From Afghanistan to Zimbabwe, an inked finger symbolizes participation in the electoral process. Indelible election ink offers a low-cost, easy to implement solution to the problem of double voting in countries with unreliable voting records and weak administrative capacity. Its use has therefore become standard practice in many emerging democracy elections. However, while the putative goal of inking is to decrease electoral fraud, it may also enable politicians and others to manipulate turnout. In this study, we articulate a theory linking ink to turnout behavior and present supporting evidence from a field experiment conducted in Uganda during the 2011 election. Our experiment was exploratory in nature, with limited scope and size. Nonetheless, it demonstrates the plausibility of a relationship between election ink and turnout.
Inking renders the decision to vote visible. An inked finger identifies a voter not just when she emerges from the polling station, but until the ink on her finger wears off. During this time, which can last for days or even weeks, her decision to vote is public. Interested actors can condition rewards and punishments on an observable indicator of behavior: the inked (or uninked) finger. Ink thereby enables turnout manipulation, which can aim either to encourage or discourage participation.
Anecdotal evidence from elections in several contexts supports the intuition that election ink facilitates turnout manipulation. In Zimbabwe, militias supporting President Mugabe used ink to encourage turnout in the second round of Zimbabwe’s 2008 presidential election. Going door-to-door in Harare townships, they promised ‘the most ferocious retribution’ for anyone lacking an inked finger (BBC News, 2008). In other places, ink enabled turnout discouragement. During the 2009 and 2010 Afghan elections, the Taliban threatened anyone with an inked finger with its amputation (or worse). In some instances, they carried out their grisly threat. Election observers noted these intimidation efforts kept many voters at home. They also observed voters begging polling station staff not to ink their fingers (Democracy International, 2010, 2011). Similar anecdotes surface about Indian elections. In Kashmir, one journalist reported during the 2002 elections that ink that is a ‘nuisance’ in most parts of India was ‘virtually a passport to death in Kashmir’ because militants threatened retribution to voters found with inked fingers (Jha, Reference Jha2002). During the lead-up to the 2013 elections in Naxal-affected districts of Chhattisgarh, election officials requested exemptions from the use of ink from the State’s Chief Electoral Officer, specifically citing depressed local turnout in the 2008 elections due to Naxalites threatening to cut off inked fingers (‘Move’, 2013).
Moving beyond anecdotes to evaluate more systematically the relationship between electoral ink and turnout presents challenges. Depending on the mixture of electoral strategies in a particular election, the effects of ink may vary, increasing turnout in some instances, depressing it in others; ignoring this heterogeneity may cause us to underestimate ink’s full impact. Virtually universal within-election adoption of ink further complicates evaluation. Not only is there little variation in observational data, but randomization of the inking practice itself seems unfeasible. Finally, few credible measures of turnout exist in emerging democracies. Standard practice involves asking people if they voted, an approach confounded by strategic misreporting and non-response.
To circumvent these challenges, we implemented a Get Out the Vote (GOTV) experiment to evaluate the relationship between ink and turnout. In the experiment, conducted in Uganda’s capital city Kampala during the country’s February 2011 national elections, we reminded people to vote and provided information to registered voters about the use of ink in the election. We also explained that ink would mark them for several days, revealing their decision to vote to other members of their community. This information was new to some, altering knowledge about electoral practices; for others, it may have primed the salience of an aspect of the election that they knew about but had not carefully considered. We find that the inking treatment reduced the probability of having an inked finger, especially for younger and less educated voters – precisely the set least likely to be familiar with inking practices and therefore most likely to respond to an informational treatment. As the scope of our study was modest, with a sample size of just over 1,100 individuals from one constituency in one country, we do not want to overstate the generalizability or conclusiveness of our findings. Nonetheless, our study provides preliminary evidence of a link between the largely unexamined practice of election inking and voting behavior.
Our study suggests that practices, like inking, which publicize turnout may not be benign in all settings. Experiments in the United States routinely find that increasing turnout’s visibility raises participation (Gerber, Green, and Larimer, Reference Gerber, Green and Larimer2008; Davenport et al., Reference Davenport, Gerber, Green, Larimer, Mann and Panagopoulos2010; Mann, Reference Mann2010; Sinclair, Reference Sinclair2012), but these results may not generalize to settings with different norms about participation and dissimilar electoral pressures. Our results echo findings in studies by Daxecker (Reference Daxecker2014), Ichino and Schündeln (Reference Ichino and Schündeln2012) and others suggesting that policies adopted to improve election quality in one domain can have unintended consequences in others.Footnote 1 Moreover, relatively innocuous interventions aimed at promoting electoral participation may backfire in some contexts (Driscoll and Hidalgo, Reference Driscoll and Hidalgo2014). Studies of turnout and other types of voting behavior in emerging democracies might benefit from these insights, which highlight the value of considering institutional features of elections in addition to individual-level factors like age, education and partisanship emphasized in most extant work (Bratton, Reference Bratton1999, Reference Bratton2008; Bratton, Mattes, and Gyimah-Boadi, Reference Bratton, Mattes and Gyimah-Boadi2005; Isaksson, Reference Isaksson2014; Kuenzi and Lambright, Reference Kuenzi and Lambright2011, Reference Kuenzi and Lambright2007). Only by placing voters in a specific political context can we make sense of their behavior.
