Version classiqueVersion mobile

Kenya’s Past as Prologue

 | 
Christian Thibon
, 
Marie-Aude Fouéré
, 
Mildred Ndeda
, 
et al.

Getting it ‘Wrong’ (Again)?

Wahojiwa vs. Wapiga Kura in the 2013 Kenyan Election1

Thomas P. Wolf

Note de l’auteur

Dr. Wolf is a Research Analyst for Ipsos-Kenya (formerly Ipsos-Synovate, formerly Synovate, formerly The Steadman Group; citations of these results use the operative company name at the time). Based on the author’s access to the data as well as the company’s leading profile among the public, through the media, this paper relies primarily on Ipsos’ survey-results. The analysis and views expressed in this paper are his own.

Texte intégral

Introduction

  • 1 In English, “survey respondents” vs. “voters”…
  • 2 Each of these two alliances was a combination of several political parties. For CORD, these were Od (...)

Raila Odinga and Uhuru Kenyatta are locked in a neck and neck race heading into the last five days of Monday’s General Election.
According to three opinion polls commissioned by the Nation Media Group, a runoff [sic] between Mr. Odinga of the Coalition for Reforms and Democracy (Cord) and Mr. Kenyatta (Jubilee) is inevitable.2
The polls show that Mr. Odinga would win during the second round of voting, largely benefitting from Musalia Mudavadi’s supporters. Mr. Mudavadi of the Amani coalition is tipped to claim third position in the March 4 General Election (Daily Nation, 2013b).

A cruel hoax is being recklessly but gleefully played on innocent Kenyans. This hoax is the self-serving computer-generated numbers touted as opinion polls by pollsters.
These pollsters, if one blindly believes them, are showing that the two leading presidential candidates are tied and we should prepare for a second round of voting. Nothing could be further from the truth.
These polls are utter nonsense and I refuse to believe one single one. I am a religious believer in pollsters, but credible ones that use science and not fiction, black magic or fantasy [Abdullahi, 2013].

  • 3 The writer is here quoting P. Githae of The People Daily.

1“How did the pollsters get it so wrong?!” Based upon the significant gap between the official results of the 4 March 2013 presidential election and the final set of survey results released less than two weeks earlier – together with the widely publicized assumption that a second round, run-off contest was inevitable – it was this question that reverberated most loudly in its aftermath. Indeed, as one Kenyan journalist put it shortly after the election: “The pollsters in Kenya have lost face. They were clueless as to what was happening on the ground” (Kabukuru, 2013a).3

  • 4 This is true notwithstanding significant variations between Kenya’s four leading survey firms’ fina (...)

2Given the repeated claims by representatives of the survey firms of the ‘scientific’ nature of their work in the face of a widely skeptical political class, and a public for whom the ‘toy’ of random sample surveys remains relatively new, this remains an important question. And this is so even if, in contrast to the previous election, the country was spared any serious reactive-violence, attributed by some – at least in part – to the “false expectations” such polls had raised then, since none had shown incumbent President Mwai Kibaki as leading.4 Moreover, and notwithstanding the Supreme Court’s dismissal of the two petitions filed (by Odinga and Africog, a local governance NGO) to contest the official results as declared by the Independent Election and Boundaries Commission (IEBC), this chapter will show that even after the election, questions remain as to just how ‘wrong’ the pollsters were, and to the extent they may have been, what could account for this.

3In addition, however, there are other (and perhaps less contentious) questions that are also addressed, including especially: (1) what did the polls show about the evolving field of candidates prior to the formation of running-mate coalitions, and what factors appear to have caused whatever major (and largely uncontentious) shifts in the survey figures that occurred during and after this period; (2) how much did these poll results change as a consequence of the completion of voter-registration that was conducted for a one-month period from 18 November 2012, and then after the official identification of presidential candidates with their deputy presidential running-mates the following month; and (3) how did the survey firms respond to their having ‘gotten it wrong’ after the results were officially declared?

4Attempting to answer these questions first requires some appreciation of the context in which these elections occurred.

2013 Election background: Competitive uncertainty and the ‘Certitude of Surveys’

  • 5 Cotrell and Ghai (2013) provide a review of the 2013 election’s democratic/ constitutional shortcom (...)

5In seeking to understand the role of pre-election polls in Kenya, of critical importance – beyond the imperative of possessing state power itself – are two related realities. The first is the tentative or ‘quasi’ nature of Kenya’s democracy, in which elections cannot be assumed to conform to those in established liberal political systems, several aspects of which shall become evident in the description of events and discussion which follows.5

  • 6 Together, they represent just over one-third of Kenya’s population. Relevant here is the fact that (...)

6The second is the increasingly competitive nature of the country’s electoral politics in recent years. Several factors account for this. One was the return to multi-party competition in 1992. Another was the insertion of presidential term-limits in that year (as part of President Daniel arap Moi’s strategy to diffuse the challenge that such an ‘opening’ of political space represented), leading to the retirement of now, two presidents (Moi and Mwai Kibaki), making presidential succession more frequent and thus national leadership itself more uncertain. A third factor has been the nearly equal division in the country between those supporting continuity and change, at least as represented by the continuation in, or barring from, presidential power, the two ethnic communities that have ‘monopolized’ it since independence: the Kikuyu and the Kalenjin (whatever the tensions between these two groups).6 Finally, and with specific relevance to the 2013 election, the fact of power-sharing under the Grand Coalition Government during 2008- 2012 fostered what “some have termed [the country’s] most liberal period since independence” (Holmquist, 2014), an interlude that could not be assumed to continue once the ‘winner-take-all’ presidential system embodied in the new constitution took effect.

  • 7 Some relevant background is provided by Wolf, who noted that as of early 2008, some 55% of Kenyans (...)
  • 8 Removing those who were still “undecided” from those interviewed, the three main survey firms produ (...)
  • 9 Regionally, thefewadditionalpublishedassessmentsofsuch (decidedlyless frequent) surveys related to (...)

7In such an environment, public affairs surveys in general and voter-intention polls in particular have both reflected this freer atmosphere and contributed to it – while receiving considerable public approval – by riveting attention on government performance and the broader competition for power.7 In large part, this was a consequence of the ‘marketing-magnet’ such surveys had become, with most media outlets eager to energetically trumpet (if less frequently prepared to pay for) them, especially after their generally impressive track-record in the previous two elections (2002, 2007) and the two constitutional referenda (2005, 20108); indeed, Kenya appears to lead across Africa in terms of the frequency and breadth of subject matter of such surveys, and their coverage by the media (Wolf, 2009: 281-82).9 As such, it had become more difficult to dismiss their results as ‘fake’ by those inclined to do so, while bolstering – perhaps too uncritically – the confidence of those shown to be initially enjoying significant leads, and later, poised for victory, in actual election campaigns.

  • 10 This claim, based on a number of interviews and casual conversations, is made even if it simultaneo (...)
  • 11 The term is used only in its most formal sense. Odinga, as prime minister, had been ‘sharing power’ (...)

8Beyond just encouraging confidence or despondency, however, in the extended ‘pre-election’ period, these voter-intention surveys played a number of significant roles, especially but not limited to the presidential contest. These include: indicating the size and cross-ethnic distribution of support with regard to assessing candidates’ viability and thus subsequently, party nominations, including potential deputy president running-mate selection; mobilizing important symbolic-rhetorical and material-financial campaign support both domestically and internationally; attracting and holding media (and diplomatic) attention; and guiding campaign strategy once the ballot-choices were set, especially in terms of the contrast between (a) seeking to win over ‘undecided’ voters and (b) maximizing turnout in areas where support was already considered strong.10 While all such factors likewise applied to the 2007 contest, the overall 2013 electoral environment was different in a number of important respects, beyond the absence of an incumbent seeking re-election.11

  • 12 For its part, the new (2009) Political Parties Act required the formal registration of any party co (...)

9Two contrasts were a function of the new constitution: first, that an outright majority is henceforth required for a first round victory, failing which a second round, run-off contest between the top two candidates is to be held; and second, that each presidential candidate must formally name a deputy-presidential running mate at the time of submitting one’s own nomination to the IEBC, rather than selecting a vice-president once in office as was previously the case.12

  • 13 The total number of such candidates was similar: nine in 2007 and eight in 2013.

10Away from the constitution, another contrast relates to the ‘menu’ of the main presidential candidates.13 Prior to the 2007 contest there was never any doubt that Kibaki would seek re-election, whereas only in August of that year did the constellation of his opponents become clear: that it would comprise two major factions/parties: Odinga’s Orange Democratic Movement (ODM), and Kalonzo’s ODM-Kenya, notwithstanding the fact that these two leaders had been united within the Liberal Democratic Party (LDP) in the 2002 contest (in support of Kibaki) and during the first (2005) constitutional referendum when they had successfully fought against the draft’s ratification (which Kibaki supported).

