Introduction
On 30 March 2013, the Supreme Court of Kenya unanimously ruled that the recently concluded presidential electoral process had fulfilled constitutional standards of credibility. Given the copious amount of evidence alleging serious errors and inconsistencies related to the voters’ register and counting and tallying procedures, the judgment was somewhat surprising. There exists significant analysis of what the judges’ rationale may have been, but there is less scholarship regarding lay public opinion about the judgment.
In fact, despite the development of ‘global norms’ regarding electoral integrity, little is known about how voters in Kenya and around the world interact with and use these standards in assessments of their own elections. Indeed, there is relatively little information regarding what voters on the ground consider to be important in this regard.
In this paper, I use results from a post-election national public opinion poll in Kenya to investigate whether Kenyans felt that the landmark 2013 general election was free and fair, and the reasons for their opinions. I test the extent to which experiences at the polling station, levels of confidence in electoral institutions, the prioritisation of ‘peace’, and ethnic identities influence voters’ assessments of electoral credibility in 2013.
My findings indicate that Kenyans’ conceptions of electoral integrity are grounded in the country's history of electoral politics and that a free and fair election is much more about what the results mean for the everyday reality of certain ethnic groups than it is about the technical aspects of election administration. I show that ethnic identity influences Kenyan voters most strongly, followed by faith in electoral institutions and the importance of ‘peace’. Voters’ personal experiences at the polling station play an insignificant role in how they ultimately judge the fairness of the electoral process.
Electoral integrity: local context versus ‘international consensus’
There is virtual consensus among politicians, practitioners and academics that ‘free and fair' elections are critical to emergent and established democracies, but for many years there was relatively little information regarding what ‘free and fair' meant in terms of measurable standards.
Anyone familiar with the often-messy business of monitoring elections knows that vote fraud can be a very complicated, shadowy, and slippery affair that causes domestic and international observers to pull their hair out by the handful. Much practical knowledge and painstaking methodological analyses have gone toward devising methods that allow monitors to distinguish massive fraud from widespread but unsystematic irregularities, but the results so far are hardly conclusive. (Schedler 2002, 38–39)
Part of the problem was – and continues to be – the lack of specific indicators. Many international election monitoring organisations1 have developed assessment guidelines, but these are often difficult to operationalise into measurable standards (Norris 2013a, 565). Moreover, it is often difficult for election monitors to escape the restrictions that come with political relationships (Ibid., 564). Overall, there has been a serious lack of systematic methodology for assessing the quality of elections (Elklit and Reynolds 2005, 147; Elklit and Svensson 1997, 33; Mozaffar 2002, 86; Norris 2013a, 564).
Over the past 15 years, however, academics have taken up the challenge to produce such standards (Alvarez, Hall, and Llewellyn 2008; Elklit and Reynolds 2002, 2005; Elklit and Svensson 1997, 42; Norris, Frank, and Martinez i Coma 2013, 132–134; Schedler 2002, 40–41; Vickery and Shein 2012, 3–11).
In 2013, Norris articulated her own definition of electoral integrity, which brought together the diverse work on this topic. Norris (2013a, 564) suggested:
Electoral integrity refers to international conventions and global norms, applying universally to all countries worldwide throughout the electoral cycle, including during the pre-electoral period, the campaign, on polling day, and its aftermath. Conversely, the notion of ‘electoral malpractice’ is used to refer to first and second order violations of these global norms. This conceptualization emphasizes four distinct features:
Global norms grounded in multilateral agreements, international conventions, treaties, and international laws … ;
A distinction between first and second-order malpractices, according to the severity of their potential consequences;
Shared standards applying universally to all countries and types of regimes … ; and
The notion of an electoral cycle, where a range of electoral malpractices can occur at any stage in a sequential process …
Notably, Norris's focus on global standards undermined the importance of case studies. Her warning that case studies are not useful for comparative analysis may be true, but Norris herself acknowledged that there are instances in which international assessments based on global norms differ significantly from public assessments (Norris 2013b, 584). It is important to understand such differences, both in order to refine international norms and to learn how to address specific local problems.
Perhaps more importantly, local opinion is important because it is voters who ultimately legitimise elected governments. It is therefore critical to understand what voters prioritise in their own assessments of elections. If an election is not credible in their eyes, international norms, no matter how widely endorsed they may be, do not make much of a difference. Indeed,
the concept of ‘fairness’ is a qualitative judgment that voters themselves are in an ideal position to make … It is voters who experience the election campaign, directly through contact with election candidates and parties, and indirectly through the mass media. Unlike outside observers, they will also cast a ballot and therefore have a degree of ‘ownership’ over the election outcome. (McAllister and White 2011, 665)
Birch agrees, noting that citizens are especially well placed to evaluate electoral processes, given widespread participation and the attentiveness that elections tend to spark (Birch 2008, 306–307). Ignoring public opinion is dangerous, as low levels of domestic public faith in elections have been shown to be linked to depressed levels of public faith in other political institutions, as well as to civil unrest (Birch 2010, 1602). Understanding the reasoning behind voter confidence is also important for voter turnout rates, with evidence that voters who believe an election will be free and fair are more likely to vote (Ibid.).