Inking as a Standard Electoral Practice
Use of election ink dates to the early 1980s, when election officials began marking voters with ink detectable only with ultraviolet light. Difficulties with administration (particularly, equipping polling stations with UV-lights) led to the adoption of visible, indelible ink as the standard by the mid-1980s (Cody, Reference Cody1985; Meislin, Reference Meislin1982; Taubman, Reference Taubman1982; Wattenberg, Reference Wattenberg1986). Inking has since become an international norm, particularly in emerging democracies outside the OECD. Nearly 30 per cent of all countries have used it, and it is particularly common in South, Central, and Southeast Asia, the Middle East, and sub-Saharan Africa. It is less common in Latin America and rare or non-existent in long-term democracies. See Figure A1 for all countries where we have found evidence of election inking.
The attractions of inking are straightforward. Establishing accurate voter registries is costly, difficult, and often controversial (Organization for Security and Cooperation in Europe (OSCE), 2010). Would-be voters in newer and poorer democracies frequently lack valid forms of identification to use for voter registration or voting. Given these limitations, multiple voting presents a serious threat to election quality. Ink offers a low-technology, inexpensive solution easily delivered with election materials and not requiring extensive training or electricity to implement.Footnote 2 Polling station staff need only mark a finger (or other part of the hand) with indelible ink after voting. Ink is typically bright purple and easily seen on all skin tones. As of this writing (late 2017), alternatives like biometric registration remain in the development stage in most countries.Footnote 3
Controversies periodically surface regarding election ink. The Malaysian government responded to public concerns about multiple voting by purchasing nearly 50,000 bottles of indelible ink from India prior to the 2008 election. It then reversed this policy and abandoned inking, citing claims that the ink might be used to mark unsuspecting citizens prior to election day, preventing them from casting votes (Weiss, Reference Weiss2009). Malaysia’s ink woes continued in its 2013 General Election, when ‘indelible’ ink proved removable with soap and water (Ostwald, Reference Ostwald2013; Welsh, Reference Welsh2013). Political operatives in the Philippines reportedly gave rewards to potential opposition voters who allowed them to ink their fingers, disqualifying them from voting (Schedler, Reference Schedler2002). Voting station workers used non-permanent ink during Afghanistan’s 2004 elections, leading to extensive allegations of ink removal and double-voting (Raman, Reference Raman2004; Baldauf, 2004). Inking controversies continued in subsequent Afghan elections, with questions about ink quality and inconsistent application (Democracy International, 2010).
Despite its controversies and pervasive use, we have found no systematic study of inking in either the policy or academic literatures. This lack of attention is puzzling because, as we suggest in the next section, inking may have significant consequences for voting behavior. Specifically, we argue that inking can facilitate the manipulation of turnout.
Inking and Turnout
Ink helps polling station officials detect and deter double voting, but in the process makes the decision to vote visible beyond the polling station, typically for days or even weeks. During this time, a range of actors – spouses, bosses, neighbors, religious leaders, community activists, party leaders, politicians, soldiers, police – can observe whether an individual has a marked or unmarked finger and from this infer his or her decision to vote.Footnote 4
Ink facilitates turnout manipulation by lowering the costs of turnout monitoring. External actors can now condition rewards or punishments on the presence or absence of ink. These rewards and punishments may assume many forms, from the subtle (a raised eyebrow, a verbal rebuke, shunning) to the direct (cash payments, physical injury, loss of employment) and may be legalFootnote 5 or illegal in nature.Footnote 6 They may or may not involve coercion and threats of violence.
External actors have a range of motivations when attempting to manipulate turnout. We highlight two: encouragement and discouragement. An encouragement strategy tries to increase turnout of a voter or group of voters. Parties frequently engage in encouragement strategies vis-à-vis their partisans, especially in close elections, to enhance their chances of winning. The Obama campaigns of 2008 and 2012, for example, made targeted attempts to raise turnout of African American voters, students and other groups strongly predisposed to vote Democrat. Nichter (Reference Nichter2008) refers to such strategies as ‘turnout buying’ when they involve an exchange of benefits for voting. Turnout encouragement can also take on a coercive tone, as it did in the example from Zimbabwe discussed earlier, where pro-government police threatened retribution to citizens who could not show a marked finger.
