1992 Elections

After incumbent president Daniel arap Moi restored multi-partyism in Kenya at the end of 1991, the Kenyan African National Union (KANU) (the political party headed by Moi) enacted legislation to strip the Provincial Administration and the Supervisor of Elections (under the auspices of the Attorney General) of their previous role in administering elections. The law revived the ECK that Moi's one-party state had kept both legally and practically powerless. The new legislation granted the ECK de jure independence and power to run elections (International Republican Institute 1993; Throup & Hornsby Reference Throup and Hornsby1998). However, Moi took few legal or rhetorical steps to distinguish the formerly powerless, non-independent commission from its new, purportedly autonomous incarnation (Andreassen et al. Reference Andreassen, Geisler and Tostensen1993; Barkan Reference Barkan1993). With the exception of one member, Moi appointed all of the election commissioners before the restoration of multi-partyism.

Because of Kenya's top-down leadership, Kenyans have tended to infer much about a governmental body from its leader, and the ECK's leadership in 1992 did not augur change. Moi kept the ECK's chair, Justice Zakaya Richard Chesoni, who had been responsible for the Commission under the single-party state, in his position. Opposition party politicians claimed Chesoni could not serve as an independent actor: he was affiliated with KANU and opposed multi-partyism (Andreassen et al. Reference Andreassen, Geisler and Tostensen1993; Commonwealth Observer Group 1993). Even more damningly, a series of newspaper articles claimed Chesoni had been mysteriously cleared of bankruptcy charges before his appointment to the new ECK, which heightened perceptions among elites that the ECK chair was beholden to the KANU and Moi (International Republican Institute 1993; Throup & Hornsby Reference Throup and Hornsby1998; Hornsby Reference Hornsby2012). Indeed, the ECK became a primary target for opposition protest in the run-up to the election (Hornsby Reference Hornsby2012). As the International Republican Institute (IRI) (1993: 10) wrote in its observation report, doubts about Chesoni ‘contributed to the distrust many Kenyans felt toward the Electoral Commission’.

The ECK demonstrated meagre capacity and poor planning skills throughout the election cycle. During the registration process, many eligible voters claimed they could not obtain identity cards required to register to vote (an issue that would return election after election). USAID and local opposition actors also reported that registration was systemically more difficult in areas supportive of the opposition parties (Throup & Hornsby Reference Throup and Hornsby1998) and that ward boundaries were gerrymandered in favour of KANU (Adar Reference Adar2000). While we do not observe the baseline level of trust before the 1992 elections, the ECK's performance during the elections does not appear to have improved confidence after the elections, and probably even decreased it, particularly among opposition members.

1997 Elections

According to an International Foundation for Electoral Systems (IFES) pre-election assessment, citizens' opinion of the ECK may have improved in the run-up to the 1997 elections given the Commission's successful administration of many by-elections between 1993 and 1996 (Klein et al Reference Klein, Scallan, Santos de Assunçao and Dauphinais1996: 14). However, citizens' perceptions of the ECK's capacity were offset by questions about its autonomy, when Moi re-appointed seven of the 11 Commissioners from the 1992 elections, including Chesoni, in the run-up to the 1997 elections.Footnote ¹⁰ Moreover, the new Commission's first action increased the number of constituencies to the benefit of KANU (Ng'ethe & Barkan, Reference Ng'ethe and Barkan1998).Footnote ¹¹ Indeed, before the elections, IFES reported that ‘Many Kenyans, however, still point to the Commission as the most important and fundamental flaw in the electoral system’ (Klein et al Reference Klein, Scallan, Santos de Assunçao and Dauphinais1996: 14).

However, the capacity demonstrated during the by-election was not repeated during the presidential elections. The ECK's performance was nothing short of catastrophic, and the elections were plagued with problems with polling stations' opening and closing times and lack of ballot papers (Steeves Reference Steeves1999). The view of the Commission as fundamentally flawed was probably also influenced by a more active media that started to question whether the procedural failures were a result of incompetence or, rather, a KANU scheme to allow Moi to win (Ajulu Reference Ajulu1998).

