From Blank Stares Come Blank Ballots:
Poor Candidate Information and the Prevalence of Null Votes in the World’s First Democratic National Judicial Elections (Bolivia 2011)
Rough draft only; please do not cite
By Todd A. Eisenstadt and Jennifer Yelle, American University
Abstract: Using the unique case of Bolivia’s 2011 Constitutional Court elections, where vote was mandatory and there was no legal campaigning or campaign spending, and nearly 2/3 of the electorate cast blank or null votes, we weigh in on one side of the US debate over state-level judicial elections, arguing that, at the national-level too, judicial elections are a bad idea. Voters do not seem to differentiate between candidates based on qualifications or experience, and instead draw distinctions based on perceptions of partisanship and ethnic identity. Using the null and blank votes cast as the dependent variable, we show that blank votes were cast more frequently in municipalities where there was NOT a substantial amount of support for the ruling MAS party. We also found qualifications had no impact on the casting of votes. We consider implications for the election of judges, concluding that, even controlling for structural obstacles to candidate differentiation in the view of the electorate (e.g., illiteracy), Bolivian voters had difficulty differentiating candidate quality. This confirms the claim made continuously by one side in the US case, arguing that judicial elections may not yield the most qualified judges.
The first democratic national judicial elections in modern history[1] were held in October 2011 in Bolivia, widely repudiating the argument sometimes made in political science literature on the U.S. states, where two thirds elect some state-level judges, that electing judges would make the bench more accountable and representative. Voters cast more blank and null votes in the race than in any other democratic election in modern history (IDEA database). Indeed, well over half of the votes cast in that remote Andean nation of 11 million people were disqualified because they were blank or otherwise “null,” and a majority of voters in an exit poll explained that they did not have sufficient information upon which to base a reasoned choice. Moreover, nearly all of the two dozen experts interviewed unanimously qualified the election results as disappointing, as the winning candidates possessed very little judicial experience, and many are not even lawyers. Indeed, the Plurinational Constitutional Tribunal justice who garnered the highest number of votes made national news, to the disdain of nearly all the experts interviewed, across the political spectrum,[2] by stating nonchalantly that difficult cases are better decided by consulting coca leaves than laws.[3]
As pointed out by Driscoll and Nelson (2012), the Bolivian judicial elections offered an excellent opportunity to explore the effects of candidate quality and voters’ choices on judicial elections, because individual campaigning was forbidden, media coverage censored, the election was a “first” (meaning that there were no incumbents to possess an advantage), and candidate pre-selection ensured gender parity and that some 35 percent of the candidates self-identified as indigenous. But these authors, using votes cast by municipality (N=328) as their unit of analysis, found that voters did not necessarily elect the most qualified judges, but perhaps used ascriptive criteria to elect the candidates “most like them” in ethnic and gender terms, and also considered candidates’ partisan affiliations, even though the elections were non-partisan, and candidates were allegedly independent of party identification.
More important than the other experimental conditions present for the election cited by Driscoll and Nelson, may have been Bolivia’s compulsory voting law, which meant that every registered citizen had to cast a ballot, including the majority, who either opposed the candidates or the process, or simply were too uniformed to cast any ballot at all. And if elections are supposed to be positive statements about competition between qualified candidates in a free and fair contest to decide the vital issue of who governs, the Bolivian judicial contest was decidedly a failure. Some 58 percent of the ballots were disqualified in the Plurinational Constitutional Tribunal race, or three times as many as those disqualified in recent municipal elections, where voting is also compulsory.