There are relatively few studies of how voters decide whether or not elections are credible, but the limited extant literature can be divided into three broad categories: personal voting experiences; faith in election-related institutions; and partisanship (Kerr 2013, 819).
Personal experience
It would seem reasonable to expect that personal experiences at the polling station impact voters’ views of the electoral process, but studies that attempt to make this link do not reach any consensus. In Mexico and in Russia, problems at the polling station did not figure as largely as consideration of the broader political context in which elections take place. That context determined how seriously voters viewed polling station irregularities (McAllister and White 2011, 675; Schedler 1999, 135).
In contrast, Hall, Monson, and Patterson showed that in the context of American elections, the way in which voters perceived poll worker performance significantly impacted voter confidence (Hall, Monson, and Patterson 2009, 519). Elklit and Reynolds's analysis of eight African countries revealed that individual experiences related to the conduct of elections had a direct bearing on how the sense of political efficacy developed in individuals and this is an important factor behind the development of legitimacy and the progression towards democratic consolidation (Elklit and Reynolds 2002, 88).
Election-related institutions
There is general consensus that electoral management bodies (EMBs) play a pivotal role in the quality of an election and therefore on how citizens perceive elections (Birch 2008, 308; Hartlyn, McCoy, and Mustillo 2008, 76). Robert Pastor emphasised this point, saying that EMBs ‘can determine whether an election is a source of peaceful change or a cause for serious instability’ (Pastor 1999, 5). In developing countries Pastor claimed that independent EMBs are especially important because the technical problems of election management can very easily converge into the political ones, thereby threatening the entire process (Ibid., 10). More recent studies confirm Pastor's claims. Lehoucq asserted that election commissions have the power to ‘make democracy stick’ (Lehoucq 2002, 31). He used the cases of Costa Rica, Chile and Uruguay to demonstrate that over time the devolution of electoral governance to independent commissions made for much smoother political competition. Now, these countries are renowned for clean, hotly contested elections. Debrah, Asante, and Gyimah-Boadi (2010, 4) echoed this opinion in their study of the Ghanaian electoral commission, which they claimed enjoys high public confidence because it ‘has shown tenacity’ in ensuring credible elections.
An EMB's degree of independence and autonomy can have serious implications for electoral processes, and this is partially why Elklit and Reynolds also asserted the importance of EMBs (2002, 87). They explained that any election evaluation must include an assessment of the EMB's independence, and they also emphasised EMBs’ centrality to elections, responsible as they are for everything from the design of the ballot, to districting, to campaign regulation and the counting of results (Elklit and Reynolds 2005, 151). Indeed, the lack of credible EMBs was bemoaned in the context of Africa in the 1990s, as it became clear that such bodies were critical to emerging democracies trying to organise regular elections (Jinadu 1997, 1–5). In the aftermath of the 2007–2008 post-election violence in Kenya, the report of the Independent Review Commission (IREC or Kriegler Commission) identified the Election Commission of Kenya's lack of independence as one of the fundamental reasons for the debacle that was the 2007 election (IREC 2008, 10). In his study of Nigerian elections, Kerr confirmed the importance of EMBs, finding that the performance of election-related institutions, especially the election commission, is what mattered most to Nigerians in their assessments of elections (Kerr 2013, 819).
Partisanship
A third dimension of citizens’ perceptions is partisan loyalty. Several studies demonstrate the importance of partisanship in shaping citizens’ views of elections (Alvarez, Hall, and Llewellyn 2008, 761; Anderson et al. 2005; McAllister and White 2011, 676; Moehler 2009, 346; Rose and Mishler 2009, 126).
Partisanship provides a perceptual frame for interpreting information regarding the quality of elections: those affiliated with the winning party are more inclined to believe that the success of their party was legitimate, while those affiliated with the losing party search for evidence to show that their favored candidate or party was shortchanged. (Kerr 2013, 821)
Despite the fact that partisanship overlaps with ethnicity in Kenya (discussed below), there is little evidence of the explicit consideration of ethnicity in voters’ opinions. Also, while there are studies that try to assess how ethnicity impacts voters’ choices in terms of candidates/parties, there is little research on how ethnicity factors into how citizens assess the credibility of elections. A related study by Alvarez, Hall, and Llewellyn (2008, 764), however, did show that race played a role in American voters’ assessments of their elections. Specifically, white Republicans were more likely to approve of certain elections than black Republicans.
Following the literature, in this study I consider the role of individual experience with electoral irregularities, confidence in election-related institutions, and ethnicity (which overlaps with partisanship in Kenya) in the shaping of voters’ opinions of the 2013 Kenyan election. I also consider local historical context, and I suggest that Kenyan conceptions of electoral integrity are deeply grounded in past experiences with electoral violence.