  • 14 He did so without ever formally resigning from ODM which would have necessitated defending his parl (...)
  • 15 There were widespread rumors that most if not all of the costs of such ‘early’ campaign efforts wer (...)

11With regard to this 2013 election, however, Odinga was early on identified as the only serious challenger against whomever the ‘Kibaki’ side would front, though it was unclear whether he would have only one credible opponent. Most prominent initially was William Ruto, MP for Eldoret North and initially Minister for Agriculture representing the Odinga/ODM side of the power-sharing Grand Coalition Government. Almost immediately after breaking with Odinga by the end of 2009,14 he sought to ‘market’ himself as a presidential candidate in his own right, though few took such ambition for the top ‘prize’ seriously. Among various factors contributing to this view were: the modest size of his quite disparate Kalenjin community (14% of Kenya’s population), his association with some of the negative aspects of Moi’s 24-year reign (including several highly visible on-going ‘corruption’ cases in the courts), his vigorous opposition to the widely popular new constitution leading up to the 2010 referendum, and his relatively modest personal wealth (given what is required for such a campaign).15 As such, most observers saw his efforts as aimed at establishing himself as the most valuable running-mate with a substantial ethnic bloc-vote in tow wherever he went.

  • 16 Kibaki defeated him by nearly a 2-to-1 margin. While many observers have suggested ‘common-sense’ r (...)

12Kenyatta, from the country’s ‘First Family’, was in a different category altogether. This was so not just because his own Kikuyu and other ‘Mt. Kenya people’ constitute nearly one-third of the electorate, and his family’s almost limitless wealth, but also because of his previous (however unimpressive) presidential attempt in 2002.16 Yet several factors militated against a Kenyatta candidacy, at least at this time.

  • 17 Though Ruto had returned to Cabinet as minister for Higher Education, he was relieved of this posit (...)
  • 18 On 11 June the Office of the Chief Prosecutor announced that it was “not opposed in principle” to h (...)
  • 19 Whereas a December election would have marked a full five years since the previous one, and the new (...)
  • 20 For a more general discussion on the impact of the ICC cases on the election, see Wolf (2013b).

13Above all was the matter of his case before the International Criminal Court, having become, along with Ruto and four others, a defendant at The Hague from January 2012 when the charges against them (first unveiled in December 2010) were confirmed, stemming from the violence that followed the 2007 election. At a minimum, there were four aspects to this fact.17 First, some expected, or at least hoped, that the leadership-integrity provisions of the new constitution (in Chapter Six) would prevent them both from contesting (Saturday Nation, 2011). Second, even if they were not barred from doing so, it was considered even less likely, at least until mid- 2012,18 that either would be able to effectively campaign due to the expected commencement of their cases well before the election.19 Third, whatever Kenyatta’s and Ruto’s common (anti-ICC) cause, many likewise assumed that due to the enmity between sizable proportions of their respective communities extending back much earlier than the 2008 post-election violence, a Kikuyu-Kalenjin alliance was an electoral non-starter (Lynch, 2011: 182-200; 211-12).20

  • 21 As one local ‘alternative’ publication put it some two years before the election: “Both the Kikuyu (...)

14A final factor militating against a Kenyatta candidacy was the profound anxiety among a small but highly influential group of political and business leaders associated with the Kibaki/Party of National Unity (PNU) side of the Grand Coalition Government. Two issues were involved here. One was an assumption (or at least concern) that even if Kenyatta could overcome the obstacles noted above, after a decade of ‘Kikuyu rule’ (under Kibaki, and notwithstanding the previous five years of ‘power-sharing’ with Odinga), the level of the nation’s ‘Kikuyu fatigue’ was too high for him (or anyone else from this community) to mount a viable campaign.21

  • 22 These two factors raised the importance of the United Democratic Front (UDF) as a potential vehicle (...)

15The other was economic anxiety over the ICC issue: that whatever the ultimate outcome of these cases, electing a president with such an ‘albatross’ around his neck (let alone having a running-mate with the same burden) would likely trigger inestimable damage to their own considerable business interests – acquired or at least considerably augmented since Moi’s departure just a decade earlier – in the form of punitive measures imposed by Kenya’s most important trading and development ‘partners.22 As such, a Kenyatta candidacy was simply not a risk worth taking. But if ‘not (yet) Uhuru’, then who?

  • 23 The elusive goal was either having the cases returned to some judicial process in Kenya, or dropped (...)

16Initially, the most likely alternative seemed to be Musyoka, given the vital role he had played in first, shoring up the legitimacy of Kibaki’s dubious December 2007 election by accepting Kibaki’s offer of the vice-presidency, and later, when he was not attending the massive ‘peace’ and ‘prayer’ rallies headlined by Kenyatta and Ruto, undertaking numerous ‘shuttle-diplomacy’ visits in an ultimately futile attempt at the international level to ward off the ICC threat (The Standard, 2011a).23 Indeed, as of March 2011, a Synovate poll found that among those who supported Kenyatta as their future presidential choice, Musyoka had a significant, if modest, lead over Ruto in terms of his preferred running-mate (35% vs. 29%; Synovate, 2011), however hypothetical these options still were, evidently reflecting a combination of his own anti-ICC efforts and Ruto’s recent ICC indictment (along with other factors mentioned above).

  • 24 Earlier, the Musyoka-led Wiper Democratic Movement party (having been re-named from ‘ODM-Kenya’) ha (...)
  • 25 As one observer was reported to have commented in response to the several changes made on both side (...)

17From mid-2011, however, Musyoka suffered increasing isolation by the rest of the ‘G-7’, the set of leaders of Kenya’s larger ethnic groups (aside from the Luo) united by an anti-Odinga agenda. At least in part, this was due to the, stark shift in the stance on the ICC issue by (fellow-Kamba) Musyoka-associate and Minister for Justice, Constitutional Affairs, and National Cohesion, Mutula Kilonzo, who suddenly began to trumpet support for it (Daily Nation, 2012a; The Standard, 2012d). Such exclusion, eventually reflected in Kilonzo’s replacement by Eugene Wamalwa, younger brother of the late Vice-President Kijana Wamalwa (while he suffered de facto demotion to the Education ministry),24 suggested (to some, including Wamalwa himself) that the latter was subsequently being considered for this role. And this ‘demotion’ was effected even if it was the vice-president himself – responsible for Kilonzo’s appointment to the cabinet in the first place – who formally made this recommendation to President Kibaki.25

  • 26 According to several accounts, Ruto would have remained as his running-mate (for example, Weekly Ci (...)

18However, in the absence of any other viable ‘Mt. Kenya’ candidate following the death of Internal Security minister (and former vice-president) George Saitoti in a mysterious helicopter crash in June 2012 (KTN, 2013), and later with the quick-collapse of the ‘agreement’ with Musalia Mudavadi in December that he stand in Kenyatta’s stead (Capital-FM, 2012),26 the ‘Kenyatta option’ to an Odinga presidency was the only credible one left.

  • 27 Significantly, however (aside from the final pre-election survey when he was paired with Musyoka), (...)
  • 28 Such fluidity – if not simply self-serving opportunism – among the political elite with regard to t (...)

19Altogether, such uncertainties on the anti-Odinga side of the political divide thus made the poll numbers during 2011-2012 even more important to key actors than they had been hitherto. For example, according to a March 2011 survey, while Odinga enjoyed a clear lead with 38%, Kenyatta and Ruto were far ‘behind’, with 18% and 8%, respectively (with another 8% either “undecided” or unwilling to reveal their presidential preferences; Synovate, 2011).27 Moreover, as late as one month to the deadline of the filing of binding coalition agreements (4 December 2012), uncertainties remained as to just which leaders, with their (in several cases, new) political parties in tow, would be aligned with whom (i.e., deputy presidential running-mates, preferred Speakers in the two houses of parliament, and so on; Sunday Nation, 2012).28

  • 29 Note, however, that as early as September, 2009 an alternative publication had reported meetings be (...)
  • 30 Two factors reportedly thwarted Odinga’s ‘advances’: his inability to provide convincing ICC-protec (...)
  • 31 He further claimed that he, Kenyatta and Saitoti had “signed a pact” that only one of the three of (...)

20Aside from the aborted effort to have Kenyatta withdraw for Mudavadi (cited above), the most striking example of such uncertainty was the eventual Odinga-Musyoka partnership, not cemented until the first week of December (Daily Nation, 2012c).29 Earlier, in August 2012, the latter claimed that he and Ruto had agreed on a joint ticket (presumably, with himself the presidential candidate; Sunday Standard, 2012). And just a week before the Kenyatta-Ruto announcement, Odinga had attempted to strike a similar deal with the latter (Daily Nation, 2012c),30 while barely a week before he joined Odinga as his running-mate, Musyoka claimed that Kenyatta and Ruto had earlier agreed to support his candidacy but had subsequently ‘betrayed’ him (The Star, 2012).31

  • 32 Such a summary-point also ignores the actual unfolding of events following the official designation (...)
  • 33 In this regard, the criticism that conducting presidential voter-intention polls without pairing ca (...)
  • 34 For example, Ipsos-Synovate’s January, 2013 poll found that 11% of Kalenjin (among those who claime (...)