The individuals encouraging turnout can be ordinary citizens rather than politicians. Citizens may encourage turnout in a bid to attract the attention and support of powerful patrons, or to enforce community norms about political participation. The encouragement mechanism involved in Gerber, Green, and Larimer (Reference Gerber, Green and Larimer2008) is likely normative and community based – the desire to be seen as a good citizen by friends and neighbors – rather than explicitly electoral in nature.Footnote 7
Parties and other actors may also engage in discouragement strategies. In addition to the Afghan and Indian examples discussed earlier, Cox and Kousser (Reference Cox and Kousser1981) document cases of parties paying opposition partisans to stay home in late nineteenth-century New York state. The Democratic Party used a variety of coercive methods to discourage African American voting in the pre-Voting Rights South (Kousser, Reference Kousser1974). Collier and Vicente (Reference Collier and Vicente2014) and Bratton (Reference Bratton2008) discuss efforts by Nigerian parties in the 2007 election to discourage opposition turnout via intimidation. A similar dynamic may have transpired in Colombia in the 1990s, where Steele (Reference Steele2011) documents a long-term strategy by counter-insurgents to displace voters in areas likely to vote for the insurgent-backed political party, and in Pakistan’s 2013 election, when the Pakistani Taliban targeted leaders and candidates from three secular parties and warned people to stay away from their rallies.Footnote 8 Parties in the Panchthar region of Nepal threatened and intimidated voters to keep them home. Affected villages recorded turnout levels under 2 per cent. Observers noted that even weeks after the elections ‘citizens of one of the communities were still fearful of reprisals against those who had managed to vote’.Footnote 9 Normative turnout discouragement strategies are less commonly discussed in the literature, but community-based election boycotts likely employ these (see Beaulieu, Reference Beaulieu2014).
Regardless of whether an actor’s goal is encouragement or discouragement, to further electoral goals or enforce community norms, election ink makes the task of monitoring – and therefore manipulating – turnout easier. It reveals turnout decisions just as surely as Gerber, Green, and Larimer’s experimental letter treatment, but in a far broader and more public fashion. It therefore seems plausible that ink might also influence turnout decisions. Moreover, the effects of information revelation may not always be turnout encouragement. Publicizing past turnout decisions through letters, public voting records, or ink is likely to depress rather than increase turnout where turnout discouragement prevails.
In sum, the use of ink in elections may lead to changes in voter turnout by facilitating turnout manipulation, which can attempt to raise or lower turnout. Inking lowers the cost of all such strategies by rendering the voting decision public, making it easier for outside actors to incentivize desired voting behaviors. Anecdotal evidence, as well as results from American GOTV experiments, support these claims.
Evaluating Inking
Rigorous empirical evaluation of the practice of inking presents challenges. Inking likely has context-specific effects that depend on the turnout manipulation strategies present in a particular place. Ignoring this heterogeneity may wash out treatment effects, but accounting for it directly can be difficult if turnout manipulation is subtle or hidden. Furthermore, the practice of inking voters has been adopted so universally that variation in treatment is rare. Election officials periodically fail to ink some voters because of political controversy or insufficient ink, but these instances are almost certainly not random, leading to standard problems with identifying treatment effects. Nearly universal adoption of inking and the importance placed on the practice by election officials would also seem to rule out direct randomization of inking itself.
Measuring turnout brings its own difficulties. Official voting records detailing individual turnout (as in the US) rarely exist in developing countries. Moreover, social norms in many places strongly favor voting, leading to potential social desirability bias in self-reports of voting. Bratton, Mattes, and Gyimah-Boadi (Reference Bratton, Mattes and Gyimah-Boadi2005, 145–6) show self-reported turnout for Afrobarometer respondents is, on average, 13 per cent higher than official turnout.
We circumvent these challenges through an experiment conducted during the February 2011 Ugandan election in a central Kampala constituency. We randomly exposed participants to reminders to vote and information about the use of ink in the election via a face-to-face intervention just before the election. We then evaluated the effect of this intervention on turnout by returning a few days after the election to check individuals’ fingers for ink.Footnote 10 The goal of the experiment was twofold: first, to evaluate whether face-to-face reminders to vote would raise participation; and second, to evaluate the effects of inking on turnout. As our study was exploratory and small in scale, we chose a constituency that was accessible with middling levels of turnout in prior elections, to avoid ceiling or floor effects that would make changes in turnout difficult to detect.
Our design had several advantages and a few limitations. We reduced the problem of strategy heterogeneity by restricting the experiment to a single constituency. We avoided the problem of response bias by measuring turnout through observation of inked fingers. And we mitigated standard problems with observational data by pursuing an experimental design. One limitation, however, is that our focus on a single constituency reduces external validity. Our findings apply to a particular setting, discussed below. Were we to run the experiment in alternative contexts, we would expect different outcomes.Footnote 11 A second limitation of our approach is that the informational ink treatment is necessarily weaker than a treatment directly manipulating the application of ink. It may only affect respondents who previously lacked information or awareness about the practice. Our results thus likely understate the true effects of inking. Nonetheless, given the challenges of non-experimental approaches and the difficulty of manipulating the use of ink itself, the informational treatment offers a valuable if indirect window on an otherwise impossible to study practice. In the next sections we discuss the experimental setting, treatments and design.