Ethnicity and elections in Kenya
Kenyan political parties have always been largely based in ethnic identity and revolved mainly around ‘big men’ of particular ethnic groups (Lynch 2006, 234). Today, as in the past, political parties are seen to primarily represent the interests of the ethnic groups of their leaders (Ibid., 235). Thus, in the context of the 2013 election the Jubilee Alliance was strongly affiliated with the Kikuyu and Kalenjin co-ethnics of Uhuru Kenyatta and William Ruto, while the Coalition for Reform and Democracy (CORD) Alliance was linked with the Luo and Kamba co-ethnics of Raila Odinga and Kalonzo Musyoka. There is no strong ideological basis on which the parties were founded, and electoral campaigning had little to do with promoting specific economic or political policies (Freedom House 2013).
This is based in a long history of patronage politics, whereby co-ethnics of those in power tend to benefit disproportionately with regard to jobs, education and other valuable development. Such targeted patronage was the norm under the first three presidents, Jomo Kenyatta, Daniel arap Moi and Mwai Kibaki (Amutabi 2009, 61; Barkan 2008, 2).
This patronage-based system becomes especially relevant during elections, when unscrupulous politicians exploit inter-ethnic grievances and fears to win votes. This has often resulted in bouts of ethnic violence, including, perhaps most traumatically, in the 2008 post-election brutality (Cheeseman, Lynch, and Willis 2014, 1, 4). Triggered by the announcement of Mwai Kibaki as the winner amidst widespread accusations of fraud, the violence pitted Kalenjin, Luo and other opposition supporters against Kikuyus, the core of Kibaki's supporters, and some Kisii. The Commission of Inquiry into Post-Election Violence, or the Waki Commission, eventually confirmed 1133 deaths and 700,000 displaced (CIPEV 2008, 272; Lynch 2009, 604). This carnage convinced Kenyans and the international community of the urgent need for real political reform (Githongo 2010).
The ethnic animosities so regularly exploited by politicians are deeply rooted in a historical narrative of purposeful inclusion and exclusion from state power.2 For Luos, the narrative of purposeful exclusion is informed by patterns of patronage, which have flowed disproportionately to Kikuyu and Kalenjin communities, but also by former presidents Kenyatta's and Moi's actions to subvert opposition groups and persons that threatened their positions, many of which came from the Luo community (Kanyinga and Long 2012, 37; Kiai 2008a, 165; Oloo 2007, 96; Throup 1993, 371; Truth, Justice and Reconciliation Commission 2013, 445).
Central to this Luo narrative is the Odinga family. Jaramogi Odinga led the opposition during President Kenyatta's rule, championing minority rights and eventually falling foul of the president for such efforts. His son, Raila Odinga, has been a leader of the opposition for virtually the entirety of his political career. He was detained and tortured for his pro-democracy work under the rule of President Moi, but his supporters have been the most incensed over the state's perceived betrayals in 2002 and 2007. In 2002, after winning the presidency with Odinga's help, Kibaki reneged on his promise to create a prime ministerial position for Odinga and to share cabinet appointments with him. The controversy and lack of opacity that marked the 2007 election, as well as the Kriegler Commission's finding that it is impossible to tell who actually won (IREC 2008, 9), contributed to the perception that the election was stolen from Odinga.
The 2013 Kenyan election was thus a milestone, partly because it was the first national election to be held since the debacle of 2007. Kenyans were afraid of a repeat of 2007, and many communities were still dealing with sustained trauma. The international community was also tense, eager for Kenya to regain its place as a beacon of stability and peace in the region (Kiai 2008a, 162). As such, the 2013 election occurred very much in the shadow of the 2007 election, not least because, as political commentator Gathara blogged, Kenyans ‘were assailed with unceasing calls for peace’. Until then, the ‘peace-ocracy’ campaign, as it came to be known, was all-pervasive (Gathara 2013).
The 2013 election was a milestone in several other ways as well. First, it was the inaugural election under the new 2010 constitution, bringing devolution into effect. Second, it was the first national election conducted by the new Independent Electoral and Boundaries Commission (IEBC), which had replaced the much-maligned Electoral Commission of Kenya (IREC 2008, 25–52). Third, the combination of new electoral laws and the introduction of electronic electoral technology was meant to improve public faith in the elections by promoting high levels of transparency and verifiability.
It was also the first election in the world to feature two candidates indicted by the International Criminal Court (ICC) for crimes against humanity, allegedly committed, somewhat ironically, in the previous election. The candidacies of Uhuru Kenyatta and William Ruto of the Jubilee Alliance were thus controversial. After the pair were cleared to run by the High Court, they made the ‘illegitimacy’ of the ICC cases a central part of their campaign and the election was interpreted by some as a referendum on the ICC (Lynch and Zgonec-Rozej 2013, 8). The Jubilee Alliance's main competition came from CORD, led by Raila Odinga and Kalonzo Musyoka.