21The above discussion suggests the general point that responses to the pollsters’ main question – “If the next presidential election were held now, whom would you vote for if that person is a candidate?” – from surveys conducted in advance of the forging of key alliances and the official identification of candidates, are less indications of actual voting intentions as opposed to more general measures of individuals’ popularity;32 moreover, most (though not all) Kenyans appear to eschew voting for candidates lacking any conceivable chance of winning – which the polls themselves help to provide – however much they might admire or identify with them.33 In the context of the 2013 election, therefore, it was only from mid-January (when the relevant question was changed to ask: “Which pair of candidates for president and deputy president do you intend to vote for in the forthcoming election?”) that the polls began to approximate the election’s voting patterns.34

22At the same time, whereas survey results are based on ‘private’ interviews in which respondents appear generally confident that their identities will not be revealed, the same cannot always be said for their choices at polling stations. Indeed, pre-election intimidation in favor of certain parties/candidates is apparently quite widespread, which could also cause voters to ‘change their minds’ as they cast real ballots (Ferree et al, 2014: 16). However significant such considerations, the basic question remains: how true a reflection of voters’ intentions are such ‘late-hour’ polls, at least at the time they are conducted?

Crunching – or ‘massaging’ – the numbers?: The Pollsters vs. the IEBC

  • 35 In the final set of poll results prior to the 2007 election, two firms gave Odinga a 2% lead and tw (...)

When a politician complains about an independent opinion poll, it could mean the polling was conducted without undue influence from vested political interests. This is a good thing given our recent history when the country moved close to a civil war over disputed presidential results.
At the time, at least one pollster was accused of having been influenced to swing the results of its research one way, and although no evidence was ever provided to prove this, the huge gap between the pollster’s results and others was puzzling [The Standard, 2011].35

For years, Odinga’s campaign relied heavily on pollsters, who continuously painted him as leading in the polls. This costly mistake made Odinga lethargic and made him live in a “make-believe-bubble” based on the assumption that victory was already assured [Kabukuru, 2013a].

The final outcome had local analysts scratching their heads. The difference [between the polls and the official results] was surprising and way outside the margin of error of what most local pollsters had predicted…
The mystery of how the polls were at variance with the final official IEBC results confound many to date [The Standard, 2014].

23With the above context in mind, the record of Ipsos’ presidential voter-intention polls over nearly a four-year period leading up to the March 2013 election may be considered (Fig. 1).

  • 36 The results were compiled from SPEC (Social, Political, Economic and Cultural) Barometers April 200 (...)

Figure 1*: Support for potential/actual (main) presidential candidates: 2009-201336

Figure 1*: Support for potential/actual (main) presidential candidates: 2009-201336

*For reasons of visual clarity, percentages are shown for only the top two candidates, their (eventual) deputy president running-mates, all others combined, and those who preferred no one/were still undecided.

24Several points may be derived from these trend-figures. First, in the initial aftermath of the formation of the Grand Coalition Government in 2008, many of Odinga’s supporters assumed that the position of prime minister would become permanent following its incorporation into a new constitution (the specification of which was well underway by the middle of 2009). Consequently, at the start of 2009, relatively few (about 20%) respondents identified him as their presidential choice for the next election. However, towards the end of that year, and especially after January 2010, when both sides of that government abruptly agreed on a presidential system (Wolf, 2010), an increasing number of these supporters (among others) expressed their intention to vote for him as president, with his overall level of support peaking just after its August 2010 promulgation at nearly 50%.

25Second, immediately thereafter (January 2011), Kenyatta’s rating rose markedly, following first, his ‘anointing’ as the undisputed future leader of the Kikuyu by cabinet minister John Michuki (The Star, 19 October 2010), and subsequently, his identification as one of six “main suspects” in the cases opened by the ICC’s Chief Prosecutor, Luis Moreno-Ocampo, in December 2010.

  • 37 Note again here that this was first survey conducted after completion of the voter registration exe (...)

26Third, from mid-2011 through 2012, and with support for all other candidates largely depressed to between 10% and 20%, the gap between Odinga and Kenyatta was relatively constant, with the former’s support nearly unchanging at about one-third of all respondents. Once their deputy president running-mates were known, however, support for both jumped significantly (by more than 10%) in January 2013, with a concomitant decline in support for all others, and dramatically so in terms of the proportion of those still undecided, which had swung upwards in December when a brief moment of major uncertainty occurred on both sides of the main political divide (as noted above): whether Kenyatta would be replaced by Mudavadi, and who would be Odinga’s running-mate. Most significant, however, Odinga still enjoyed a clear lead (of 6%), and one that was statistically unchanged since September 2012 (when both were 10% lower (36% vs. 30%).37

27Finally, the last surveys in mid-late February by four firms revealed that this gap had been closed, putting the two main pairs of candidates in a statistical tie, though neither within 5% of a first round win, as the balance needed to achieve this remained with a combination of support for all the ‘minor’ candidates combined (about 8%, most of this for Mudavadi) and those still undecided and/or who declined to reveal their preferences (about 3%), as shown in Table 1:

  • 38 The Table is adapted from Wikipedia (2013). Sample sizes for Strategic PR and Consumer Insight were (...)

Table 1: Final pre-election survey results of four Kenyan firms38

Table 1: Final pre-election survey results of four Kenyan firms38
  • 39 One month earlier, Infotrak released results showing with 51%, Kenyatta with 39%, 7% the remaining (...)

*No breakdown is available for “others” / “undecided” or those who refused to answer (RTA) in the Consumer Insight poll.
Note39

  • 40 For example, and whatever the accuracy of the official results, turnout at the Coast was just 70% i (...)
  • 41 After the Supreme Court controversially ruled that ‘rejected’ votes should be excluded in calculati (...)

28Such figures, collectively having far less variance than was the case in 2007 (Wolf, 2009: 288), underscored the fact that the contest between the top two candidates was so close that (aside from any irregularities) voter turnout would likely determine which of the two would win most votes, as well as whether either could attain a first round victory.40 When the IEBC declared the official results, this is just what Jubilee managed to (barely) achieve, obtaining 50.07% of all votes cast.41 So where did this leave the survey firms, and Ipsos-Synovate in particular?

The post-election poll debate: Proof or posture?

29In the wake of the IEBC’s declaration, the survey firms found themselves in a quandary: remain silent and thus imply admission of a modest but widely-noted professional error, or alternatively, defend their polls and thereby add their voice to those questioning the IEBC’s figures, even if far less directly than had CORD and Africog in filing their petitions with the Supreme Court.

  • 42 This was two weeks before the Supreme Court issued its judgments on 30 March.
  • 43 These excluded the five counties which were among both of their ‘most-votes’ totals, due to their h (...)
  • 44 Its county-level survey figures were also largely correct. These included especially those for the (...)

30In the event, only Ipsos-Synovate chose to publicly revisit its final survey results, which it did within a week of the IEBC’s declaration of results. In seeking to straddle this divide, it undertook an analysis aimed at defending its credibility while avoiding the then-ongoing controversy about the integrity of the official results. In its media release of 15 March, therefore,42 it based its analysis on the official, IEBC figures (since no alternative ones existed), and (among other calculations) compared the reported turnout rate in the two sets of fifteen counties where Kenyatta and Odinga won most votes,43 which revealed a 6% advantage for the former (Ipsos-Synovate, 2013c). Given how populous these 30 counties are, this gap explained most (but not quite all) of Jubilee’s winning margin, and total votes. Aside from this ‘anomaly’ for the Kenyatta figure, however, the release underscored the fact that all of the official results for the seven remaining candidates were within its survey’s margin-of-error (as shown in Table 2) :44

Table 2: Ipsos-Synovate vs. the IEBC (Media Release)

Table 2: Ipsos-Synovate vs. the IEBC (Media Release)
  • 45 As compared with 5.74% of the vote these candidates collectively obtained, the four pollsters’ esti (...)
  • 46 This included a statement one month before the election from the Finance Minister, Njeru Githae, wh (...)
  • 47 In addition, while none of these ‘minor’ candidates filed an election petition, several of them wer (...)