Setting
We conducted the experiment in Kawempe South constituency in the heart of Kampala, Uganda’s capital city. Kawempe is a large, socially, economically and ethnically diverse area. Built over a flood plain and settled by waves of migrants, it contains dense neighborhoods of closely spaced small one or two room structures, many informal, with some larger homes interspersed. Markets, schools, churches, mosques, clinics, farms and small businesses coexist with residences. Small streets and paths meander and branch off roads clogged with buses, motorcycles, pedestrians, cars and animals. Kawempe is also home to Makerere University, Uganda’s premiere post-secondary institution. It is neither particularly poor nor rich relative to other urban areas of Uganda.
Kawempe’s history of oppositional politics differentiates it from much of Uganda. The National Resistance Movement (NRM) has dominated Ugandan politics since the mid-1980s when its leader, Yoweri Museveni, seized power during a civil war. After resisting pressures to democratize, Museveni removed a ban on multiparty politics in 2005. He won the presidency in the country’s first multiparty election in many decades in 2006, defeating perennial rival, Kizza Besigye of the Forum for Democratic Change (FDC) 59.3 per cent to 37.4 per cent. Ironically, political repression and consolidation of power around Museveni appear to have increased in the period after the introduction of multiparty competition (Barkan, Reference Barkan2011; Makara, Rakner, and Svåsand, Reference Makara, Rakner and Svåsand2009; Mwenda, Reference Mwenda2007; Tripp, Reference Tripp2010). Museveni bolsters his dominance through a ‘carrot and stick’ mixture (Conroy-Krutz and Logan, Reference Conroy-Krutz and Logan2012) of good economic performance, control over state resources and service delivery, constitutional manipulation, lavish campaign spending, election fraud, and intimidation of the opposition, all in the context of a highly militarized state(Juma, Reference Juma2011; Barkan, Reference Barkan2011; Tripp, Reference Tripp2010; Mwenda, Reference Mwenda2007).
Fractured, weak and underfunded, the Ugandan opposition has had few means to challenge the NRM in most of Uganda (Tangri and Mwenda, Reference Tangri and Mwenda2010). Rural areas in particular remain firmly tied to Museveni (Barkan, Reference Barkan2011). In Kawempe, however, the opposition won a majority in 2006, supporting Besigye with 55.5 per cent of the vote to Museveni’s 41.4 per cent (Electoral Commission, 2006). In 2011, Museveni once again faced an opposition that was weak and divided throughout most of the country, but was relatively strong in the capital city (Conroy-Krutz and Logan, Reference Conroy-Krutz and Logan2012). Kawempe thus represents an opposition-leaning area in a single party dominant, semi-authoritarian state with a long history of political repression. As Barkan (Reference Barkan2011, 7) writes: ‘Kampala serves as something of a “Potemkin Village of Democracy” – impressive at first encounter, but not representative of the country or the regime’.
Our results may therefore be particular to Kawempe and likely do not generalize to Uganda as a whole. Kawempe is urban and politically diverse. Rural, politically homogeneous areas might exhibit different dynamics. In places firmly dominated by the ruling party, we would expect turnout encouragement to prevail and ink to have turnout enhancing effects. In opposition strongholds we would expect turnout discouragement by the ruling party. Kawempe is somewhere between these extremes. It leans opposition, but the NRM also has a strong and visible presence, typically winning at least 40 per cent of the vote. It is politically vibrant, and we are aware of no tradition of election boycotts or norms disfavoring political participation. We therefore anticipate that the effects of ink are weaker in Kawempe than in many other contexts, and our choice of case makes null results more likely, if anything.
Treatments
The experiment had three treatments: a non-political ‘control’ treatment, a ‘reminder to vote’ treatment, and an ‘ink prime’ treatment. Table 1 summarizes treatment groups. All were delivered face-to-face via household survey. We trained interviewers to deliver treatments in a neutral, informative fashion, and to emphasize that they represented an American research organization.
Treatments and control
The control consisted of a short, benign set of questions on demographic factors (age, education) and mobile phone usage. The ‘reminder to vote’ treatment included the mobile phone survey, plus a generic message reminding the recipient that there would be presidential and legislative elections on 18 February. It also stated ‘all registered voters will have the opportunity to vote in this election and voting is very important’. The purpose of the reminder to vote treatment was to evaluate the generalizability of American GOTV experiments, which have typically found personal delivery of a pro-turnout message raises turnout (Adams and Smith, Reference Adams and Smith1980; Eldersveld, Reference Eldersveld1956; Gerber and Green, Reference Gerber and Green2000, Reference Gerber and Green2001; Green, Gerber, and Nickerson, Reference Green, Gerber and Nickerson2003; Michelson, Reference Michelson2003; Miller, Bositis, and Baer, Reference Miller, Bositis and Baer1981; Ramirez, Reference Ramirez2005). As several studies have demonstrated similar effects in a variety of different countries and settings (Giné and Mansuri, Reference Giné and Mansuri2011; Aker, Collier, and Vicente, Reference Aker, Collier and Vicente2011; Bochel and Denver, Reference Bochel and Denver1972; Fieldhouse et al., Reference Fieldhouse, Cutts, Widdop and John2013; John and Brannan, Reference John and Brannan2008; Guan and Green, Reference Guan and Green2006), we hypothesized that the reminder treatment would increase turnout.