The 2013 election was relatively peaceful, but it was strikingly divisive. One particularly illustrative example is the ‘Tyranny of Numbers’, a pre-election report predicting a first-round Jubilee victory, based solely on an ‘ethnic census’ and the coalition's ‘bankable’ vote from co-ethnics (Maina 2013b, 3). The report is overly simplistic in many ways, but it does illustrate the severity of ethnic polarisation and the types of calculations and strategies which reinforce such polarisation.
There were also numerous administrative problems which called the integrity of the election into question. Irregularities in the voters' register, in the polling station-level and constituency-level tallying forms, and the widespread failure of electoral technology, which was meant to act as a check on manual counts, contributed to serious doubts regarding the verifiability of the results.
Ultimately, after the IEBC declared Kenyatta the winner, Odinga's coalition and civil society organisations, represented by the Africa Centre for Open Governance (AfriCOG) and the Kenyan Asian Forum, filed two separate petitions in the Supreme Court.3 Odinga's petition alleged that Kenyatta did not garner the required number of votes, and AfriCOG argued that the electoral process did not fulfil legal standards of transparency and verifiability. Ultimately, the Court unanimously ruled against petitioners, and it argued that the problems illustrated in the petitions did not prove that the final result would have been different. Legal scholars have since cited numerous problems with the judgment, casting doubt on the Court's impartiality and competence. The judgment's failure to even note concerns about the irregularities, malpractices and errors in the electoral process has motivated many to question the relationship between the IEBC, the Court and the current regime (Maina 2013a).
Empirical analysis
I test the following hypotheses in the context of the March 2013 general election in Kenya:
H1: Citizens who are from the same ethnic communities as the leaders of the winning coalition are more likely to assess the election as free and fair.
H2: Citizens with positive evaluations of election-related institutions, namely the IEBC and the judiciary, are more likely to assess the election as free and fair.
H3: Citizens who prioritise peace are more likely to assess the election as free and fair.
H4: Citizens who personally experienced irregularities at the polling station are less likely to assess the election as free and fair.
Data and methods
I examine the four potential explanations (ethnicity, confidence in institutions, prioritisation of peace and experience with irregularities) of voters’ assessments in the context of the 2013 Kenyan presidential election using data from a national public opinion poll. Infotrak Research and Consulting, commissioned by AfriCOG, surveyed 2342 registered Kenyan voters in 32 out of the 47 counties. AfriCOG allowed the author to use the survey data for this study. The dependent variable, perceived election quality, captures respondents’ assessments of the 2013 election. Respondents were asked, ‘Do you think the recently concluded election was free and fair?’ Potential responses included ‘Yes’, ‘No' and ‘Don't Know'. I rescaled the original variable into a dichotomous variable, with 1 representing ‘Yes' and 0 representing ‘No'. I tested the ‘Don't Know' respondents separately, scaling the ‘Don't Know' responses as 1 and all other responses as 0. See Table 1 for a description of all variables used.
| Independent variable | Infotrak's question | Choices given by Infotrak to respondents | My coding |
|---|---|---|---|
| Ethnicity | Respondents were asked to name the ethnic group to which they belonged. | Infotrak pre-listed 34 ethnic groups, and respondents could also identify as ‘other'. | Dummy variables for Kikuyu, Kalenjin, Kamba, Luo, Luhya, Mijikenda, Maasai, Somali and Swahili groups |
| Confidence in electoral institutions | On a scale of 1–5, how would you rate your level of confidence in the following institutions? | Not confident at all; Slightly confident; Average confidence; Largely confident; Extremely confident | 0: Not confident at all |
| 1: Slightly confident | |||
| 2: Average confidence | |||
| 3: Largely confident | |||
| 4: Extremely confident | |||
| Prioritisation of peace | On a scale of 1–5 where 1 = strongly disagree, 2 = disagree, 3 = neither disagree nor agree, 4 = agree, 5 = strongly agree, how strongly do you agree or disagree with the following statements? | strongly disagree; disagree; neither disagree nor agree; agree; strongly agree | −2: strongly disagree |
| −1: disagree | |||
| 0: neither agree nor disagree | |||
| 1: agree | |||
| 2: strongly agree | |||
| Personal experience with irregularities | I will read out some statements regarding the recently concluded elections. Kindly tell me whether you agree or disagree with the statements by indicating YES or NO. | Respondents were given a list of scenarios about polling station operations. They indicated whether or not they had experienced the described scenario. | Dummy variables for scenarios experienced by respondents |
| Dependent variable | |||
| Assessment of electoral credibility | Do you think the recently concluded election was free and fair? | Yes; No; Don't Know | Dummy variables for Yes and No; Don't Know tested separately as dummy |
For the first hypothesis, I include dummy variables for the ethnic groups of the leading presidential candidates and their running mates (Kikuyu, Kalenjin, Luo and Kamba), as well as several other smaller groups, some of which also had co-ethnics contesting the presidential race (Luhya and Maasai), and some of which did not (Mijikenda, Somali, Swahili). I evaluate the second hypothesis using a scaled variable of citizens’ confidence levels in the IEBC and the judiciary overall.4 I test the third hypothesis using a variable which scales respondents’ levels of agreement with the statement ‘Peace is more important than free and fair elections.’