31An additional factor emphasized in the Media Release was the ‘evaporation’ of 2.14% from the six minor candidates (combined), most of which appears to have gone to Kenyatta, a critical gain considering his miniscule margin above the 50% hurdle.45 Whether such attrition stemmed from a desire of their supporters to avoid the high cost of a possible second round contest (which had received considerable media attention),46 and/or to avoid casting ballots for candidates – however admired during the campaign – who anyone with the least knowledge of the polls knew had no chance of winning, is unclear, and was left to speculation.47

32Beyond the figures presented in the Media Release, it later emerged that this firm had slightly weakened its own case by failing to remove from its calculation the 188 respondents (i.e., 2.95%) who declined to reveal their vote-intentions. Had this been done, the numbers for all the candidates would have been higher, but especially those of the leading two pairs: Kenyatta-Ruto to 46.4%, and Odinga-Musyoka to 45.4%, putting Jubilee’s total within a range of 45.1%-47.7% (based on an error-margin of about +/- 1.3%).

  • 48 Recall here that this survey’s fieldwork dates were 15-19 February. Here also note that while the P (...)
  • 49 At the same time, not one of the other three survey firms – all of which completed their final pre- (...)

33Beyond such re-calculations, however, the firm’s analysts called attention to Jubilee’s gain of 6% since its January survey – a trend, it was argued, that most likely continued over the last two weeks of the campaign.48 It was suggested, therefore, that the question to be asked was not why its poll results were ‘wrong’, but rather, how close to the IEBC’s figures would they have been had an additional survey been conducted the day before the election?49

  • 50 Two lawyers involved, one with each petition, separately estimated that such excluded evidence amou (...)

34Given the drama surrounding the nationally-televised Supreme Court proceedings of the several presidential election petitions, overall reaction to the company’s analysis was muted. Yet, while Jubilee supporters remained silent, several complaints were received (via SMSs and social media) from those associated with CORD on the grounds that since the IEBC figures themselves were ‘incorrect’ (a claim impossible to refute), Ipsos-Synovate, whether by design of default, was simply lending legitimacy to a ‘stolen’ election. And subsequently, the fact that much of the evidence amassed in support of the two petitions was disallowed on technical grounds (and thus never examined by the Court) eliminated a potential opportunity to at least partially ‘rescue’ the pollsters.50

  • 51 Repeated claims were made at campaign rallies and media appearances of getting 55%, such as by Muku (...)
  • 52 He similarly declined to reveal in which country the firm is based.
  • 53 For its part, several months after the election The African Report offered a number of ‘straw-men’ (...)

35Once the drama associated with these petitions had subsided, intermittent criticisms of the pollsters (and especially of Ipsos-Synovate), presumably due to its higher credibility, or at least profile) were forthcoming from both sides. For example, a senior CORD politician complained to the author that “you misled us”, referring to their apparent assumption – as noted above, based on the polls – that a run-off election was inevitable (Personal Interview, Nairobi, 19 August 2013). For their part, and against the backdrop of Jubilee’s robust pre-election prediction of an outright first round victory51 – and their extremely narrow win (in terms of reaching the 50% +1 threshold) – when this subject arose, the most common response by its officials and supporters was a repeat of pre-election accusations that the survey firms were either incompetent, pro-CORD, or both. For example, interviewed for a story looking back at the election one year later, TNA Secretary-General Onyango Oloo ‘revealed’ that the Jubilee campaign had made use of a foreign survey firm (which he declined to identify),52 on the grounds that “we didn’t trust the local ones, since they failed to take into account the ethnic factor” (The Standard, 2014).53

  • 54 Thatiah (2014) displays similar ignorance of national random-sampling in his claim that whereas the (...)
  • 55 Note, however, there is no legal prohibition against doing so, and all raw data that is not client- (...)
  • 56 This is just what was called for by a local columnist (Warigi, 2013), echoing (in more detail) a si (...)
  • 57 The report cited is entitled, “A Report on the Analysis of the Registered Voters and the Prediction (...)
  • 58 Specifically, the several senior Jubilee officials consulted for his story reportedly told the jour (...)

36At the very least, such a criticism is misplaced, however, since random sampling (among registered voters) should accurately capture the electorate’s ethnic profile (among other demographic factors),54 even if (as noted above) local firms decline to release results with such ethnic correlations.55 What could have added value, however, are some finely-tuned voter turnout scenario-estimates based on such factors (and using past experience), which none of the local firms offered,56 a task that might have been performed by either local or foreign analysts using existing data. Indeed, the Jubilee campaign strategy document described by one local journalist appears to have served exactly this purpose (Kabukuru, 2013b).57 Yet this story made no reference to the use of any foreign survey firm; it said only that a team of Jubilee experts had “crunched the numbers” before declaring that ‘UhuRuto’ would win on the first round with more than 6 million votes. As such, it seems doubtful that any such ‘foreign firm’ was used, at least for the actual data-gathering, thus raising the question as to why the impulse to denigrate local firms in the first place?58

Conclusion: Pollsters’ past, pollsters’ future?

  • 59 For more general assessments of the IEBC’s performance that raise questions about the official pres (...)

37As was the case in the previous election (Kanyinga et al, 2010; USAID, 2008; Throup, 2008; KPTJ, 2008), controversy will remain as to the credibility of the official presidential election results.59 A key difference, however, is that whereas in 2007 the dispute was over which candidate obtained the most votes, in 2013 the issue was whether Jubilee actually achieved a first round victory (despite an initial claim by Odinga that he actually won the most votes (NTV, 2013). Given that the message from all the pollsters, as noted above, was the likelihood (if not certainty) of a second round, run-off contest, it was the failure of this to materialize that constituted the main challenge to their credibility. In other words, if Jubilee had obtained even a single vote less than the outright majority it needed for a first round win, the widespread conclusion would have been that ‘the pollsters were right’, even if the statistical distance between ‘right’ and ‘wrong’ in this case is infinitesimal.

38The failure of the second round contest to materialize, therefore, invited a number of quite scathing criticisms, such as this one from a U.S.-based Kenya doctoral student:

  • 60 Issenberg (2012) provides numerous examples of this distinction, and methods to exploit it, in the (...)

The results were nowhere near what the polls had predicted a week to the election. Most opinion polls showed the race as close, with Mr. Odinga leading by two percentage points. My own pre-election analysis several weeks before the election warned against belief in the national polls as they were over-estimating Mr. Odinga’s support. The outcome of the election showed that even I had underestimated the extent to which the national polls had over-estimated Mr. Odinga’s popularity. The polls’ biggest error was to base their results on interviews of registered voters, as opposed to likely voters.60

  • 61 While helping to explain Jubilee’s win, this factor is irrelevant to the pollsters’ collective ‘fai (...)
  • 62 Curiously, Opalo failed to even acknowledge that both Odinga and Africog had immediately announced (...)

In reality Mr. Kenyatta won the election long before voting day. His strongholds, relatively wealthier and more urban, had higher voter registration rates than Mr. Odinga’s (85.4% vs. 69.7%).61 This coupled by [sic] a higher turnout rates [sic] on polling day (88.6% vs. 84%) pretty much assured Mr. Kenyatta’s election as Kenya’s 4th president [Opalo, 2013].62

39While making no specific reference to this assertion, however, it was subsequently contradicted in the analysis of the election-day exit poll conducted by a team of Western academics. They asked the question: “Do the exit poll results confirm the IEBC counts?”, and found that they did not. Specifically, they estimated that after removing from their sample those who had refused to be interviewed, but taking into account the dominant voting patterns in the areas where they voted, “Overall, Odinga took around 45.3% of the vote and Kenyatta 45.6%, a statistical tie”, results that “are consistent with the final Ipsos pre-election tracking poll implemented ten days before the election.” They were therefore behooved to ask:

  • 63 This finding was at a slight yet significant variance from the domestic ‘civil society’ observer pa (...)

Could last minute swings in the share of votes for the third through sixth place finishers to Kenyatta have pushed him over the 50% threshold? It is extremely unlikely: the sum of all the presidential candidates besides Odinga who lost votes moving from the Ipsos poll to the IEBC tally only produces 2.8%. This implies that about half of the 5.27 percentage point increase Kenyatta received when moving from the Ipsos poll to the IEBC count would have had to have come from Odinga supporters – the most unlikely group to switch their votes. Moreover, because the exit poll interviews people directly after they voted, it should capture all last minute swings in support. In sum, registration, turnout, and shifts in candidate preference do not seem to explain the difference between the exit poll and the IEBC results [Ferree et al, 2014: 12-15].63

40What, then, is one to make of this discrepancy? Until and unless further (and conclusive) evidence regarding the conduct of the election emerges, we are bound to suffer the uncertainties posed by this author some two weeks before the actual event:

  • 64 A full year after the election, an Ipsos survey found that only 39% of all respondents were “certai (...)

While restoring credibility – and thus legitimacy – to the electoral process after the debacle of 2007-8 is a far more important goal than is any research objective related to pre-election polls, voting behavior, or the study of politics more broadly, being able to pursue any of the latter depends absolutely on the achievement of the former [Wolf, 2013].64

  • 65 A post-election survey found that only 13% backed the proposition that “a true election result is m (...)

