On the 24th of June 2021 at around 6 a.m., an insidious hashtag, #KatibaMbichi, appeared on Kenyan Twitter timelines. Its trend seemed to be driven by a number of faceless bots, and retweeted by a series of catfishes that sent it to the number one spot on the Kenyan Twitter trends.
Our investigations have uncovered how such malicious, coordinated, inauthentic attacks that seek to silence members of civil society, muddy their reputations and stifle the reach of their messaging, is a growing problem in Kenya. Twitter, especially, has been central to these operations due to the influence it has on the country’s news cycle.
The proliferation of digital media platforms in Kenya carries the promise of a renewed definition of freedom of speech. Moreover, Twitter has been a vital tool of expression for many Kenyan citizens, many of whom use it to hold their leaders to account and to call out their failures. But civil society members and journalists have increasingly come under attack thanks to disinformation campaigns in the country.
Through a series of interviews with anonymous influencers involved in these campaigns, we accessed their inner workings and gained crucial insights into how they are organized.
An examination of the campaigns has provided our team with a window into the shadowy world of Twitter influencers for political hire in Kenya. Many of the accounts and individuals involved promote brands, causes and political ideologies without disclosing that they are part of paid campaigns.
Twitter features such as the trending algorithm are exploited to achieve the goals of these campaigns by amplifying them. Certain verified accounts on the platform are complicit in leading these attacks. The goal of these campaigns is to exhaust critical thinking and poison the information environment by muddying the truth.
Our investigations examined two months’ data between 1 May 2021 and 30 June 2021, with a particular focus on the Constitutional Amendment Bill—famously known as the Building Bridges Initiative—that was being promoted in Kenya at the time.
With the aid of Twint, Sprinklr and Trendinalia, we trailed the attacks by mapping and analysing specific hashtags that the influencers used on Twitter. This involved mapping certain accounts that posted malicious content targeting Kenya’s activists and judicial officers. The flagged hashtags often displayed synchronized publishing timestamps within the metadata, with a lack of content on most days, followed by one very sharp burst of activity and then fizzling out.
In total, using Sprinklr, which has access to Twitter’s full historical archive, we flagged 23,606 tweets and retweets released by 3,742 accounts under the 11 hashtags. We also obtained 15,350 of these tweets using the Twint package on Github to carry out further analysis of the content.
How disinformation is spread
The Twitter campaigns we looked at were those that were pro-BBI and directly attacked citizens and prominent civil society activists that were vocally opposed to the proposed reforms, and also sought to discredit civil society organisations and activists by portraying them as villains who were being funded by Kenya’s Deputy President William Ruto—he opposes the BBI process.
The well-coordinated attacks are launched through WhatsApp groups to avoid detection. The WhatsApp group admins give direction about what to post, the hashtags to use, which tweets to engage with, and whom to target. They also synchronize the posting to enable the tweets to trend on twitter.
There is money to be made in attacking civil society. Our sources confirmed that they get paid between US$10 and US$15 to participate in three campaigns per day. Those higher up the ranks are on a monthly retainer that can go as high as US$500. Those who are on a retainer supervise the hashtags and ensure that they trend on the days they are posted.
Who the disinformation targets
From our analysis, the top three most frequent victims were Kenyan journalists, judges, and known activists. Prominent anti-BBI activists under the Linda Katiba movement who petitioned the courts against the BBI were the targets of some of the most vicious attacks.
The attacks peaked in early May with the specific goal of trying to discredit the anti-BBI campaign. Jerotich Seii, a key member of the Linda Katiba campaign who was targeted, said in interview that she had to spend a lot of time trying to prove that her activism efforts were genuine and that she was not a front for someone else. “The disinformation attacks against me focussed on painting me as someone with ulterior motives who isn’t interested in the welfare of Kenyans. I had to spend a good chunk of my time defending my position as someone who is actually a patriot who does what they do out of love for their country,” said Seii.
From our analysis, the top three most frequent victims were Kenyan journalists, judges, and known activists.
All this is leading to self-censorship by some of the activists on the platform as they feel that it is pointless to use a platform that cannot deliver any meaningful engagement. One activist we spoke to said that she had significantly scaled down her Twitter activity because of all the trolling she had experienced.
The Kenyan High court struck down the BBI on 14 May on the grounds that the initiative was unconstitutional and the Court of Appeal followed suit on August 20th. The ruling not only strained the already bad relationship between Kenya’s Judiciary and the Executive, it also led to wave after wave of disinformation attacks seeking to question the judges’ judicial independence and the accuracy of their decision.
A notable change in these attacks was how the visual aesthetics of the content within the campaigns evolved; newspaper editorial cartoon-style caricatures and memes were employed, a likely indication of a change of leadership or strategy at the top that sought to make the content more palatable and shareable.