In the ‘ink prime’ treatment, interviewers delivered the mobile phone survey and then the following script:
I would like to speak with you briefly about the upcoming election. Are you aware that there will be a presidential and legislative election on February 18? Are you aware that visible ink will be used to mark the fingers of those who have voted in this election? [Response noted.] Are you aware that this ink will be easy to see and will last for several days? [Response noted.] Ink is not just about preventing fraud. It will also allow everyone to know if you’ve voted or not. It is important to show your neighbors that you are responsible and that you are involved in influencing what happens in your community. All registered voters will have the opportunity to vote in this election and voting is very important. Since you are a registered voter, please remember to turn out and vote.
This treatment represented a milder version of the pro-turnout letter in Gerber, Green, and Larimer (Reference Gerber, Green and Larimer2008), which threatened to publicize non-voters’ behavior to neighbors. Several studies have implemented treatments similar to Gerber, Green, and Larimer’s in a variety of American settings (Davenport et al., Reference Davenport, Gerber, Green, Larimer, Mann and Panagopoulos2010; Gerber, Green, and Larimer, Reference Gerber, Green and Larimer2008; Mann, Reference Mann2010; Nickerson, Reference Nickerson2008; Panagopolous, Reference Panagopolous2010), but, to our knowledge, none have done so in an emerging or partial democracy.
The ink treatment manipulates knowledge or awareness about the link between ink and the visibility of turnout rather than the practice of inking itself. Informational treatments like the ink prime work one of two ways: they change the factual knowledge base of subjects (a learning effect) or alter the salience of particular considerations (a priming effect) (Conroy-Krutz, Moehler, and Aguilar, Reference Conroy-Krutz, Moehler and Aguilar2016). In focus groups prior to the experiment, we found many knew ink would be used in the upcoming election, but not all understood it would make turnout decisions visible for an extended period of time and many had not thought through the political implications of this revelation of information.Footnote 12 We therefore had reason to believe the ink treatment would induce either learning or priming in a subset of respondents.Footnote 13
Sampling and Randomization
We visited 1,137 households in the week prior to the election.Footnote 14 We selected ten Kawempe South parishes and designated a supervisor and team of enumerators to each.Footnote 15 Each enumerator received a packet of surveys, which came in three versions, corresponding to the treatment arms in Table 1. We randomized survey order for each enumerator. By working through the packets in order, enumerators achieved treatment randomization. Our procedure ensured randomization across treatment arms, but not a representative sample of Kawempe. Our sample is biased toward people at home during the period in which we delivered the survey and those willing to participate.Footnote 16
During the three days following the election, enumerators recontacted treated individuals.Footnote 17 They repeated demographic questions and asked a new set of political and social questions, including vote choice, political party identification, strength of ethnic attachment and political and social pressures to vote. Near the end of the survey, enumerators asked to check the respondent’s finger for ink and recorded whether the respondent had a mark (indicating they voted) or lacked a mark (indicating they did not). We achieved a recontact rate of over 99 per cent; fewer than 2 per cent of successfully recontacted participants refused the follow-up survey, producing an overall attrition rate of 2.1 per cent.Footnote 18
Results
We first examine whether the ink prime had the intended effect of increasing awareness about the public nature of voting. We asked: ‘Regardless of whether you actually voted, do you think that your neighbors know whether you turned out or stayed home, or do they not know?’ Overall, most respondents in our sample (63 per cent) believed that their neighbors could ascertain whether or not they voted, suggesting that voting in Kawempe is a relatively public act.Footnote 19 If the visibility treatment had the intended effect, respondents exposed to it should have been more likely to believe their neighbors knew their turnout choice. Indeed, the ink prime increased the probability of believing neighbors knew turnout decision by about 7 per cent.Footnote 20 The simple reminder treatment served as a placebo for this outcome and, as expected, had no effect.