41As suggested above, looking through a broader, ‘governance’ lens, the main contrast between the last two elections and their aftermaths was the fact of the ‘institutionalization of grievance’, in that the ‘loser’ (again, Odinga) agreed to take his dispute to the new Supreme Court, and abide by its ruling against him;65 the question of the ‘true’ results as suggested (if not ‘proven’) by either pre-election or exit polls thus became moot in practical terms.

  • 66 The period could be extended further backwards into the later Moi years, if the ‘cooperation arrang (...)
  • 67 Again, see Holmquist (2014). A case in point is the recent enactment of a raftofsecurityamendments, (...)

42Whatever the case, given this ‘slice’ of history as related to the 2013 election, the fact of Jubilee’s post-election near-monopolization of power at the national level – in stark contrast to the de facto or de jure power-sharing arrangements over most of the previous decade66 – leaves the future ‘space’ available to civil society voices, including survey firms and the media through whom they convey their findings, open to serious question.67 In this regard, their fate in terms of further, more restrictive legislative fetters, cannot be ruled out. As such, how the media, the public, leading political figures and especially the government respond to and/or treat the pollsters is likely to be a more precise ‘barometer’ of the country’s governance environment over the near future than are the survey results they obtain from, and share with, the Kenyan public.

Bibliographie

Bibliography

Abdullahi,A. ‘Opinion polls are a hoax: don’t believe them’, Sunday Nation, 2013. http://elections.nation.co.ke/Blogs/Opinion-polls-are-a-hoax-dont-believe-them/-/1632026/1702594/-/x98u9iz/-/index.html [archive].

Africa Confidential. ‘The Long, Long Vote Count’, 54, no. 13, 21 June 2013.

Africog/KPTJ. Election Series: ‘Election Day 2013 and its Aftermath’, 2013. http://kptj.africog.org/wp-content/uploads/2014/03/Election-Day-2013-and-its-Aftermath.pdf [archive].

Alshahid. ‘Kenya’s referendum: latest opinion Polls show Kenyans support draft law’, July 23, 2010; http://english.alshahid.net/archives/9964.

Basedau, M., G. Erdmann and A. Mehler., eds. Votes, Money and Violence: Political parties and elections in Sub-Saharan Africa. Uppsala: Nordik Afrika Institut, 2007.

BBC News Africa. ‘Kenya media: President Kenyatta signs ‘draconian’ bill, December 17, 2014; www.bbc.com/news/world-africa-25418234 [archive].

Bratton, M., R. Mattes and E. Gyimah-Boadi. Public Opinion, Democracy and Market Reform in Africa. Cambridge: Cambridge University Press, 2005.

Capital-FM. ‘‘Dark forces’ arm-twisted me – Uhuru’, December 18, 2012; http://www.capitalfm.co.ke/news/2012/12/dark-forces-arm-twisted-me-uhuru/2/ [archive].

Carter Center. “Observing Kenya’s March 2013 National Elections: Final Report”. November, 2013.

Cottrell, J. and Y. P. Ghai. “Ethnicity, Nationhood and Pluralism and 2013 Elections”. In eds. Y. P. Ghai and J. C. Ghai, Ethnicity, Nationhood and Pluralism: Kenyan Perspectives, 107-135. Nairobi: Colorprint, 2013.

Chweya, L.“Western Modernity, African Indigene, and Political Order: Interrogating the Liberal Democratic Orthodoxy”. In ed. L. Chweya. Electoral Politics in Kenya. Nairobi: Claripress, 2002: 1-27.

The East African. ‘East African govts to crack down on civil societies, NGOs’, November 1-7, 2014a.

The East African, ‘Kenyatta signs controversial security law’, November 20-26, 2014b.

Election Observer Group (ELOG). ‘The Historic Vote: Elections 2013’, Nairobi; https://2012-2017.usaid.gov/sites/default/files/documents/1860/THE%20ELOG%20REPORT.pdf [archive].

Ferree, K. E., C. C. Gibson and J. D. Long. “Voting behavior and electoral irregularities in Kenya’s 2013 Election”. Journal of Eastern African Studies 8, no. 1, 2014: 153-172.

Gallup. ‘The Grand Coalition Government’, Public Presentation, Serena Hotel, Nairobi, September; http://www.gallup.com/poll/111634/kenyans-speak-about-their-postelection-crisis.aspx,2008 [archive].

Gibson, C. C., and J. Long. “The presidential and parliamentary elections in Kenya, December 2007”. Electoral Studies 28, no 3 (September), 2009: 497-502.

Holmquist, F. ‘Kenya: On Edge, Rights at Risk’, 2014; http://www.jadaliyya.com/pages/index/18734/kenya_on-edge-rights-at-risk [archive].

Infotrak Research and Consulting. ‘Infotrak Poll on the Race to State House January’, 2013; www.nation.co.ke/blob/view/-/1663206/data/447361/-/k4phgf/-/poll/pdf [archive].

Infotrak-Harris. ‘An Infotrak Harris Poll on Kenyans’ Voting Intentions for the August 4, 2010 Referendum on the Proposed New Constitution’.

Inter Press Service News Agency. ‘Kenya’s Electoral Opinion Polling Marred by Suspicion’, February 27, 2013; http://www.ipsnews.net/2013/02/kenyas-electoral-opinion-polling-marred-by-suspicion/ [archive].

Ipsos. SPEC Barometer (Unreleased Data), May 2014a.

Ipsos. SPEC Barometer (Media Release), September 19, 2014b.

Ipsos-Synovate. Political Barometer Survey (Media Release), January 25, 2013a.

Ipsos-Synovate. Political Barometer Survey (Media Release), February 22, 2013b.

Ipsos-Synovate. ‘4th March IEBC Election Results vis a vis IPSOS February SPEC Poll: Voter Turnout Explains (nearly all) the Difference’ (Media Release); March 15, 2013c.

Issenberg, S. The Victory Lab: The secret science of winning elections. New York: Crown Publishers, 2012.

K-24 News. ‘Kabando: Jubilee will win on first round’, January 23, 2013; www.news24.co.ke/MyNews24/Kabando-Jubilee-will-win-in-first-round-20130129 [archive].

Kabukuru, W. ‘The 10 mistakes that Odinga made’, New African, April 4, 2013a; http://newafricanmagazine.com/the-10-mistakes-that-odinga-made/ [archive].

Kabukuru, W. ‘The return of President Kenyatta’, New African, April 4, 2013b; https://newafricanmagazine.com/3658/ [archive].

Kanyinga, K., J. Long and D. Ndii. ‘Was it Rigged?: A Forensic Analysis of Vote Returns in Kenya’s 2007 Elections’. In Tensions and Reversals in Democratic Transitions: The Kenya 2007 General Elections, eds. K. Kanyinga et al, Nairobi: Society for International Development/ Institute for Development Studies, University of Nairobi, 2010: 373-411.

Kegoro, G. ‘A Review of the Supreme Court Judgment’, Saturday Nation, April 20, 2013.

Kenya Stockholm Blog. ‘New Uhuru Party in Secret Recruitment’, August 26, 2011; http://kenyastockholm.com/2011/06/23/uhuru-sneaked-1-2-billion-in-budget-to-pay-fake-company/ [archive].

KPTJ. ‘Unwelcome Evidence’, Truth and Justice Digest, Issue 3, September 15, 2008; https://africog.org/reports/unwelcome_evidence.pdf [archive].

KTN (Kenya Television Network). Inside Story: ‘The 2013 Elections’, 2014. http://www.nairobiexposed.com/2014/04/22/jicho-pevu-inside-story-2013-elections-part-1-part-2-video/ [archive].

KTN (Kenya Television Network). Inside Story: ‘Death in Ten Minutes’, 2013; https://www.youtube.com/watch?v=En-SOqq5bYw.

Lynch, G. ‘Electing the ‘alliance of the accused’: the success of the Jubilee Alliance in Kenya’s Rift Valley’. Journal of Eastern African Studies 8, no. 1 2014: 93-114.

Lynch, G. I Say to You: Ethnic Politics and the Kalenjin in Kenya. Chicago: University of Chicago Press, 2011.

Maina, W. ‘Five key reasons Kenya’s Supreme Court failed crucial election petition test’, The East African, April 20-26, 2013.

Makulilo, A. ‘Where have all researchers gone?’Use and abuse of polls for the 2010 elections in Tanzania”. International Journal of Peace and Development Studies 3, no. 32012: 33-56.

Mars Group. ‘Kenya Audit Report on 2013 Kenyan Presidential Election Results’, 2013. http://www.marsgroupkenya.org/blog/2013/09/03/mars-group-kenya-report-audit-report-on-2013-kenyan-presidential-election-results/ [archive].

Nairobi Law Monthly. ‘Secrets of Election 2013’, 5, no. 12 (December), 2014: 20-29.