For the fourth hypothesis, I use a scaled variable which indicates whether or not respondents experienced at least one polling station-level irregularity.
In order to check for collinearity amongst the institutional variables, I used variance inflation factors, which are shown in Table 2. Since they are all below 10, it is reasonable to conclude that they are not collinear.
| Election assessment on IEBC + judiciary + Supreme Court + Chief Justice: | |
| IEBC | 1.39 |
| Judiciary | 4.97 |
| Supreme Court | 2.90 |
| Chief Justice | 3.22 |
| Election assessment on IEBC + judiciary + irregularities + price of peace: | |
| IEBC | 1.34 |
| Judiciary | 1.33 |
| Irregularities | 1.01 |
| Price of peace | 1.01 |
I also control for location (rural/urban), gender, level of education and age. I used dummy variables for location and gender and numerical scales for education and age. All Infotrak respondents were registered voters, such that results reflect the opinions of those who had enough political interest to go through voter registration. I would have liked to control for socioeconomic class, but the Infotrak variable was based on respondents’ reported income and profession, data which is not always accurate (Mattes, Bratton, and Davis 2003, 4). In addition, more than half the sample did not respond to the question, which resulted in a sample size of less than 1000.
Descriptive statistics
Free and fair?
Overall, a majority of the respondents (63.5%) felt that the 2013 presidential election was free and fair, but over one-third expressed doubts in electoral credibility, with 24.4% reporting that it was not free and fair and 12.2% reporting that they did not know if the election was credible.
A close look at why respondents viewed the election in the way that they did reveals that the definition of ‘free and fair’ is not necessarily straightforward. Respondents’ top three reasons for a free and fair assessment were the absence of problems at the polling station, ‘general transparency’ and the absence of violence. Strikingly, the relative peace was essentially as commonly cited as evidence of a free and fair election as the lack of personal experience with irregularities.
Experience with irregularities
Such a high level of public approval of the election is especially surprising considering how widespread electoral irregularities were. Nearly 80% of respondents (78.9%) indicated that they experienced an irregularity. The list of irregularities included names missing from the register, non-registered people being allowed to vote, serious delays in voting because of long lines, IEBC staff incompetence, failure of electoral technology, registered voters being turned away from the polls, bribery of voters, coercion of voters to select particular candidates, inadequate materials at the polling station, violence and a difference in the results as witnessed at the county/constituency level and results announced in Nairobi.
Confidence in institutions
Just under half of all respondents had high confidence in the Supreme Court, the Chief Justice, the IEBC and the judiciary, while between a quarter and a third of respondents had low levels of confidence in these institutions. This is a drastic drop in confidence from just before the election, when 76% and 93% of respondents to a Kenya National Dialogue and Reconciliation Monitoring Project (KNDR) survey reported confidence in the judiciary and the IEBC, respectively (KNDR 2013, 38). The KNDR and Infotrak surveys did use different samples, but the results are different enough to indicate a significant drop in public approval. The most striking case is that of the IEBC. Almost one-third of all respondents (32.7%) revealed that they have no confidence at all in the election commission. The Supreme Court is not far behind, with 31.3% of respondents expressing no confidence in the highest court of the land.
Statistical analysis: voters’ conceptions of electoral credibility
Table 3 shows the results of the LOGIT regression (Table 3).
| Independent variable | Coefficients (marginal effects) | Standard error |
|---|---|---|
| Ethnicity: Kikuyu | .131*** | .020 |
| Ethnicity: Kalenjin | .102*** | .019 |
| Ethnicity: Luo | −.198*** | .063 |
| Ethnicity: Luhya | −.094** | .038 |
| Ethnicity: Kamba | −.008 | .029 |
| Ethnicity: Mijikenda | −.131 | .068 |
| Ethnicity: Maasai | .059* | .028 |
| Ethnicity: Somali | .021 | .041 |
| Ethnicity: Swahili | −.025 | .101 |
| Confidence in the IEBC | .085*** | .009 |
| Confidence in the judiciary | .056*** | .009 |
| Prioritisation of peace | .030*** | .007 |
| Personal experience with irregularities | −.029 | .020 |
| Location | −.026 | .017 |
| Gender | .011 | .019 |
| Age | −.005 | .005 |
| Education | −.005 | .009 |
*Significance level 10%.
**Significance level 5%.
***Significance level 1%.
The results indicate that Kenyans assessed the credibility of the 2013 election based on (in descending order) their ethnicity, confidence in electoral institutions, their prioritisation of peace and their personal experience with irregularities at the polling station.