Next we present a basic breakdown of the outcome variable: finger marking status (Table 2). A last-minute decision by election administrators in Uganda to use permanent markers instead of indelible ink complicated our measurement strategy. The permanent marker made a more subtle mark on the skin, which meant reading fingers required cooperation on the part of participants. A small number (4 per cent) of respondents refused to allow their fingers to be examined. More respondents (20 per cent) made it difficult for enumerators to get a clear reading. Rather than treating these unclear or refused fingers as a nuisance category, we view them as a strategic choice that reveals important information about the mindset of Kawempe citizens. In the following analysis, we group unclear fingers and refusals together as ‘concealers’.Footnote 21 We then use multinomial logit to examine concealers, marked and unmarked fingers as an unordered three-level categorical variable.Footnote 22
Finger status
The large ‘unclear’ category means we cannot map our marked finger variable directly to turnout because the true turnout level depends on the fraction of unclear fingers/refusals that actually voted. We can, however, put bounds on the turnout level: 65 per cent of our sample had a clearly marked finger and only 11 per cent had a finger that was clearly not marked. This gives us a turnout range of 65–89 per cent in our sample.Footnote 23
Table 3 shows average treatment effects on finger status for the two treatment variables (and 95 per cent confidence intervals).Footnote 24 Treatment effects represent the changes in probability for each category of the finger status variable given a particular treatment. Changes across a row must sum to 0. The simple reminder treatment produced no discernible shifts between categories of the dependent variable. The ink prime treatment, on the other hand, increased the probability of not having an inked finger by around four percentage points. This effect falls just shy of the 95 per cent confidence level, with a magnitude similar to those in American GOTV experiments, most of which fall in the 1–8 per cent range.Footnote 25 As our sample undoubtedly included many people who already knew about inking and had factored it into their decisions about whether to vote, our results likely understate the full impact of inking on behavior.
Average treatment effects (95% confidence intervals)
Notes: We estimate a multinomial logit on the three level finger status variable, controlling for the reminder and ink prime treatments (control is reference category) as well as parish fixed effects to account for the blocking design, generating first differences for each treatment. Raw output of the model is presented in Table A5 of the Appendix, ‘baseline model’.
The ink prime treatment should have had its largest effects on respondents not already aware of the use and implications of ink in elections. We do not have direct pre-treatment measures of information about inking (asking these would have confounded the experiment). We can, however, examine differential treatment effects for respondents at varying levels of education and age. Less educated residents likely had limited access to information about the electoral process, including the use of ink. Younger voters, not having voted in previous elections, also likely had less knowledge about the use of ink and its implications.Footnote 26 We therefore expect the ink prime to have had stronger effects for these groups versus older, better educated participants.
Indeed, controlling for education, age and interaction terms for these variables and the ink prime substantially sharpens the effects of the ink treatment (now significant at the 0.01 level): both interactions are significant at the 95 per cent level.Footnote 27Table 4 shows the effects of the ink prime treatment for respondents without a formal education versus those with a university degree or higher, and young (18–24) versus older respondents (60+).Footnote 28 The effects of the ink treatment on those with a university degree were indistinguishable from 0 while treatment effects for people without a formal education were substantial. Similarly, older people had no identifiable response to the inking treatment – if anything it may have increased their probability of having a marked finger (although this result is not statistically significant). Young respondents, in contrast, proved very sensitive to the inking treatment, displaying a significant increase in the probability of having an unmarked finger and a large drop in the probability of having a marked one. Given the almost symmetric and compensatory responses in the probability of being inked or not, and changes in the probability of being concealed centered around 0, we believe the treatment not only reduced the probability of being inked, but actually reduced turnout itself in young and less educated voters.
Heterogeneous ink prime treatment effects for education, age and gender
Note: Simulated from a multinomial logit model with parish fixed effects, N=1,124. 95% confidence intervals in parentheses. See Table A5 for raw coefficients and full details.
We might also expect effects to be strongest in more vulnerable groups, in particular, female voters. Work by Grossman, Humphreys, and Sacremone-Lutz (Reference Grossman, Humphreys and Sacremone-Lutz2014) suggests that female voters in Uganda are more marginalized and less engaged than male voters. They may therefore be especially vulnerable to turnout discouragement strategies. We find suggestive evidence of a gender effect: the ink treatment had a significant depressive effect on female voters but no effect on male voters. The z-statistic for the interaction term of female and the ink prime was not significant, however.
We also looked for evidence of turnout discouragement for the subset that, in this setting, we would expect to be most evident: opposition supporters. Partisan effects are challenging to detect in Kawempe. Given Uganda’s single party dominant state and history of opposition oppression, there are compelling reasons for individuals to misrepresent or conceal partisanship. When we asked respondents in our post-treatment survey to report vote choice in the presidential election, 22 per cent refused to answer the question, a telling indication of their reluctance to reveal partisanship.Footnote 29 Most of those who did answer said they supported Museveni. We also asked them if they were close to any particular party: 9 per cent refused to even answer this question, and 40 per cent claimed not to be. Again, most of those who self-identified as partisans indicated that they supported the NRM. As the NRM won neither the presidential nor parliamentary election in Kawempe South, we conclude that many opposition supporters concealed their true preferences by refusing to answer or misrepresenting their choice. It is impossible to use such data as a true measure of electoral support.