Daily Nation. ‘Election to cost Sh25 billion’, January 4, 2013a; http://www.nation.co.ke/News/politics/Election-to-cost-Sh25-billion-/-/1064/1657848/-/7k1nf2z/-/index.html [archive].

Daily Nation. ‘Second round inevitable, Raila will win in run-off’, February 26, 2013b; http://elections.nation.co.ke/news/Raila-and-Uhuru-in-tight-race-poll-shows/-/1631868/1705518/-/8e5956z/-/index.html [archive].

Daily Nation. ‘Kalonzo left out of Uhuru-Ruto political alliance’, February 16, 2012a.

Daily Nation. ‘Hunt for Uhuru vehicle stepped up’, April 3, 2012b.

Daily Nation. ‘Raila and Kalonzo finally sign deal’, December 4, 2012c; http://www.nation.co.ke/News/politics/Raila-and-Kalonzo-finally-sign-deal/-/1064/1635856/-/a3xmex/-/index.html [archive].

Daily Nation. ‘Mutula king of the jungle’ January 1, 2011; http://allafrica.com/stories/201101030165.html [archive].

Daily Nation. ‘Opinion polls correctly predicted public mood’ August 5, 2010; http://www.nation.co.ke/Kenya-Referendum/Opinion-polls-correctly-predicted-public-mood-/-/926046/971884/-/7fvmis/-/index.html.

Saturday Nation. ‘Low turnout at Coast may have been Cord’s undoing’, April 19, 2014; http://mobile.nation.co.ke/news/Coast-2013-General-Election-Cord-Turnout/-/1950946/2286164/-/format/xhtml/-/12uc3wlz/-/index.html [archive].

Saturday Nation. ‘Secrets of Raila meeting with ex-ODM ally’, October 13, 2012; https://nation.africa/kenya/news/politics/secrets-of-raila-meeting-with-ex-odm-ally--832016 [archive].

Saturday Nation. ‘Hague cases set to change the 2012 election game plan’, April 9, 2011.

Sunday Nation. ‘Presidential aspirants race against time’, November 4, 2012.

Ndegwa, C. ‘Opinion polls misleading Kenyans,’ October 5, 2012; http://safariafricaradio.com/index.php/kenya-elections/102-elections-2012/1897-opinion-polls-misleading-kenyans [archive].

Njagi, D. ‘What if Kenya’s election violence points to corrupt pollsters?’, The Africa Report, July 22, 2013; http://www.theafricareport.com/News-Analysis/what-if-kenyas-election-violence-points-to-corrupt-pollsters.html [archive].

NTV. ‘Raila claims he won elections’, March 18, 2013; http://ntv.nation.co.ke/news2/topheadlines/raila-claims-he-won-elections/.

Ongoya, E. ‘Supreme Court too casual in ‘Raila Vs IEBC & Others’; https://www.standardmedia.co.ke/commentary/article/2000082170/supreme-court-too-casual-in-raila-vs-iebc-others, 2013 [archive].

Opalo, K. ‘Post Election Report: Kenyan Elections 2013’, by Joshua Tucker on March 11, 2013 https://themonkeycage.org/2013/03/post-election-report-kenyan-elections-2013/ [archive].

Republic of Kenya. Report of the Independent Review Commission (IREC), Nairobi, Government Printer, 2008.

The Standard. ‘Uhuru: I will not quit, only MPs can push me out of post’, January 26, 2012a.

The Standard. ‘Kanu plans Uhuru ouster over The Hague trial’, January 28, 2012b.

The Standard. ‘Wiper castigates defiant Mutula over ICC remarks’, January 31, 2012c.

The Standard. ‘Storms gather in G7 as Kalonzo ally pays price for utterances’, February 17, 2012d.

Sunday Standard. ‘Opinion polls controversy and why Jubilee hired a foreign pollster’, March 11, 2014.

Sunday Standard. ‘Leaders defend Uhuru, Ruto leadership skills’, January 13, 2013; https://www.standardmedia.co.ke/politics/article/2000074887/leaders-defend-uhuru-ruto-leadership-skills [archive].

Sunday Standard. ‘Kalonzo, Ruto to form coalition for easy win’, August 19, 2012; https://www.standardmedia.co.ke/politics/article/2000064347/kalonzo-ruto-to-form-coalition-for-easy-win [archive].

The Star. ‘Jubilee takes its pitch to Mt Kenya’, February 28, 2013; http://www.the-star.co.ke/news/article-109772/jubilee-takes-its-pitch-mt-kenya [archive].

The Star. ‘TNA, URP betrayed me – VP’, December 3, 2012.

The Star. ‘Ban Opinion Polls – PNU’, May 13, 2011a; http://www.the-star.co.ke/news/ article-63793/ban-opinion-polls-pnu

The Star. ‘Kalonzo Emissaries Approach Raila’, August 31, 2011b; www.the-star.co.ke/news/article-51207/kalonzo-emissaries-approach-raila [archive].

The Star. ‘Michuki remarks can incite chaos’, October 19, 2010; http://www.the-star.co.ke/news/article-81768/michuki-remarks-can-incite-chaos.

Synovate. ‘SPEC Barometer Media Release’, April 15, 2011.

Synovate. ‘Referendum Baseline Public Opinion Poll’ (Media Release), July 23, 2010.

Thatiah, I. ‘When Uhuru Kenyatta pulled fast move in TV debate to surge ahead of the pack’ Sunday Nation, November 23, 2014.

Throup, D. ‘The Count’. Journal of Eastern African Studies 2, no. 2 (July), 2008: 29-304.

de Torrente, N. ‘Understanding the 2011 Ugandan elections: the contribution of public opinion surveys’. Journal of Eastern African Studies 7, no. 3 (August), 2013: 530-548.

USAID. ‘Kenya’s President Lost Disputed Election, Poll Shows’, Frontlines, August 2008.

Voice of America. ‘Kenya Cabinet Reshuffle Changes Presidential Race’, March 26, 2012; https://www.voanews.com/archive/kenya-cabinet-reshuffle-changes-presidential-race [archive].

Warigi, G. ‘Pollsters should focus on voter turnout as that will determine next president’, Sunday Nation, February 9, 2013.

Weekly Citizen. ‘Secret plot to isolate Uhuru begins’, 15, no. 47, November 19-25, 2012.

Weekly Citizen. ‘Will a Kikuyu succeed Kibaki?’, 14, no. 18, May 9-15, 2011.

Weekly Citizen. ‘Kibaki humiliates Kalonzo on phone’, 12, no. 38, September 21-27, 2009.

Wikipedia, ‘Kenya presidential election, 2013’; https://en.wikipedia.org/wiki/2013_Kenyan_general_election.

Wolf, T. ‘Opinion Polls: Factors that might influence outcome’, Sunday Nation, February 24, 2013.

Wolf, T. ‘International Justice vs. Public Opinion? The ICC and ethnic polarisation in the 2013 Kenya Election’. Journal of African Elections 12, no. 1 (June), 2013: 143-177.

Wolf, T. ‘Kenya’s New Constitution: Triumph in hand, testing times ahead?’, in Judiciary Watch Report: ‘Constitutional Change, Democratic Transition, and the Role of the Judiciary in Government Reform: Questions and Lessons for Kenya’, Vol. 8, 2010: 23-94, International Commission of Jurists-Kenya.

Wolf, T. ‘Poll Poison’?: Politicians and polling in the 2007 Kenya Election’. Journal of Contemporary African Studies 27, no. 3 2009: 279-304.

Notes

1 In English, “survey respondents” vs. “voters”…

2 Each of these two alliances was a combination of several political parties. For CORD, these were Odinga’s Orange Democratic Movement (ODM) and Kalonzo Musyoka’s Wiper Democratic Movement, and for Jubilee, William Ruto’s United Republican Party (URP, registered in January, 2012) and Kenyatta’s National Alliance (TNA, registered five months later).

3 The writer is here quoting P. Githae of The People Daily.

4 This is true notwithstanding significant variations between Kenya’s four leading survey firms’ final polls: an Odinga advantage between 2% and 10% (Wolf, 2009: 288); an election-day exit poll gave Odinga a 6% margin (Gibson and Long, 2009). Such an accusation-attribution was subsequently repeated often, such as by several Kibaki-aligned MPs in their (eventually successful) effort to enact the Publication of Electoral Opinion Polls Act in 2012 (The Star, 2011a).

5 Cotrell and Ghai (2013) provide a review of the 2013 election’s democratic/ constitutional shortcomings. For an earlier conceptual critique of ‘liberal’/’Western’ elections with reference to Kenya, see, Chweya (2002), and for Africa more generally, see, Basedau et al (2007).

6 Together, they represent just over one-third of Kenya’s population. Relevant here is the fact that a Kalenjin (Moi) served as deputy to Kenya’s first president (Jomo Kenyatta) during most of the latter’s fifteen years in power, and that for all of his twenty-four year incumbency (excepting the final three months), the country’s second president (Moi) partnered with Kikuyu vice-presidents (that could be selected at will, according to the previous constitution, from among elected Members of the National Assembly).