First, ethnicity played the strongest role in determining how most of the surveyed Kenyans assessed elections. For the Kikuyu, Kalenjin and Luo, the groups most closely aligned to the top two candidates, the results are most statistically significant. The smaller groups were also influenced by ethnicity, but it was generally to a lesser degree.
In general, this finding contradicts respondents’ self-reported explanations, where the most commonly cited reason for a negative assessment was technological failure. This is somewhat unsurprising, given that respondents who are accustomed to surveys, as is the case in Kenya, are reluctant to admit ethnic biases (Bratton and Kimenyi 2008, 276). Table 5 shows the breakdown of opinion by ethnicity.
The vast literature describing patronage politics explains the rationale behind ethnic voting, but there is less scholarly examination of how patronage politics impacts voters’ electoral assessments. When considering voting patterns, it is easy to understand that voters who wish to maximise their chances for socio-economic advancement will vote for candidates who are most likely to facilitate such advancement (Horowitz 1993; Lindberg and Morrison 2008, 101; Vicente and Wantchekon 2009, 300). In the context of a patronage-based system where elites have historically channelled benefits to co-ethnics, it is unsurprising that voters would vote for co-ethnics.
In this case, however, the finding suggests that ethnicity also plays a role in the actual assessment of election credibility. That is, Kenyans aligned to the Kikuyu–Kalenjin Jubilee Alliance were more likely to trust the electoral process, while those supporting the Luo-dominated CORD were more likely to doubt the credibility of the process. This suggests that voters’ conceptions of ‘free and fair' are more about the outcome than the administration of the election. After all, nearly 80% of respondents experienced irregularities, but generally ethnicity still exerted the strongest influence.
Such opinions are grounded in historical lessons. When Moi came to power, he followed Kenyatta's example, redistributing positions within the civil service to favour members of his own Kalenjin group (Amutabi 2009, 61). If voters believe that an electoral victory for a non-co-ethnic will result in discrimination, it stands to reason that their conception of a fair election will hinge largely on victory for a co-ethnic. Perhaps a free and fair election, in the minds of Kenyans, cannot be separated from what particular communities will face, depending on who wins. The Jubilee government's cabinet appointments are a stark reminder of why it is still rational for Kenyans to believe that a co-ethnic will reward his or her own. Prior to the election, Kenyatta promised to appoint a Turkana to head the Ministry of Energy and Petroleum, which would have ensured that the community whose home area is the site of Kenya's newly discovered oil would have a say in further exploration there. In the end, however, Ruto appointed a fellow Kalenjin to run this ministry (Opalo 2013). Given such examples, it is unsurprising that only 10.9% of surveyed Luos were extremely/largely confident that the new government would treat their community fairly, compared with 55.4% of Kikuyus.
The results also indicate that smaller communities are not as strongly influenced by ethnicity, perhaps because their numerical weakness means it is virtually impossible for one of their co-ethnics to win power. Indeed, smaller groups like the Mijikenda, Somali and Swahili did not appear to consider ethnic identity in their assessments of elections. Even the Kamba, who did have a co-ethnic candidate standing as Odinga's running mate, were not strongly influenced by ethnic identity. This could be because another prominent Kamba politician, Charity Ngilu, was a part of the Jubilee coalition, thus splitting the Kamba vote.
In order to test whether it is indeed the case that smaller groups are more influenced by factors other than ethnicity in their assessments of elections, I used the entire dataset to test the impact of all independent variables except ethnicity on electoral assessments and compared those results with the same test on a subset of only the small groups. See Table 4.
| Independent variable | Coefficient (marginal effects) | Standard error |
|---|---|---|
| (a) Assessments of elections (entire sample) | ||
| Confidence in the IEBC | .111*** | .009 |
| Confidence in the judiciary | .075*** | .010 |
| Prioritisation of peace | .049*** | .007 |
| Personal experience with irregularities | −.041 | .021 |
| Location | −.022 | .019 |
| Gender | .019 | .020 |
| Age | −.003 | .005 |
| Education | −.000 | .009 |
| (b) Assessments of elections (sample with only smaller groups, excluding Kikuyu, Kalenjin, Luo, Luhya) | ||
| Confidence in the IEBC | .179*** | .024 |
| Confidence in the judiciary | .129*** | .026 |
| Prioritisation of peace | .053*** | .018 |
| Personal experience with irregularities | −.039 | .057 |
| Location | −.066 | .048 |
| Gender | .015 | .049 |
| Age | −.005 | .013 |
| Education | .024 | .022 |
*Significance level 10%.
**Significance level 5%.
***Significance level 1%.
The role of irregularities is relatively unchanged, but the coefficients for confidence in electoral institutions and for the prioritisation of peace are stronger, suggesting that smaller groups were less swayed by ethnic considerations when evaluating elections.