2002 Elections

In the run-up to the 2002 elections, it is probable that the ECK saw a large jump in public confidence because of its increased autonomy and capacity. First, Moi did not try to amend the constitution which only allowed him two terms after the reinstatement of multi-partyism. This had the dual effect of granting the ECK more opportunity to exercise independence and signalling to Kenyans that the playing field would be more fair. Second, legislators passed a new law, mandating vote counting at the polling station-level (Ndegwa Reference Ndegwa2003), which boosted citizen confidence in the electoral process (Steeves Reference Steeves2006). Less polarisation and lack of an incumbent made the election process easier to manage, and, therefore, made it easier for Kenyans to have trust in the ECK.

While lauding these improvements, many observers noted the ECK's continued lack of independence (Carter Center 2003).Footnote ¹² Additionally, technical challenges continued. For instance, the Commission announced Mwai Kibaki, the main opposition candidate from the National Alliance Rainbow Coalition (NARC), as the president, based on unofficial results, and he was inaugurated before the official final election results were released. Nevertheless, as seen in Figure 1, it is unsurprising that popular distrust was very low after the 2002 elections when Afrobarometer started collecting survey data. As the Commonwealth Observer Group (2002: 25) wrote, ‘the collation process may not have been rapid or elegant, but it commanded the confidence of those involved’. Another reason for the ‘commanding’ confidence in the ECK may have been associated with the turnover in parties with NARC besting KANU.Footnote ¹³

2007 Elections

By 2007, the ECK was in the limelight again, where it suffered a massive collapse in trust after elections. A competitive electoral environment pitted incumbent President Mwai Kibaki against his former ally-turned-competitor, Raila Odinga, and Kenyans looked to leadership and performance of the ECK as signals of trustworthiness.

The Commission itself did not inspire confidence, though it is likely that only keen observers were playing close attention, given the success of the 2002 elections. While the Commission still had 22 commissioners in 2007, President Kibaki had appointed them all, including 19 in the run-up to 2007. Only five of them had conducted an election before. According to monitoring reports written after the election (European Election Observation Mission 2008: 15), ‘The President's single-handed manner in which the appointments in 2007 were made undermined public confidence in the electoral authority’. Branch & Cheeseman (Reference Branch and Cheeseman2009: 25) go so far as to write, ‘the complete collapse of the ECK would not have occurred if it had not been packed by government stooges'. Figure 1 shows the significant collapse of trust across ethnic groups and winners and losers after the 2007 elections.

The story of the ECK's disbandment is well known.Footnote ¹⁴ As in the 2002 elections, the ECK announced Kibaki as the winner, again apparently without first tallying the votes. However, to Kenyan citizens, the outcome probably appeared less legitimate than it had in 2002, as Kibaki was announced the winner with only the state broadcaster present and then inaugurated one hour later, rather than in the large, public ceremony that had occurred in 2002 (Cheeseman Reference Cheeseman2008: 166). Later, ECK Chairman Samuel Kivuitu, a co-ethnic of Kibaki, admitted he had been under extreme pressure to announce the results and stated he had not actually known who had won the elections when he made the announcement. Indeed, the same trends in relationship to electoral management that occurred in the previous elections occurred again: accusations of the ECK being bought off, lacking independence and mismanaging the counting process. On this occasion, however, there was evidence that the ECK's announcement was indeed incorrect, and, furthermore, the election polarised the country along ethnic lines. Widespread post-election violence ensued, claiming the lives of an estimated 1,000 to 1,500 Kenyans and displacing upwards of 300,000.Footnote ¹⁵ According to research conducted by Dercon & Gutiérrez-Romero (Reference Dercon and Gutiérrez-Romero2012), 42% of Kenyans considered the manipulation of the ECK as the main trigger for violence in their communities.

2013 Elections

As a result of the 2007 post-election violence and the Kriegler report issued in its aftermath, a grand coalition government spearheaded the writing of a new constitution. As part of the constitutional reform process, two separate interim commissions were established while the new constitution was being drafted: one responsible for boundary delimitation, and the other – named the Interim Independent Electoral Commission (IIEC) – responsible for elections.Footnote ¹⁶ After the referendum on the new constitution in 2010, the interim bodies were united and made permanent under the Independent Elections and Boundary Commission (IEBC) (see Barkan Reference Barkan2013; European Election Observation Mission 2013).

To ensure the IEBC's independence, any individual with appropriate qualifications could now, for the first time, apply to become an IEBC commissioner, and each nominated applicant had to undergo a vetting procedure. The constitution also banned political party members and those running for elected office from applying. On the back of his success in running the constitutional referendum, the chair of the interim body, Issak Hassan, was appointed the new chair of the IEBC.