7 Some relevant background is provided by Wolf, who noted that as of early 2008, some 55% of Kenyans believed that such vote-intentions surveys “strengthen democracy”, whereas only 12% felt that they “weaken” it, the remainder being uncertain about this issue, or having no opinion (2009: 293). In a mid-2014 survey, 72% expressed either “a lot” or “some confidence” in survey firms, as opposed to 22% with “only a little” or “no confidence” in them (the latter figure being just 8%; Ipsos, 2014b).

8 Removing those who were still “undecided” from those interviewed, the three main survey firms produced the following results in their final polls: Synovate – 73%; Infotrak – 72%; Strategic Public Relations – 77%, as compared with the official result of 72% (Daily Nation, 2010).

9 Regionally, thefewadditionalpublishedassessmentsofsuch (decidedlyless frequent) surveys related to elections range from the scathingly negative by Makulilo (2012) for Tanzania’s 2010 exercise, to the more positive in connection with Uganda’s 2011 election by de Torrente (2013). For an early pan-African treatment of the increasing profile of more general (non-election related) survey research, see Bratton et al (2005).

10 This claim, based on a number of interviews and casual conversations, is made even if it simultaneously invites more systematic research.

11 The term is used only in its most formal sense. Odinga, as prime minister, had been ‘sharing power’ with Kibaki, according to the terms of the National Accord, while Kenyatta, who was serving as one of two deputy prime ministers (representing the Kibaki/PNU side of the government), had ‘incumbent-like’ status as a function of this very fact. As such, neither of these candidates could easily re-invent himself as an ‘outsider’.

12 For its part, the new (2009) Political Parties Act required the formal registration of any party coalitions with the Registrar of Political Parties.

13 The total number of such candidates was similar: nine in 2007 and eight in 2013.

14 He did so without ever formally resigning from ODM which would have necessitated defending his parliamentary seat in a by-election. Earlier in that year, Odinga had suspended Ruto from the Cabinet in connection with allegations of a maize-importation scandal.

15 There were widespread rumors that most if not all of the costs of such ‘early’ campaign efforts were covered by ‘benefactors’.

16 Kibaki defeated him by nearly a 2-to-1 margin. While many observers have suggested ‘common-sense’ reasons why Moi imposed him as the Kenya African National Union’s (KANU) candidate, this subject – including Kenyatta’s own initial receptivity to Moi’s wish – remains to be thoroughly researched.

17 Though Ruto had returned to Cabinet as minister for Higher Education, he was relieved of this position in August 2011 due to corruption charges he was facing in the Kenyan courts – not because of his ICC indictment. The explains why Kenyatta continued to serve as Deputy PM (an appointment which Kibaki pointedly declined to revoke), even though he resigned his cabinet position (as Minister of Finance, as did fellow ICC-indictee Francis Muthaura as Secretary to the Cabinet and Head of the Civil Service) shortly after the confirmation of charges against him (The Standard, 2012a; The Standard, 2012b). Kenyatta also relinquished the chairmanship of KANU, whose National Executive had generated their own pressure on him over the ICC issue (Daily Nation, 2012b).

18 On 11 June the Office of the Chief Prosecutor announced that it was “not opposed in principle” to having the trials commence after March 2013.

19 Whereas a December election would have marked a full five years since the previous one, and the new constitution sets the election in August, in January 2012 the High Court ruled that it should be held five years after the sitting parliament had been sworn in – thus the 4 March 2013 date. Barring constitutional amendment, however, the next election shall be held on the second Tuesday in August 2017.

20 For a more general discussion on the impact of the ICC cases on the election, see Wolf (2013b).

21 As one local ‘alternative’ publication put it some two years before the election: “Both the Kikuyu and Kalenjin factors in the Kenyan presidency are… about to be retired for a while with the 2012 election, which will almost certainly take the top office to a region away from Rift Valley and Central (Weekly Citizen, 2011). Recall here that by this time, the country’s 50 years of independent rule had been evenly divided between these two communities at the presidential level.

22 These two factors raised the importance of the United Democratic Front (UDF) as a potential vehicle for an alternative candidate in case Kenyatta was not able to contest. Such possible use was ironic, however, given that it had initially been registered in mid-2011 on Kenyatta’s behalf by a number of key figures around Kibaki reportedly including his Personal Secretary Dr. N. Wanjohi, Intelligence chief M. Gichangi, Ndaragwa MP J. Kioni, and businessman J. Wanjui, among others when it became clear that he would have to abandon KANU but could not take over Kibaki’s 2007 election vehicle, PNU, that was now in the hands of G. Saitoti (Personal Interview, Nairobi, 28 September, 2014; Kenya Stockholm Blog, 2011).

23 The elusive goal was either having the cases returned to some judicial process in Kenya, or dropped altogether.

24 Earlier, the Musyoka-led Wiper Democratic Movement party (having been re-named from ‘ODM-Kenya’) had threatened to take “disciplinary action” against the Justice Minister if he continued to make such “insensitive” remarks. These included his statement shortly after the confirmation-of-charges that it was “immoral” for anyone to be associated with the accused, which many observers subsequently concluded had triggered this ‘demotion’ (The Standard, 2012c); at the same time, naming his two pet cheetahs ‘Mutula’ and ‘Ocampo’, and their cage ‘The Hague’ in his rural home (Daily Nation, 2011) may not have helped, either.

25 As one observer was reported to have commented in response to the several changes made on both sides of the power-sharing divide, “The biggest target was Justice Minister Mutula Kilonzo”, whose removal from that position “is a big boost” to Kenyatta and Ruto (Voice of America, 2012).

26 According to several accounts, Ruto would have remained as his running-mate (for example, Weekly Citizen, 2012). In the event, Mudavadi pursued his own lack-lustre presidential campaign, its appeal evidently seriously ‘wounded’ by this humiliation.

27 Significantly, however (aside from the final pre-election survey when he was paired with Musyoka), only during the latter part of 2010 did Odinga’s rating exceed 40% (as shown below) – a fact that must have been worrisome for his supporters, and presumably for him as well.

28 Such fluidity – if not simply self-serving opportunism – among the political elite with regard to the previous election is discussed by Wolf (2009: 290- 91).

29 Note, however, that as early as September, 2009 an alternative publication had reported meetings between Odinga and Musyoka operatives with a view towards crafting such an alliance, given Musyoka’s fears “that he might fall out of favour with those who call the shots in Central Province” (Weekly Citizen, 2009); such negotiations were reported again some two years later (The Star, 2011).

30 Two factors reportedly thwarted Odinga’s ‘advances’: his inability to provide convincing ICC-protection ‘guarantees’, as well as to offer anything approaching what the Kenyatta side could in financial terms, such funds considered vital for him to build his URP (Personal Interview, Nairobi, 23 March 2013).

31 He further claimed that he, Kenyatta and Saitoti had “signed a pact” that only one of the three of them would contest the presidency.

32 Such a summary-point also ignores the actual unfolding of events following the official designation of candidates that can shift voter-appeal, including especially campaign strategy as the election approaches, at least for those voters whose support is not a function of ‘automatic’ patronage-reward or ethnic loyalty.

33 In this regard, the criticism that conducting presidential voter-intention polls without pairing candidates with deputy presidential running-mates is misleading, given the new constitution’s requirements, must be accepted (Ndegwa, 2012). Yet, as indicated above, the field was so fluid until nearly the deadline for the presentation of nomination papers that it would have been quite speculative to gauge the popularity of all such potential pairings.

34 For example, Ipsos-Synovate’s January, 2013 poll found that 11% of Kalenjin (among those who claimed to have made up their minds) expressed an intention to vote for Odinga (Ipsos-Synovate, 2013a), a figure which dropped to 9% in its final, February, poll (Ipsos-Synovate, 2013b). In the election, Jubilee obtained over 90% of the vote in 20 out of 41 of Kalenjin-dominant constituencies (Lynch, 2014: 100).

35 In the final set of poll results prior to the 2007 election, two firms gave Odinga a 2% lead and two others gave him a 10% lead (Wolf, 2009: 288). It is unclear, therefore, which particular “pollster” this article was referring to.

36 The results were compiled from SPEC (Social, Political, Economic and Cultural) Barometers April 2009-February 2013. The verbatim question, through the end of 2012, was: ‘If the next election were held now, aside from President Kibaki, who would you vote for, if that person was a candidate?’ As noted above, for the last two (2013) polls, the question was: ‘Which pair of candidates for president and deputy-president do you intend to vote for?’

37 Note again here that this was first survey conducted after completion of the voter registration exercise. For it, and the subsequent, February, survey, only those (randomly-selected) who could produce proof of having registered were interviewed.