Our clearest indication of partisan leanings is whether or not our enumerators noted pro-NRM displays on the living quarters of respondents. These displays could show electoral support for Museveni, but do not conclusively reveal partisanship since strategic incentives to misrepresent apply here as well. We find no detectable interaction between NRM displays and the ink prime.Footnote 30
We also considered whether visible demographic cues might provide information about partisanship that would help us identify opposition supporters in the data. Here again we ran into challenges. While ethnicity or race reveals information about partisanship in some contexts, it is not particularly informative in Kawempe. Visible cues do not provide clear information about individual ethnicity in Kawempe (Habyarimana et al., Reference Habyarimana, Humphreys, Posner and Weinstein2007), and ethnicity does not correlate strongly with stated partisanship in our sample (see Table A8). The only demographic indicator we identified that was both visible and correlated with partisanship was age. Age is relatively easy to see and correlates with reported vote choice in Kawempe, with younger age groups less likely to say they voted for Museveni (see Tables A8 and A6). The salience of youth as an opposition marker might have been especially high in February 2011, as this coincided with youth-led upheavals in Egypt and elsewhere. As already discussed, the ink prime had a particularly strong effect on young voters. While we have interpreted this as a likely product of this group being less experienced and informed, it is also consistent with a partisan effect. Younger voters, fearing they would be associated with the opposition, may have been more sensitive to the ink prime.
Turnout Discouragement?
We have shown that an experimental ink prime treatment likely reduced turnout in Kawempe during the 2011 election, especially among younger and less educated potential voters. These results are consistent with the employment of turnout discouragement strategies in Kawempe, but obviously cannot tell us who might have been discouraging turnout, why they were doing it, or provide direct evidence of such efforts. Are there additional indications that strategies to reduce turnout may have in fact been in use? Here we consider two possibilities: community-based social sanctioning to enforce norms against participation and political efforts to suppress turnout by the Museveni regime.
We find no evidence of community-based social sanctioning of turnout in Kawempe. If anything, Kawempe norms seem to favor participation. In our post-treatment survey, we asked several questions exploring social mechanisms to encourage or discourage turnout. The responses to these questions suggest that Kawempe residents believed their neighbors knew whether or not they voted but were relatively unconcerned about social sanctions for this behavior. To the extent that they feared negative reactions, it was for not voting. Thus, 63 per cent said their neighbors knew whether or not they voted but when asked whether their neighbors would treat them positively, negatively or no different if they learned the respondent did not vote, 65 per cent responded that it would not make a difference. For the subset that believed it would make a difference, most believed their neighbors would respond negatively to not voting. These pressures to vote seem particularly strong for older respondents and those who had lived in Kawempe for a long time.Footnote 31
Kawempe residents also report pro-turnout messages from their religious leaders. Although Kawempe has many relatively new residents (more than half of our sample had lived in the community for five years or less), most residents, even newer ones, participate in community life through religious organizations. Over 80 per cent of our sample reported attending church or mosque at least once per week, and many attended multiple times per week. These religious organizations encourage political participation: 58 per cent of our sample reported that religious leaders at their service encouraged them to vote frequently or sometimes; only 16 per cent indicated never experiencing pro-turnout messages from religious leaders. In sum, we see little indication that social norms or community pressure dampened participation. If anything, social pressure operated in the opposite direction, particularly for older, well established Kawempe community members.
If social pressure does not explain our results, what does? We believe the answer lies in efforts by the ruling NRM to discourage turnout, either by actively threatening likely opposition supporters, or more likely through subtle attempts at intimidation that changed the voting calculus of some voters just enough to keep them at home. The ruling NRM and its leader Museveni certainly had incentives to discourage turnout in Kawempe. Given its historical support for the opposition, lower turnout there would bolster the NRM’s electoral dominance and starve the opposition of one of its few reliable sources of votes.Footnote 32
We have no evidence of intentional, centrally directed, turnout punishment in Kampala. Intimidation can be difficult to detect, however. Intimidators have ways to signal their intent to their desired audience without revealing it to others. Moreover, it may only take a modest or even unintentional threat to affect turnout levels. Since voting confers low individual benefits, even small shifts in perceptions about the probability of sanctions may be enough to shift behavior(Driscoll and Hidalgo, Reference Driscoll and Hidalgo2014). Hence, the absence of overt acts of intimidation is not evidence that some voters did not feel intimidated, and that this intimidation kept them at home.
In fact, ample evidence suggests that many voters felt intimidated during the 2011 Ugandan election. In 2008, more than two years prior to the election, 45 per cent of Kampala’s residents polled by the Afrobarometer answered that they personally feared becoming a victim of election related political intimidation or violence.Footnote 33 Numbers were similarly high in late 2010 and early 2011. Fears were especially prominent amongst opposition supporters (Conroy-Krutz and Logan, Reference Conroy-Krutz and Logan2012).