It is also worth noting the results of tests that examine what contributed to the opinions of those respondents who said they did not know whether the election was free and fair. Here, confidence in institutions does not have a clear influence, as those who trusted the IEBC were more doubtful about electoral credibility while those who trusted the judiciary were less doubtful. Respondents who did not believe that a flawed election was worth peace were more likely to report that they did not know if the election was free and fair.
Also, Luos and Kambas were the only groups more likely to report that they did not know about the credibility of the election. In fact, almost one-third (31.4%) of surveyed Luos fell into this category. The same pattern is evident among the other non-Jubilee-aligned groups. When given the choice, respondents from the losing ethnic groups expressed doubts in the fairness of the election, but they did not all resoundingly assert that the election was not free and fair. See Table 5.
| Was the election free and fair? | |||
|---|---|---|---|
| Ethnic group | No (%) | Don't Know (%) | Yes (%) |
| Embu | 5.80 | 5.80 | 88.41 |
| Kalenjin | 5.45 | 7.39 | 87.16 |
| Kamba | 32.91 | 15.61 | 51.48 |
| Kikuyu | 4.63 | 3.97 | 91.40 |
| Kisii | 47.78 | 12.32 | 39.90 |
| Luhya | 46.59 | 20.88 | 32.53 |
| Luo | 54.55 | 31.36 | 14.09 |
| Maasai | 12.64 | 8.00 | 79.31 |
| Meru | 11.94 | 4.48 | 83.58 |
| Mijikenda | 42.31 | 19.23 | 38.46 |
| Somali | 21.82 | 7.27 | 70.91 |
| Swahili | 47.37 | 10.53 | 42.11 |
Source: Infotrak poll, May 2013.
This may be a sign of movement away from ethnicity-based voting. Other signs of change include new restrictions on ‘hate speech’, the explicit prohibition of ethnically based political parties and diversity requirements for all parties: see the Constitution of Kenya, Article 91(2)(a) (Republic of Kenya 2010); the National Cohesion and Integration Act, 2012, Article 62(1) (Republic of Kenya 2012); and the Political Parties Act, 2011, Article 7(b) (Republic of Kenya 2011)). These new laws have yet to be strictly implemented, but it is possible that future elections will move away from ethnic politics.
Second, in line with findings by Kerr (2013), Lehoucq (2002) and Pastor (1999), respondents who expressed confidence in the IEBC and the judiciary were more likely to approve of the election. Results were statistically significant for both the IEBC and the judiciary. Given Kenya's legacy of weak political institutions, the fact that citizens now use their confidence in institutions to evaluate elections is encouraging and speaks to reform efforts, especially in the courts, where there is now vetting of judges, plans to establish a High Court in each county and general efforts to greatly improve efficiency.
The correlation between high confidence in institutions and confidence in the electoral process also reveals the enormous power of institutions to bolster public faith in the process and in the benefits of democratisation (Bowler and Donovan 2002, 372). Independent and well-functioning institutions are critical to the consolidation of democracy, primarily because they provide political actors with incentives to play by the rules of the game. ‘Political forces comply with present defeats because they believe that the institutional framework that organises the democratic competition will permit them to advance their interests in the future’ (Przeworski 1991, 19). Indeed, Odinga's willingness to go to the Supreme Court and his respect for its decision showed his confidence in the new judiciary and stood in stark contrast to 2008, when his supporters preferred to air his grievances in the streets (Cheeseman, Lynch, and Willis 2014, 13).
Of course, public faith in these institutions has taken a blow since the election. More than a quarter of respondents (26.2%) felt either not at all confident or only slightly confident in the courts and almost one-third (32.7%) of respondents felt the same way about the IEBC. Indeed, more than half of the respondents who indicated that they would not vote in the next election indicated that it was because they had no confidence in the IEBC.
Without more changes to the system, it is possible that public confidence will erode, which could lead to decreased electoral legitimacy in the long term. Indeed, there are some worrying signs. According to legal scholars, the Supreme Court's judgment was severely lacking. Not only did it fail to address evidence presented by the petitioners, it relied on ‘mean-spirited and cramped Nigerian legal precedent’, which had actually been deplored by Nigerian lawyers (Maina 2013a, 2013b). In addition, the IEBC has, to date, failed to release polling station-level results for the five non-presidential elective offices, and 2585 polling station results for the presidential election are still missing.
The third most powerful influence on citizens’ perceptions of the election is their prioritisation of ‘peace’. Not only did a majority of respondents agree/strongly agree with the statement that peace is worth a flawed election, those who tended to agree with the statement were more likely to assess the election as free and fair.