38 The Table is adapted from Wikipedia (2013). Sample sizes for Strategic PR and Consumer Insight were obtained from their managements as they do not appear in the Wikipedia entry.

39 One month earlier, Infotrak released results showing with 51%, Kenyatta with 39%, 7% the remaining six candidates, and 3% undecided (Infotrak, 2013).

40 For example, and whatever the accuracy of the official results, turnout at the Coast was just 70% in a region where Odinga received 75% of the vote (and where the level of voter registration was also below its share of the national population), which underscores the fact that even a slightly higher turnout there could have forced a run-off, other things begin equal (Saturday Nation, 2014).

41 After the Supreme Court controversially ruled that ‘rejected’ votes should be excluded in calculating these results, the figures for all candidates increased slightly, especially Kenyatta’s (to 50.5%) and Odinga’s (to 43.7%).

42 This was two weeks before the Supreme Court issued its judgments on 30 March.

43 These excluded the five counties which were among both of their ‘most-votes’ totals, due to their high populations, starting with Nairobi.

44 Its county-level survey figures were also largely correct. These included especially those for the seven largest counties, for which three positions had been polled – governor, senator, and women’s representative – with the survey obtaining correct results for all, aside from two (close) women’s representative contests (in Kakamega and Nakuru).

45 As compared with 5.74% of the vote these candidates collectively obtained, the four pollsters’ estimates for all of them (combined) ranged from 7.6% to 10.4%.

46 This included a statement one month before the election from the Finance Minister, Njeru Githae, who told Kenyans to “make up your minds” on the first round since Treasury did not have the additional Ksh. 5b a run-off would cost (Daily Nation, 2013a), and another one by Kenyatta a week before the election who promised to use this money for a new youth fund if Jubilee secured a first round victory (The Star, 2013).

47 In addition, while none of these ‘minor’ candidates filed an election petition, several of them were said to have (quietly) accused the IEBC of malpractice.

48 Recall here that this survey’s fieldwork dates were 15-19 February. Here also note that while the Publication of Electoral Opinion Polls Act (2012) prohibits the publication of such results within five days of an election, it imposes no restriction on the conduct of research during this period.

49 At the same time, not one of the other three survey firms – all of which completed their final pre-election surveys nearly a week after that of Ipsos (on 26 February, as shown in Table 1) – placed Jubilee (even barely) ahead of CORD, as did Ipsos-Synovate.

50 Two lawyers involved, one with each petition, separately estimated that such excluded evidence amounted to “about 80% of what had been amassed for the case” (Personal Interview, Nairobi, 23 April 2014; Personal Interview, Nairobi, 4 September 2014). For critical appraisals of the Court’s decisions, see, for example, Kegoro (2013), Maina (2013), and Ongoya (2013).

51 Repeated claims were made at campaign rallies and media appearances of getting 55%, such as by Mukurwe-ini MP, Kabando wa Kabando during a television news-discussion over a month prior to the election (K-24, 2013), and by Mvita MP, Najib Balala, at Jubilee’s final rally at Uhuru Park, Nairobi (KTN, 2013a).

52 He similarly declined to reveal in which country the firm is based.

53 For its part, several months after the election The African Report offered a number of ‘straw-men’ arguments to convince readers that the pre-election results the pollsters had served up were “corrupt” (Njagi, 2013).

54 Thatiah (2014) displays similar ignorance of national random-sampling in his claim that whereas the last set of polls had Kenyatta and Odinga in a virtual tie, when Jubilee strategists “factored in population density”, Kenyatta was actually ahead by 7%.

55 Note, however, there is no legal prohibition against doing so, and all raw data that is not client-privileged are always available for purchase, as least from Ipsos.

56 This is just what was called for by a local columnist (Warigi, 2013), echoing (in more detail) a similar suggestion by the author (Wolf, 2013).

57 The report cited is entitled, “A Report on the Analysis of the Registered Voters and the Predictions Thereof.”

58 Specifically, the several senior Jubilee officials consulted for his story reportedly told the journalist in question that they were alerted to the ‘weakness’ of Ipsos-Synovate’s results by its published findings for Nakuru County in its January 2013 survey, which they asserted had “greatly underestimated” ‘UhuRuto’s’ assumed popularity there (Personal Interview, Nairobi, 18 August 2014). This assertion is curious, to say the least. Ipsos-Synovate’s media release of its January survey included no county-level results, since error-margins at this level (for most counties) would have been too high: +/-5.9% for Nakuru (based on 278 respondents). Moreover, if Jubilee had, nevertheless, been able to obtain the results for this county, they would have found that Jubilee enjoyed a 78%-19% advantage over CORD– in statistical terms almost identical with the official election results nearly two months later – 82.4% to 17.6% – (especially taking into account the associated margin-of-error), surely no cause for “hiring a foreign pollster.” Ipsos-Synovate did release results for all key races in Nakuru one month later (based on mobile phone surveys for this and several other counties), but these were published by the media-client who had hired Ipsos-Synovate for the purpose less than two weeks before the election (The Standard, 2013), far too short a period for any firm – local or foreign – to plan, undertake, usefully analyze, and make strategic use of the resultant data. Based on 750 respondents (and thus yielding an error-margin of +/- 3.6%), it showed Jubilee with a 74%-17% advantage over CORD in the presidential race.

59 For more general assessments of the IEBC’s performance that raise questions about the official presidential results, see, Africa Confidential (2013), Africog (2013), Carter Center (2013), MARS Group (2013), and KTN (2014). More recent revelations about the ‘rot’ within the IEBC have only added to doubts about its integrity (Nairobi Law Monthly, 2014).

60 Issenberg (2012) provides numerous examples of this distinction, and methods to exploit it, in the U.S. where the relevant data are far more reliable; just how it could have been applied in the Kenya context is unclear.

61 While helping to explain Jubilee’s win, this factor is irrelevant to the pollsters’ collective ‘failure’, since (as noted above) their sampling frames were adjusted on this basis after the voter registration figures were released (early in January 2013).

62 Curiously, Opalo failed to even acknowledge that both Odinga and Africog had immediately announced their intentions to contest the results.

63 This finding was at a slight yet significant variance from the domestic ‘civil society’ observer parallel vote count exercise. Based on reported results at 7% (952) of all 31,977 polling stations, it estimated Jubilee’s vote at 49.7%, with a 2.7% margin-of-error, yielding a possible range of 47.0%-52.4% (Election Observer Group, 2013: 63). At the same time, if the ‘true’ turnout at all the polling stations where the exit poll was conducted cannot be verified, how were the results at each one weighted to reflect it?

64 A full year after the election, an Ipsos survey found that only 39% of all respondents were “certain” that Jubilee had obtained an outright majority (with a 70%-9% split on this issue between Jubilee and CORD supporters), while 29% were “certain” that they had not (again, with an 8%-54% split between these two sub-sets of the sample; Ipsos, 2014a).

65 A post-election survey found that only 13% backed the proposition that “a true election result is more important than maintaining peace”, with 85% of the contrary opinion (Long et al, 2013: 151).

66 The period could be extended further backwards into the later Moi years, if the ‘cooperation arrangement’ with Odinga’s NDP from 1998 and then its absorption into KANU in March, 2002 are taken into account.

67 Again, see Holmquist (2014). A case in point is the recent enactment of a raftofsecurityamendments, severalofwhichareviewedasviolatingcritical constitutional rights (The East African, 2014b). Earlier, the government proposed legislation aimed at ‘reining-in’ especially governance/human rights NGOs (part of a clear regional trend; The East African, 2014a). These follow what have been described as potentially “draconian” restrictions on the media embodied in legislation enacted the previous year (BBC, 2013).

Table des illustrations

Titre Figure 1*: Support for potential/actual (main) presidential candidates: 2009-201336
Légende *For reasons of visual clarity, percentages are shown for only the top two candidates, their (eventual) deputy president running-mates, all others combined, and those who preferred no one/were still undecided.
URL http://books.openedition.org/africae/docannexe/image/1482/img-1.jpg
Fichier image/jpeg, 40k
Titre Table 1: Final pre-election survey results of four Kenyan firms38
Légende *No breakdown is available for “others” / “undecided” or those who refused to answer (RTA) in the Consumer Insight poll.Note39
URL http://books.openedition.org/africae/docannexe/image/1482/img-2.jpg
Fichier image/jpeg, 28k
Titre Table 2: Ipsos-Synovate vs. the IEBC (Media Release)
URL http://books.openedition.org/africae/docannexe/image/1482/img-3.jpg
Fichier image/jpeg, 63k

Auteur

Research Analyst for Ipsos-Kenya.

© Africae, 2014

Conditions d’utilisation : http://www.openedition.org/6540

Acheter

Volume papier

amazon.fr
Rechercher dans OpenEdition Search

Vous allez être redirigé vers OpenEdition Search