On the performance side, the IEBC introduced new technology in the form of biometric voter registration (BVR) and a new computerised results transmission system to remedy two serious problems from the 2007 elections: an inaccurate register of voters and untimely results transmission (Barkan Reference Barkan2013). Despite hiccups, the IEBC generally maintained public confidence in the run-up to the 2013 elections across both sides of the ethnically polarised political divide.Footnote ¹⁷ In fact, our data and others' showed the IEBC to be the most trusted governmental institution in the run-up to the elections, despite a series of issues and accusations, particularly pertaining to the acquisition of the new technology.Footnote ¹⁸

Support for the IEBC after the 2013 election, however, followed Kenya's previous pattern of collapsing post-electoral confidence, but this time it became polarised across groups. According to the election results, Kenyatta won, narrowly escaping a second round run-off with Odinga (see Ferree et al. Reference Ferree, Gibson and Long2014). A series of near-disastrous technological failures marred election day itself (see Barkan Reference Barkan2013). Odinga and his coalition immediately cried foul play. Immediately after the election, Odinga gave a series of speeches with the theme of ‘Democracy on Trial’, in which he stated, ‘What Kenya witnessed was the failure of virtually every instrument it had deployed for the election … the constitutional process has been thwarted by another tainted election.’Footnote ¹⁹ Therefore, according to Odinga, the method to undo this democratic shortcoming was to seek remedy in the Supreme Court.Footnote ²⁰ The Supreme Court, however, found in favour of the IEBC and did not reverse the result, and the IEBC itself remains intact, though confidence in its functioning does not, especially among Odinga supporters.

Why was there differential trust in support for the IEBC after the 2013 election when many of the same elements were present as 2007? Several scholars have focused on the fact that peace prevailed after the elections (Cheeseman et al. Reference Cheeseman, Lynch and Willis2014; Deacon Reference Deacon2015; Harrington & Manji Reference Harrington and Manji2015; Shah Reference Shah2015). Deacon (Reference Deacon2015: 203) describes Kenyans' views of the election in religious terms as a ‘second chance’. It may be that the lack of violence, coupled with the Supreme Court decision, which worked ‘to shield the IEBC from scrutiny’ (Harrington & Manji Reference Harrington and Manji2015: 176), was sufficient for electoral winners to maintain confidence in the IEBC.Footnote ²¹

Overall, we have seen waves of public confidence in the electoral process. Reforms are initiated, and Kenyans appear willing to believe that the EMB is functioning well in the run-up to an election. However, with the exception of the 2002 elections, EMBs have failed to maintain confidence among a broad cross-section of the population after administering an election.

Winner/loser status

Citizens' electoral support for the winner/loser of the most recent election is perhaps one of the most widely explored determinants of institutional confidence (Anderson et al. Reference Anderson, Blais, Bowler, Donovan and Listhaug2005; Moehler Reference Moehler2009; Cantú & García-Ponce Reference Cantú and García-Ponce2015). Supporters of the winning candidate are more likely, relative to others, to positively evaluate regime institutions. Several studies have noted the presence of a winner/loser effect with regards to confidence in EMBs in Africa (Moehler Reference Moehler2009) and elsewhere (Ugues Reference Ugues2010).

There are two reasons why electoral outcomes may shape citizens' orientations toward EMBs. First, campaigns and elections are emotionally intensive periods, the outcomes of which usually affect citizens' psychological disposition (Anderson et al. Reference Anderson, Blais, Bowler, Donovan and Listhaug2005). The positive feelings that winners experience after an election usually foster confidence in regime institutions, while losers' frustration with outcomes might lead to more sceptical evaluations of institutions, especially those that are associated with their candidate's loss. Lack or loss of trust may be particularly acute if certain ethnic or social groups are continually losers. Second, various studies have shown that losing candidates are usually more critical of electoral authorities and EMBs, and publicly highlight deficiencies in EMB performance as justifications for their loss (Bratton Reference Bratton1998; Pastor Reference Pastor1999; Beaulieu Reference Beaulieu2014). As a result, we expect losers will be more accepting of information questioning the impartiality and efficiency of EMBs, and this may in turn structure their attitudes towards the EMB. We consider two ways in which the experience of winning and losing might influence EMB confidence. The first adheres to the traditional literature, where winner/loser status is defined by whether a citizen voted for the winning/losing candidate (Anderson et al. Reference Anderson, Blais, Bowler, Donovan and Listhaug2005).