What is the impact of the slander?
The data that we gathered from Trendinalia (which collects data on Twitter trends in Kenya) shows that sufficient amplification was achieved for 8 of the 11 hashtags we identified that became trending topics. This amplification was achieved partly through the use of verified accounts. One anonymous influencer we spoke to said that owners of certain verified accounts involved in these campaigns would often rent them out to improve the campaign’s chances of trending. “The owner of the account usually receives a cut of the campaign loot from the person that rented it from them once it’s over,” the influencer said.
The demand for this service by the political class in Kenya is markedly strong. During the months of May and June alone, we counted at least 31 artificial political hashtags, including the ones linked to the BBI process. This translates to at least one manipulated disinformation campaign that Kenyans have to deal with every two days.
Curiously, there is little evidence that these operations actually sway people’s opinions. However, they do have an effect on how Twitter users interact with their information environment. The goal of such operations is to overwhelm, to create an environment where nobody knows what is true or false anymore. The objective is to exhaust critical thinking and muddy the truth.
During the months of May and June alone, we counted at least 31 artificial political hashtags, including the ones linked to the BBI process.
Typically, a post by any of the prominent activists or judicial officers is bombarded with so much aggression, insults, and dismissive comments that the space for a good conversation is lost. The point is always to ensure that sober-minded people are disincentivized from amplifying the topic after encountering so much aggression in the replies and the quote tweets.
The role of Twitter Inc.
To many Kenyans, Twitter matters. The platform has become a very critical avenue of expression, networking, running ads, and a means of obtaining information. It is also an important avenue for active citizenship as #KOT (Kenyans on Twitter) is one of Africa’s loudest and most lively internet communities.
On the darker side however, some of the features on Twitter are being exploited for nefarious purposes. The platform is failing Kenyans—and Africans more broadly. Political actors are using it to try to control political narratives by poisoning the platform and harassing dissenting voices.
Specifically, Twitter’s trending algorithm, which selects and highlights content without examining its potential for harm, often serves as an on-ramp for users who are trying to find information on the platform. Our sources said that Twitter trends is the primary key performance indicator by which most of their campaigns are judged. They admitted that without it their jobs would not exist. “The main goal is to go trending on Twitter. I’m not sure what our jobs would look like without that target,” said one source.
The evidence available points to the fact that, for the executives at Twitter, this is not a new phenomenon. The trending algorithm in particular, which is a big part of how Twitter works, has been abandoned to disinformation campaigns and attacks.
Twitter’s Moderation Team should pay close attention, keenly monitor and regulate its trending section. Activists, such as Sleeping Giants, have repeatedly called for Twitter to “untrend” itself. This could be done by either removing the feature completely or by disabling it during critical times such as during election periods.
The evidence available points to the fact that, for the executives at Twitter, this is not a new phenomenon.
Arguably, Twitter does not have an incentive to fix this. It sells ads for “promoted trends” and “promoted tweets” within the feeds of hashtags on its trending topics section to business clients. This puts Twitter squarely in the middle of the mess as it profits from this harmful activity.
Ad Dynamo, an agency that sells Twitter Ads in Kenya, currently offers promoted trends for US$3,500 per day within the country. The overall message this sends is that it is ok to sow hate on the platform so long as Ad Dynamo owners can place ads next to the trending content and make a profit from it.
As Kenya heads towards elections in 2022, the demand for these services will increase and many political parties will seek out malicious coordinated trending models and create the risk of a repeat of the 2007 political violence.
Support The Elephant.
The Elephant is helping to build a truly public platform, while producing consistent, quality investigations, opinions and analysis. The Elephant cannot survive and grow without your participation. Now, more than ever, it is vital for The Elephant to reach as many people as possible.
Your support helps protect The Elephant's independence and it means we can continue keeping the democratic space free, open and robust. Every contribution, however big or small, is so valuable for our collective future.
The Pollsters Have Had Their Say, Now Let the Real Voters Have Their Way
This is the fifth in a series of articles that will review and comment on surveys related to the August 2022 general election, providing analytical tools to enable the reader to assess their credibility and potential impact.
Now that the election is upon us, this, my 5th piece for The Elephant, will compare the final round of pre-election survey results released last week—by law no new opinion poll can be published within 5 days to the election—and consider several other issues related to them, individually and collectively. It will then consider the findings of several dubious survey entities; some examples of weak and incorrect interpretation of poll results by the media; aspects of the Act regulating the publication of polls during the pre-election period; a consideration of several factors related to recent change of presidential voting intentions; and the assumed “bandwagon” effect as a consequence of polls.
To provide some empirical and analytical context, it begins with my (slightly updated) piece that appeared in the Standard of 3 August.