The focus on peace above all else is not surprising. The devastation of 2008 was the worst political violence since independence, crippling Kenya's reputation as a bastion of peace and stability in the region. In the aftermath of one of the worst attacks, in which 50 people were burned alive inside a church, analysts speculated about the possibility of genocide (Schaller 2008, 341). In addition to the loss of life and trauma inflicted upon survivors and witnesses, the violence also severely impacted economic growth, abruptly ending ‘the most robust economic performance in recent memory' (AfriCOG 2010, 5). AfriCOG research indicated that business turnover declined by 30% in 2008 and a further 25% in 2009 (Ibid., 2). The violence sent hundreds of thousands of displaced Kenyans to camps, where they languished in dire conditions. Needless to say, Kenyans feared a repeat of such violence (Long et al. 2013, 141). The fear was due partly to sporadic outbreaks of violence in the lead-up to election day, including communal clashes, gang violence and suspected acts of terrorism. Some of the worst violence impacted police officers, 42 of whom were killed in just one attack in Baragoi in November 2012. It is not clear that all these incidents were directly related to the election, but there is some evidence that candidates were involved in supporting at least some of the violence (HRW 2012). Against this backdrop, and with memories of 2007 still fresh, it is hardly surprising that there was a surge of ‘peace messages’ urging Kenyans to reject violence and keep the peace, whatever the result.
Sadly, however, most of this peace propaganda was one-dimensional; the messages did not contain civic education information, nor did they address the root causes of the violence. In fact, many messages implied that ordinary Kenyans were the ones to blame for the devastation of 2007–2008. For example, ‘Diaspora Messenger’ urged Kenyans to vote without violence:
People across the world care about the future of Kenya and we urge you to VOTE IN PEACE. We know it is possible and it's why we've made this petition to demonstrate our support … [The 2010 constitutional referendum] … demonstrated Kenya could rise above ethnic and racial differences to vote peacefully … We call on all Kenyans to respect the rule of law in the lead up to election day, on election day and afterward. Stand together and the rest of the world will stand with you. (Diaspora Messenger 2013)
In reality, of course, the lion's share of responsibility for the violence lay with political elites who used divisive and inflammatory ethnic rhetoric to ignite fears and who organised, armed and financed the attacks (CIPEV 2008, 121–124).
In the end, peace largely prevailed, but the messaging appears to have convinced Kenyans that ‘peace at any cost’ is most desirable. Describing the Kenyan media's coverage of the 2013 election, Michela Wrong reported that the ‘feisty’ Kenyan media had been replaced with a ‘zombie army’, revealing a society terrified by its own capacity for violence. Wrong went on to describe glaring irregularities, including results on the screen above the IEBC Chairman's head not adding up even as he was congratulating the media on its ‘exemplary behavior’. She asked whether the media's behaviour ‘allowed another major abuse to occur right before their eyes’ (Wrong 2013).
The correlation between a positive electoral assessment and the belief that a flawed election is worth peace reveals how grounded local assessments are in historical context. The fact that a peaceful election is seen as a credible election is likely due to the violence that ravaged Kenya in the aftermath of the 2007 election. Kenyans feared a repeat of that violence so intensely that they were willing to overlook serious problems in the name of peace. It is worth noting that the large majority of the 2008 violence occurred after the polls had closed. The violence did not actually impact the electoral process as much as it did the perception of the outcome. Again, free and fair in Kenya appears to be understood more in the context of the outcome than the process itself.
It seems doubtful that Kenyans would have assessed an election as credible if there had been violence, even if that violence had not impacted the technical administration of the election. This is important, because it emphasises the importance of local context. Certainly, it may be true that some electoral standards are now widely accepted ‘global norms', but those norms alone may not attract legitimacy ‘on the ground'.
Fourth, and perhaps most surprising, is the relatively minor role of personal experience with irregularities. Personal experience with irregularities does decrease the likelihood that a voter will view the election as free and fair, but this variable lacks statistical significance. This finding carries perhaps the most important implications for electoral credibility. If voters care relatively little about errors made at the polling station, that suggests that their conception of free and fair does not hinge on administrative and technical procedures which could impact final results. Given that international monitors focus much of their attention on the credibility of polling station operations, this finding reveals a significant discrepancy between ground-level and policy-level understandings of fair elections and democracy.
It seems, then, that citizens’ endorsement of elections may be more about how that election fared against their own personal hopes and fears than about predefined international standards. Future research on free and fair elections will benefit from the consideration of voters’ conceptions of their own elections, because it is what matters to them that will ultimately determine political legitimacy of their elected governments.
Conclusion
It may be true that a growing number of countries are accepting ‘global norms’ of electoral integrity, but it is important to remember that local context can also be an important contributing factor to the development of voters’ perceptions of credible elections. Despite election observers’ focus on the more technical aspects of election administration, Kenyans on the ground in 2013 were relatively unmoved by their personal experiences of irregularities at the polling stations. Instead, their assessments of the election were influenced more by their ethnic identities, their confidence in electoral institutions and the value they placed upon ‘peace’. These findings underscore the importance of local context, and they suggest that local, ‘on the ground’ conceptions of free and fair elections may be more about what the result means for certain groups than about electoral administration.