The second examines the effect of citizens' ethnic affiliation with the winning/losing candidate. Studies have shown that in contexts of high ethnolinguistic diversity, ethnicity matters: it influences political mobilisation, distribution of resources and appointments of political elites (Posner Reference Posner2005). Furthermore, membership in and attachment to an ethnic group structures citizens' political attitudes and political behaviour (Eifert & Posner Reference Eifert and Posner2010). The same is true for Kenya, where ethnic identity has been salient during elections (Lynch Reference Lynch2006, Reference Lynch2011; Gibson & Long Reference Gibson and Long2009; Shah Reference Shah2015). Ethnicity serves as a reliable predictor of voting behaviour, especially when co-ethnic candidates are running for office and have a realistic chance of winning (Bratton & Kimenyi Reference Bratton and Kimenyi2008; Ferree et al. Reference Ferree, Gibson and Long2014). Therefore, we expect citizens who are affiliated with the ethnic group of the losing candidates (parties) to report lower levels of EMB confidence, based on a rationale similar to that for those who voted for the losing candidate. We propose the following two hypotheses:

EMB Performance

Some scholars argue that institutional confidence may be indicative of citizens' rational evaluations of the performance of the institution (Miller & Listhaug Reference Miller, Listhaug and Norris1999; Mishler & Rose Reference Mishler and Rose2001). Building on these insights, we argue that citizens' confidence in EMBs may be associated with their assessments of various dimensions of the institutions' performance during different stages of the electoral cycle.

Hartlyn et al. (Reference Hartlyn, McCoy and Mustillo2008) and Elklit & Reynolds (Reference Elklit and Reynolds2002) highlight autonomy and capacity as two dimensions of EMB performance that are particularly important for citizens' attitudes towards electoral integrity. Autonomy gauges the extent to which electoral authorities can conduct their duties without manipulation by the incumbent political parties or other interest groups. Popular assessment of EMB autonomy may be based on the citizens' perception of the integrity and character of the EMB chairperson, as well as the impartiality of poll officials. Capacity captures the degree to which electoral authorities effectively manage various aspects of the electoral process. This can be evaluated based on several factors, such as whether polling stations open on time, or voters find their names on the voter register.Footnote ²² When citizens believe that EMBs fail to exercise adequate autonomy and/or capacity, this may lead them to lose confidence in the institution, as was the case in Kenya following the 2007 elections (Dercon & Gutiérrez-Romero Reference Dercon and Gutiérrez-Romero2012). Contrastingly, perceived improvements in EMB performance can enhance confidence in the institution, as seen in Nigeria's 2011 and 2015 elections (Lewis & Kew Reference Lewis and Kew2015). Consequently, we hypothesise that:

Political Awareness

In addition to institutional performance, other studies have emphasised the importance of citizens' political awareness. That is, whether citizens have the motivational and cognitive resources to acquire and process information about specific institutions may have crucial implications for their confidence judgements (Gibson et al. Reference Gibson, Caldeira and Baird1998). The link between political awareness and institutional confidence is complex. In the first place, it depends on certain individual traits such as citizens' (1) motivation to invest time and other resources into acquiring information about institutions, (2) cognitive capacity to understand new and to reconsider existing information on institutions and (3) social location that provides varying opportunities to learn about institutions (Cook et al. Reference Cook, Jacobs and Kim2010). Another key dimension of the awareness-trust nexus is the availability of ‘factual’ information about the institution and the identities of the sources of information (government officials, political actors and the media).

The conventional wisdom is that when citizens become more knowledgeable, they will espouse greater institutional trust (Gibson et al. Reference Gibson, Caldeira and Baird1998). However, this expectation is not always borne out, because in some contexts, informed citizens are likely to have higher expectations of institutional performance and become more distrusting when institutional performance falls below these standards (Norris Reference Norris2011). In fact, studies in Africa often find political awareness to be negatively associated with institutional trust (Bratton et al. Reference Bratton, Mattes and Gyimah-Boadi2005; Moehler Reference Moehler2009).