The undecideds: the overall picture
By all accounts, the forthcoming presidential election is likely to be a close contest. Two factors underlie this assessment: first, not all of the mainstream survey firms agree in giving Raila Odinga a clear lead over William Ruto (beyond their respective margins of error), even if three do so (Infotrak, TIFA and Ipsos—in contrast to Radio Africa). Secondly, those respondents declining to identify their voting intentions either because they declared they were “undecided” or have chosen to remain silent altogether, remain a significant proportion of potential voters in the findings of nearly all the four surveys: between 4 and 8 per cent. In other words, it can be assumed that it is they who will decide who the next president is, whether they turn out to vote, or stay away, thereby leaving this decision to those who do vote.
Such figures raise two questions: which of these two candidates will obtain the most votes, and will either also obtain the 50 per cent + 1 required for an outright win?
As shown in TIFA’s poll conducted during 21 June-16 July, both candidates increased their overall rating by about 5 per cent (Ruto from 39 to 44 per cent, and Odinga from 42 to 47 per cent). Two factors account for this: a decline in those claiming to be undecided (from 10 to 4 per cent) as well as those who gave no response (from 4 to 2 per cent). Combined with a drop by half in expressed voting intentions for George Wajackoyah (also from 4 to 2 per cent), this explains where the combined gain of 10 per cent for the two main candidates came from.
The remaining proportion who declined to declare their voting intention will be critical. Is there anything about their identity and attitudes also captured in this survey that might indicate which way they will go?
This analysis—based on previously unreleased TIFA survey data from the above-mentioned survey purchased by the Standard—explores two key questions related to the above: just who are these respondents, and does their identification give any clue as to whether and how they will vote? How many Kenyans have changed their voting intentions since the beginning of this year, and for those who have, why have they done so?
Getting inside the numbers
The first point to make is that the proportion of respondents declining to name a preferred presidential candidate has continued to decrease since the beginning of the year. For example, according to TIFA, it dropped by about half, from 30 per cent in January to just 14 per cent in late June. This stands in contrast to the relevant figures for the same period prior to the last three elections: in 2007, 1 per cent; in 2013, 5 per cent; and in 2017, 9 per cent. (It should be noted that the surveys which yielded these figures from the three previous elections were all conducted face-to-face at respondents’ households, unlike the three at issue here. It may be the case that in the former setting, where interviewers and respondents are able to establish a more “personal” relationship, it would be more “awkward” for a respondent to avoid answering this question.)
As noted, the TIFA figure is 5.2 per cent (among all respondents who said they are registered voters, but excluding those who said that they will “definitely not vote just 2 per cent). Yet with only two weeks remaining when this survey was conducted, it cannot be assumed that all such respondents have not, in fact, made up their minds; they may just be too shy to reveal their voting intentions, for one reason or another.
How many Kenyans have changed their voting intentions since the beginning of this year, and for those who have, why have they done so?
One way to test this assumption is to pose this question: what proportion of these respondents who claim to be undecided about their presidential vote have, in fact, made up their minds about the other choices they will confront on their ballot papers?
To answer it, TIFA asked all “undecideds” (on their presidential choice) if they have decided who will receive their votes for governor, senator and MP. Altogether, more than half (58 per cent) reported that they had yet to make up their minds on any of these choices. By contrast, just over one quarter (28 per cent) indicated that they had indeed made up their minds about all three.
It thus appears that these latter respondents—those who had made up their minds on the lower races—very likely decided, for whatever reason, not to reveal their presidential preference although they actually have one. By contrast, it may be assumed that the others are being honest, since if they have yet to decide their vote-choice for any of the other positions, their attention to the campaigns—and perhaps to the election itself—is obviously low (even if they are classified as “potential voters”, that is, those who claimed to be registered but excluding those who said that they “definitely will not vote).
Main reason(s) for being undecided
Another clue about the “real” position of these self-professed undecided voters is provided by the distribution of responses to the next question that all of these respondents were asked: “What is the main thing that will help you decide which presidential candidate you will vote for?” Whereas a substantial proportion (28 per cent) identified “more information about the policies/the manifestos of the candidates” as the main factor that would allow them to decide, more than twice as many (61 per cent) said that they are “not sure” what would help them make this decision. A handful mentioned either “guidance/advice from family, friends or respected local leaders” or “instructions/advice from local candidates/politicians”. At this late stage of the electoral cycle, how anyone could not be certain what they need to know—or whose opinion they need to hear—in order to make this important decision seems improbable, suggesting that this vague response is nothing more than a ruse.
Political party self-identification
Another “window” into the minds (or at least the political profiles) of the “undecideds” is gained by getting the answer to another question: what proportion of them do/do not self-identify with any political party? The assumption here is that far more of those who do may be expected to have made up their mind about their presidential voting intentions, even though, as noted above, they may be shy about revealing this.