Like Driscoll and Nelson, this article tests hypotheses about the election versus appointment of judges. However, unlike those authors, who evaluated the electoral results consistent with the U.S. judicial election literature, we ask a larger question. Rather than using candidate professional qualifications, age, past government affiliation, and demographic variables to assess the dependent variables of which candidates actually won, we instead call into question the entire process. Instead of evaluating the determinants of who won, we elaborate various means of showing that the elections undermined citizen confidence in the judicial branch and then, given this overwhelming citizen and elite perception of electoral failure, we look at determinants of null and blank votes. We consider the Plurinational Constitutional Tribunal, created in 2009 to interpret Bolivia’s large corpus of laws relating to indigenous and other constitutional rights, as we feel this election (among the four judicial elections held simultaneously), presents the best “hard case.” That is, if voters were not interested in the magistrates who would adjudicate this extensive, concrete and newly created set of laws, then they would probably also not be interested in the Judicial Council (which hears administrative and punitive cases against other magistrates) and the Agricultural Court, which adjudicates land disputes.[4]
The US-centered literature focused on whether voters can differentiate candidates sufficiently on qualifications and platforms. However, in developing democracies, partisanship, ruling party clientelism and structural variables (poverty and literacy) may be important.
This article seeks to test both of these literatures, which emerge from diverse contexts in order to explain the anomalous results. Bolivia achieved the highest rate of blank or null votes recorded on any of the 468 compulsory voting elections in 28 nations found in International Institute for Democracy and Electoral Assistance data base ( , no others achieved 58 percent “spoilt/non-effective” outcome. Indeed, the 58 percent null or blank ballots out of 4.7 million votes cast in Bolivia’s October 2011 election was a higher percentage of blank or null votes than in any other modern democratic election on record (IDEA data). Even within Bolivia, the election was highly anomalous. Turnout in 2011 judicial election was 60 percent (despite compulsory voting law), and 58 percent of the votes cast were invalid. In the six other national elections (legislative and presidential contests in 2009, 2005, and 2002), the average turnout was 84 percent and only 5 percent of the votes were ruled as invalid. What explains this dramatic change?
Our central claim is that the difference between judicial candidate qualifications is not very great, and, given the partisan nature of the establishment of the judicial elections and also the conduct of these elections (despite the allegations of neutrality), these elections were not destined to allow voters to select the best qualified and most experienced jurists. Indeed, several legal experts interviewed cited Gualberto Cusi Mamani, the highest vote-getter among Plurinational Constitutional Tribunal victors, who last March told the national media in Bolivia that when he has a difficult case, such as one with constitutional implications, he seeks to “read” coca leaves. Populations were not able to differentiate between candidates because they did not have sufficient information about 1) the candidates themselves and 2) what makes a good judge. Furthermore, there was absolutely no information about platforms.
Much of Bolivia did not have access to campaign materials. A ban on campaigning meant that all voters were privy to the same state-provided campaign literature. Yet these written brochures aid the literate voters but leave the illiterate voters at a loss of verbal campaign events. Indeed, we find that illiterate voters were more likely to have no information about the election and file ineligible (blank or null) votes.
Even without this lack of information about the candidates, populations often have trouble discerning which candidate would make a better judge. The American political science literature argues that people are often unable to rank qualifications sufficiently for this vocation because the large majority of the population does not understand the role of judge sufficiently. We find evidence for this as well in Bolivia, mending the gap between the narrowly focused U.S. judicial elections literature and other countries experiments with the democratization of the judiciary. This lack of information causes voters to turn to other possible cues for discerning which candidate to choose—if any. While the bans on campaigning were posited to hinder a possible clientelistic arrangement with poor and rural voters (the vast majority throughout Bolivia), we still find evidence that these structural conditions (poverty, indigeneity, ties to President Morales’ Movement Toward Socialism, or MAS party, and rural locales) vulnerable to clientelistic influencecorrelated with increased blank and null balloting in the judicial elections.