Ugandans had good reason to be nervous about election-related violence. The previous election in 2006 saw several instances of such violence. Museveni suppressed demonstrations in Kampala in September 2009 by sending in the police, who shot and killed twenty protestors. Reports of human rights abuses by the army and police increased in the period between the elections (Barkan, Reference Barkan2011). The NRM primaries in summer 2010 featured violence (Conroy-Krutz and Logan, Reference Conroy-Krutz and Logan2012). In interviews we conducted after the election, Kawempe residents complained of intimidation, bribery, rigging and vote stuffing by security forces and NRM representatives.
In this context, Museveni may have been able to discourage turnout in opposition areas without engaging in overt and conspicuous efforts to do so. Even if intimidation did not change hearts, minds and votes (Conroy-Krutz and Logan, Reference Conroy-Krutz and Logan2012), it may have kept opposition voters at home.Footnote 34
If Museveni did attempt to discourage opposition voters from going to the polls, we would expect to see a negative correlation between opposition support and turnout in the 2011 election. We also would expect opposition support to correlate with use of violence and intimidation. Using data collected in a pre-election survey in January and February 2011, we find the level of opposition support positively correlated with expectations of violence, with borderline statistical significance. Moreover, areas with more opposition support had significantly lower turnout levels.Footnote 35 None of these pieces of evidence offers a ‘smoking gun’ for politically motivated turnout discouragement, but they paint a coherent picture of an electorate on edge, fearful of the consequences of voting. For anxious voters in Kawempe, any additional information that their turnout decision would be visible may have been enough to keep them at home.
Conclusion
During the 2011 Ugandan general election in an opposition-leaning neighborhood of a coercive single-party dominant state, raising awareness about the use of election ink reduced inked fingers, especially among young and less educated voters. This result is consistent with turnout suppression, supporting our intuition that electoral inking may have undesirable second-order behavioral consequences in some contexts. Kawempe is in many ways unusual in Uganda and our study was small-scale and exploratory. We make no claims these results generalize to all parts of the country. Indeed, our expectation is the effects of inking on turnout are highly context specific. In Museveni strongholds, we would expect inking to have the opposite effect, that is, encouraging turnout. However, we believe similar dynamics and locales such as Kawempe might be found within many other countries.
We hope our findings elicit greater consideration of the largely unquestioned practice of inking in emerging democracies. We do not advocate the whole-scale abandonment of inking in these elections. Ink is a low cost and easy to implement solution to important problems of electoral administration. In many elections, its behavioral effects are likely to be mild and benign. During elections featuring turnout manipulation strategies, however, electoral administrators should consider whether inking facilitates coercion and, if so, whether the risks of using ink outweigh the benefits.
Our results also suggest the outcomes reported in American GOTV studies should be reconsidered. The findings of these studies likely depend on a context in which community norms favor participation and electoral strategies promote turnout. In other contexts, like Kawempe, public revelation of turnout may have different, less benign results.
Finally, our results highlight the importance of broadening the study of political behavior in developing democracies to explicitly consider electoral context. Prior work on turnout in Africa has focused primarily on individual-level factors like education, income and partisanship, often borrowing independent variables from studies of US elections (Bratton, Reference Bratton1999, Reference Bratton2008; Bratton, Mattes, and Gyimah-Boadi, Reference Bratton, Mattes and Gyimah-Boadi2005; Isaksson, Reference Isaksson2014; Kuenzi and Lambright, Reference Kuenzi and Lambright2011, Reference Kuenzi and Lambright2007). To truly understand voters in developing democracies, we must look beyond individuals to how electoral practices, institutions, community norms and political strategies interact with individual factors to shape behavior. Our study suggests it is especially important to understand practices and institutions that reveal information about otherwise private political action. We have focused on election ink, but a variety of practices and institutions can reveal such information, including voter files recording information about individual turnout, geographic segregation of politically similar individuals, and violations of ballot secrecy. When politicians can accurately deduce the partisan leanings of voters, they gain access to a range of powerful strategies to shape voter behavior: intimidation, vote buying and turnout manipulation (Kitschelt and Wilkinson, Reference Kitschelt and Wilkinson2007; Shaffer, Reference Shaffer2007; Stokes, Reference Stokes2005; Nichter, Reference Nichter2008; Ferree and Long, Reference Ferree and Long2016). We hope our work inspires further theoretical and empirical analysis of the interaction between such practices and institutions and political behavior.
Supplementary Material
Replication data and instructions can be found in Harvard Davaverse at https://dx.doi.org/10.7910/DVN/LZMEIV and online appendices at https://doi.org/10.1017/S0007123418000121.
Acknowledgements
We gratefully acknowledge the help of Barak Hoffman and Craig McIntosh, who contributed in essential ways to design and/or implementation stages of this project. We also thank David Nickerson, James Fowler and the participants of seminars at University of Notre Dame and Stanford University for their comments. Doreen Lwanga provided expert advice in the field. Daniel Arnon, Inbok Rhee and Danielle Villa provided excellent research assistance. We greatly appreciate James Long, who generously contributed to the design and implementation of this project. This research was funded by the Kellogg Institute for International Studies at the University of Notre Dame.