Building upon these insights, we argue that greater political awareness may not simply enhance or diminish institutional trust; instead, the awareness-confidence nexus could be contingent on citizens' political status as a winner or loser. Informed voters are likely to seek out information from co-partisan leaders and partisan media when making their trust judgements. As a consequence, we expect:

Studies have shown that losing candidates/parties are usually very critical of the performance of election administrations (Beaulieu Reference Beaulieu2014). After the 2013 elections, Odinga's accusations against the IEBC were widely disseminated in print and electronic media and further formed the basis of ODM's legal challenge before the Supreme Court. Therefore, we believe informed supporters of the ODM would (1) be acutely aware of criticisms of the IEBC (relative to less informed losers), (2) consider ODM's framing of information to be credible (relative to winners) and (3) probably incorporate this framing into their evaluations of the IEBC.

Dependent Variable: Changes in IEBC Confidence

The main dependent variable, Change in IEBC Confidence (Δ IEBC Confidence), measures reported changes in citizen confidence in the IEBC following the 2013 general elections. Specifically, respondents were asked, ‘From what you may know or hear about the way the March 4, 2013 general election was managed, has your confidence in the IEBC declined, remained the same, or increased?’ Changes in IEBC Confidence is coded on a 1–3 scale with 3 representing increased confidence. Although the survey included a direct question on citizens' level of confidence in the IEBC,Footnote ²⁴ which has been the standard approach in the literature, we decided to assess relative changes in citizen confidence in the IEBC. This is because, when compared with the straightforward confidence question, change in confidence provides a more valid approach for measuring how the events surrounding the 2013 elections may have explicitly altered (positively or negatively) citizens' levels of trust in the IEBC.Footnote ²⁵

Independent Variables

To examine winner/loser status, we construct three binary variables, based on whether citizens reported that they (1) voted for the winner of the presidential election, Uhuru Kenyatta (Winners); (2) voted for one of the other seven presidential candidates (Losers);Footnote ²⁶ or (3) did not vote at all (Non-voters). We expect Losers to report greater declines in IEBC confidence when compared with Winners and Non-Voters.

We gauge the significance of ethnicity through a series of dummy variables, indicating the ethnic groups in which respondents report membership. For ease of interpretation, we focused on the five largest ethnic groups in the data, who are also the most politically salient ethnic groups in Kenya: Kikuyu, Luo, Luhya, Kalenjin, Kamba, and created a sixth variable, Other Ethnic Groups, to include all other ethnic groups. Following H2, we expect co-ethnics affiliated with the main losing presidential and vice-presidential candidates, that is co-ethnics of Odinga (Luo) and Musyoka (Kamba), to be more likely to have withdrawn support for the IEBC, compared with co-ethnics of the winning presidential (Kenyatta/Kikuyu) and vice-presidential (Ruto/Kalenjin) candidate, or of other ethnic groups without dominant presidential and vice-presidential candidates.Footnote ²⁷

We rely on two variables to assess citizens' evaluations of EMB performance. The first, IEBC Performance Index, is constructed using five questions that assess whether the IEBC and its staff are (1) independent from political control, (2) competent, (3) united, (4) of high integrity and (5) impartial in boundary delimitation. Responses to these five items were highly inter-correlated and represented a reliable scale (Cronbach's alpha = 0·90) that was re-coded to range from 0 to 1, with 1 indicating highest perceived performance. The second variable, Quality of Voter Register, probes respondents' opinions on whether the voter register used in 2013 was acceptable. The constitutional reform process emphasised developing an independent, competent and effective EMB, and many reports issued after the 2007 post-election violence saw the overhaul of the voter register as integral to rebuilding confidence among stakeholders. These two variables, then, provide adequate insight into how Kenyans assessed critical dimensions of EMB performance during the 2013 elections.

Because political awareness is a multidimensional concept that is often difficult to operationalise, guided by the existing literature, we include three indicators. The first taps into the frequency of access to different types of news media (Media Exposure). The second variable measures whether citizens report receiving any information about the voting process (Info on Voting), and the third captures citizens' reported level of education (Education).

We also include a series of controls that other studies have considered important, such as citizens' level of political participation based on whether citizens attended political party primaries (Attended Primaries), gender (Female), personal economic conditions (Index of Lived Poverty)Footnote ²⁸ and urban/rural residency (Urban).