organised during the 2009 elections to study the effect of a voter education campaign. The hope was that the campaign would re-establish confidence in the electoral process, thereby increasing turnout. At face value, the campaign was successful. On average, it increased voter turnout by 5% in treated locations (Aker et al. 2017). The campaign took three different forms. The first was the distribution of a free newspaper that focused on neutral information about the elections. The second was a text messaging hotline to which citizens could report electoral problems. The third was civic education delivered via a leaflet and text messages providing information about the elections.

In a follow-up paper (Fafchamps et al. 2018), we examine whether the success of the intervention is partly due to positive peer effects – if citizens vote to fulfill a civic obligation, by making this obligation more salient the education campaign may generate social pressure to participate. We focus on the peer effects triggered by the campaign within each village and we investigate whether the effect of the campaign is stronger for those who are socially or geographically close to targeted individuals in the village. In terms of outcome variables, we use survey measures of individual turnout, information and interest in politics, and a behavioural measure of political participation. To estimate the peer effects, we use three measures of social and geographical proximity. The first, kinship, corresponds to the proportion of people sampled in the village who are relatives of the respondent. The second, chatting, corresponds to the proportion of people sampled in the village with whom the respondent talks on a regular basis. Third, geographical proximity is proxied by the (negative of) the average distance between the respondent’s house and the houses of the other people sampled in the village.

The impact of peer effects

As expected, the education campaign, on average, increased the voter turnout (Aker et al. 2016). This effect, however, is smaller for more central individuals. We find negative peer effects on voter participation. In contrast, peer effects on information and interest in politics are positive and in line with the average effect of the campaign.

We interpret these findings as consistent with a model of costly political participation. In this framework, voter turnout may be induced either by the probability of affecting the electoral process, or by non-instrumental motivations such as civic-mindedness. By giving information about the credibility of the elections, the campaign intends to reassure voters about the integrity of the process. In so doing, it may also raise civic-mindedness. Both effects are conducive to increased turnout, in line with the average effect of the campaign. However, peer effects can mitigate this positive influence on turnout if central voters realise that, because of the campaign, turnout will increase and their vote is less necessary to achieve a politically acceptable electoral result. These voters may then decide to free ride on others’ political participation.

In Mozambique, there was no doubt about who would win the 2009 election. Thus, voters could not have expected to be pivotal in deciding the election result. Nevertheless, voters might care about other electoral outcomes such as the village turnout or win gap. It is possible – even likely – that a low turnout or win gap would be interpreted as indicating disapproval of the government. This in turn might lead to some form of punishment (e.g. reduced supply of local public goods). In such a political environment, being pivotal does not mean casting the ballot that determines who wins the election, it means bringing the turnout or win gap above the threshold below which the community faces reprisals. Our data suggest that pivotal reasoning based on the win gap was one of the determinants of political participation in Mozambique. This evidence corroborates our interpretation of the negative peer effect on participation as free-riding through pivotal reasoning.

These results have implications for the design of voter education campaigns. While social networks tend to magnify treatment effects on soft outcomes such as interest in elections, they can attenuate turnout by circulating information about voting intentions, thereby triggering free-riding.

Editors' note: This column first appeared on


Aker, J C, P Collier, and P C Vicente (2017), “Is information power? Using mobile phones and free newspapers during an election in Mozambique”, Review of Economics and Statistics 99(2): 185-200.

Fafchamps, M, A Vaz, and P Vicente (2018), “Voting and peer effects: Experimental evidence from Mozambique”, Economic Development and Cultural Change, (forthcoming).

Feddersen, T (2004), “Rational Choice Theory and the Paradox of Not Voting”, Journal of Economic Perspectives 18(1)" 99-112.

World Bank (2017), World Development Report 2017: Governance and the Law, IBRD/The World Bank, Washington DC.