We test these claims using municipal-level results from the 2011 judicial election as well as looking at data from Bolivia’s most recent completed census (still the one from 2001), and from a survey conducted in major cities after the election. Our premise comes from preliminary analysis of the survey by the international polling firm IPSOS. When asked[5] how informed the respondent felt about the candidate (none/little/enough/very informed), the majority of respondents felt little informed. When broken down into education groups, the majority of respondents answered “little informed” most frequently within their educational group except for the least educated who felt “not informed.” Thus most illiterate voters responded that they felt less informed than the average voter. Indeed, for all groups, when asked why they think people might not vote in the judicial election, the overwhelming majority (1,394/2,080 = 67%) said lack of information about the candidates. When broken down by education level this is the most popular response for each group.
The second most popular reason given (253/2,080 = 12%) to why the respondents think citizens might not vote in the judicial election, is ‘because of a suspicion that the candidates were affiliated with President Evo Morales’ Movement to Socialism Party (MAS in Spanish). These responses tended to be given by more educated respondents ( 39/197, 19.8%, of all college educated said this and 8/24, 33.3%, of all post graduates said this). In a similar vein, 18.2% of all non-educated people (4/22) said that people would not vote because they want to approve of the opposition. In the following section, we discuss the literature on voting in judicial election as considered in the U.S. debate, but also the clientelism literature, and then develop null hypotheses based on conventional explanations, as well as a couple based on the clientelism literature.
Literature Review: Broadening the US-Centered Judicial Elections Debate
The Bolivian election touches upon two overlapping debates in the judicial elections literature. The most prominent question in the literature is whether voters are qualified to choose between candidates, meaning, do they understand the qualifications which make a successful judge (e.g., previous adjudication record, etc) or are they forced to rely on other, often problematic, cues (e.g., party affiliation, sex, religious-ethnic cues, name familiarity). This debate overlaps with the larger, more normative, debate of whether judicial systems should emphasize accountability, via elections, or independence, via judicial appointments. Using this US-based literature on judicial elections, we are able to theoretically situate the Bolivian experiment and show that components of the US literature have applicability outside of the fifty states, but we are also able to expand upon this literature by introducing new variables found in Bolivia, such as clientelism. We first explore the theoretical underpinning of the US judicial elections literature and then contextualize the case of Bolivia with regard to this literature. We then delve deeper into the case of Bolivia and discuss the importance of clientelistic relations in affecting the judicial elections. We test these hypotheses and conclude that the narrow judicial elections literature is limited in explaining voter behavior in Bolivia and that a more historical, institutional perspective is needed to understand the (lack of) voting in Bolivia. The results of this analysis have large consequences for the more normative debate over whether appointing judges or electing judges is most in the public interest. We show that voting for judges does not introduce accountability (as is argued in the literature) when institutions have a history of clientelistic type behaviors.
To begin, we turn to the literature on voting behavior in judicial elections. In this line of research, the question is often whether voters have enough information about the candidates to make a decision between candidate qualifications or whether they are forced to rely on other, often problematic, cues such as gender or religious-ethnic cues. Those scholars who are skeptical of the virtues of judicial elections often state that the voting population does not have the requisite knowledge to judge the candidates in these elections, thus they rely on other problematic cues for their decision and the result is a system which produces biased/inferior selection of candidates. On the other hand, proponents of judicial elections state that many of these other indicators (religion, gender, etc) are not necessarily problematic and are used throughout all democratic election processes in legislative and executive elections as well.
In most regards judicial elections resemble legislative elections. Scholars have found the whole gambit of voting cues to be a significant factor affecting voting behavior: party cues matter for voters’ decisions (Dubois 1979a, 1980a:70-79, Barber 1971, Adamany and Dubois 1976:756-60 found in Dubois 1984:398; Gann Hall 2007:1156); incumbency advantage matters (Dubois 1979a:771-75, 1980a:79-82 found in Dubois 1984:398 Griffin and Horan 1981:22-23 found in Dubois 1984:403); name familiarity matters (Dubois 1979a:771-75, 1980a:79-82 found in Dubois 1984:398); religious-ethnic cues matters (Nagel, 1973:20-21, 23 found in Dubois 1984:398); gender matters (Dubois 1984:419; Dubois 1980a:81, found in Dubois 1984:398); relative position on ballot (Dubois 1980a:81, found in Dubois 1984:398); Newspaper endorsements matter (Dubois 1984:419); ballot roll-off (voting during other state-wide elections) matter (Gann Hall 2007:1156); voter pamphlets matter (Beechen 1974:244 found in Dubois 1984:410); and occupational history matters (e.g., lawyer, municipal judge, etc written on ballot) (Dubois 1980a:81, found in Dubois 1984:398). Thus the literature suggests that voting behavior in judicial elections is very similar to voting behavior in legislative elections but with two very important differences: campaign spending and voter awareness. Unlike other elections where campaign spending is posited to be a highly potent factor affecting voter choice, voter decisions in judicial elections are not seen to be greatly affected by campaign spending. When it does matter, Dubois (1984:430) found that it only matters for small towns and Bonneau (2007: 496) found that it only matters for challengers, not incumbents.
In terms of voter awareness, scholars have found that a lack of awareness of courts, judicial contests, and the identity of specific candidates is a salient factor affecting a voter’s choice (or non-choice) in a judicial candidate (Klots 1973; Jacob 1966; Ladinsky and Silver 1967; Adamany and Dubois 1976; Johson et al 1978; Roper 1981 found in Dubois 1984:403). However, not all scholars agree (see: Lovrish and Sheldon 1983: 245-46; Philip et al; 1976; found in Dubois 1984; Gann Hall and Bonneau 2006:29-30; Bonneau and Gann Hall 2003:342-343). This contention boils down to the question of whether voters understand the judicial process enough to be able to choose which candidate is more qualified. Scholars who reject the virtues of judicial elections believe that the lack of information about the judicial elections significantly impedes the ability of the election to turn into a mechanism for accountability; whereas, proponents of judicial elections believe that the amount of discretion—even if small—used by judges mandates that they be held accountable to the people, for this is the foundation of democratic theory.
A recent debate by Carrington (2011), a proponent of judicial appointment, and Gaylord (2012), a proponent of judicial election highlight all this point of contention. In making his case for judicial elections, Gaylord (2012) uses the statistical analysis of Bonneau and Hall (2009). Bonneau and Hall analyzed US state Supreme Court elections and found that voters are indeed able to distinguish between different qualifications of judges. As explained by Bonneau and Hall (2009:13 cited in Gaylord:73-74), "[w]hen casting votes among supreme court candidates, voters distinguish challengers who have experience on the bench from challengers who lack it and thus are less suitable alternatives to incumbents." However, this statement is hardly an accepted truth. Carrington (2011:1984) provides many examples of flawed judicial elections in the United States due to a lack of voter awareness. He notes how a reoccurring result has been "the occasional success of grievously unqualified candidates who happened to share a name with a popular figure." He give the example of Donald Burt Yarborough, "who was elected in 1976 at the age of thirty-five to the Supreme Court of Texas because voters thought that he was the same Don Yarborough for whom they had voted in races for other state offices (who was, in turn, mistaken for Ralph Yarborough, a United States Senator." (2011:1984). In the case of Bolivia, Driscroll and Nelson (2012) did not find any evidence that voters were able to distinguish between candidates qualifications. They find that voters acted on other cues, mainly ethnicity and party affiliation. We have also noted through surveys conducted by IPSOS this lack of candidate awareness[6]. However, unlike Driscroll and Nelson, we find that that a more important facet of voting behavior was the lack of voting in an election where voting was compulsory. While lack of awareness of candidates may have shaped voting behavior for those who chose a candidate, essentially turning votes towards partisan affiliates and co-ethnics; we find that the bigger effect on voting behavior was the intense allegations of MAS corruption in the judicial elections design, essentially spurring wide-spread non-participation. We argue that this rejection of the judicial election--through null votes--is not so much about which voter cues were emphasized and to what extent, but rather, is a deeper matter of institutional legitimacy.