The data here indicate that this is indeed the case, with over 90 per cent of those who self-identify with any political party also naming a presidential candidate, as compared to slightly over half of those who do not. Thus, once again, the latter exhibit a much lower level of electoral interest.
At the same time, these “undecideds” regarding the presidential contest, but who do, nevertheless, self-identify with a party are nearly evenly split: 3 per cent doing so with ODM and 2 per cent with UDA, and 1 per cent each with regard to the two corresponding coalitions.
Indeed, only with the benefit of credible official results will it be known if at least some of those declining to reveal their voting intentions have actually concealed them—as happened with significant proportions of respondents in US surveys that were “wrong” with regard to Donald Trump’s victory in the 2016 election and in the UK’s “Brexit” vote the same year. Assuming that is the case, who between Ruto and Odinga will benefit most when the real votes are counted?
Confidence in election integrity
Another factor evidently affecting the naming of a preferred presidential candidate can be put in the form of another question: does confidence in election integrity influence the willingness or propensity to name a preferred presidential candidate? According to the data, only very marginally so. Two sets of figures may be cited. First, slightly more of those who did not name a candidate believe that false election results (for whatever positions) are either “very likely” or “somewhat likely” after the 9 August votes are counted: 56 per cent vs. 52 per cent. Conversely, rather more of those who named a candidate are “quite certain” that this will not happen as compared with those without a (stated) preferred presidential candidate: 40 per cent vs. 32 per cent. (More of the latter are also “not sure” if any election “wins” will be stolen or not: 12 per cent vs. 8 per cent.)
Two demographic variables: education and English language competency
The impact of education on the inclination or ability to make a decision about one’s presidential vote is significant. For example, among those who mention an intention to vote for any candidate, only 26 per cent have not gone beyond primary school, whereas more than double this figure (i.e., 56 per cent) who do not mention any candidate are within this lower education bracket. And regarding language ability, only 33 per cent of those did not name a candidate say they can speak English, as compared with 57 per cent of those who did (clearly largely a correlate of one’s education level). Taken together, such contrasts related to these two variables suggest that at least part of the failure to name a candidate appears to be insufficient access to political information as opposed to a propensity to conceal one’s intentions.
Flip-side: who has changed their presidential preference, and why?
To probe deeper into the process of deciding whom to vote for (which can be mainly either an individual or collective/communal one, although these two factors vary across the country both between and within various communities), TIFA sought to discover just how many people have recently changed their voting intentions since the beginning of the year. Such changes are of three types: from being undecided to supporting a particular candidate; from supporting a particular candidate to being undecided; and to changing support from one candidate to another. Even without data, some of the most salient reasons can be suggested: because the candidate dropped out of the race; because one’s most admired or local political/community leader moved into or out of a relationship with one or the other presidential campaign team or alliance; because of campaign or other messaging; or because of the candidates’ choices of running-mates.
In this endeavour, the first requirement was to determine how many people say that they have done this, and who they are, especially in partisan-political terms.
Altogether, a substantial minority—13 per cent—report having done so. Significantly, a clear majority of these—57 per cent—say that they have changed from one candidate to another, as shown in the following chart:
Moreover, slightly more of those now supporting Odinga than Ruto (a 5 per cent margin) are among these 57 per cent. However, the DP has the “upper hand” in terms of winning over those who had initially been undecided (by a 12 per cent margin). Finally, among those who had previously intended to vote for one presidential candidate but are now preparing to vote for another one or are now undecided (the latter, as noted above, just 5 per cent of the total sample), Raila and Ruto have “lost” a nearly equal number of potential voters (36 per cent and 33 per cent, respectively). Nearly all of the remainder (that is, among those who have “switched” to a new candidate) previously supported someone who never entered or who dropped out of the race (especially Musalia Mudavadi or Kalonzo Musyoka).
Indeed, only with the benefit of credible official results will it be known if at least some of those declining to reveal their voting intentions have actually concealed them.
In partisan terms, an equal (and relatively small) proportion of declared supporters of ODM and UDA (led by Odinga and Ruto respectively) report having changed their presidential voting intentions since the start of the year (11 per cent). However, far more of those who continue to self-identify with Jubilee have done so (26 per cent), and almost as many among those who support any of the numerous other parties (22 per cent), the take-away being that those attached to a party lacking a presidential candidate have been rather more inclined to “move” in one direction or another in terms of their ultimate presidential choice.
But to what do the respondents attribute their changes in voting-intentions? For most (42 per cent), it has been campaign promises and manifesto content, although “the influence of others” and “the choice of running mate” also register significant numbers (13 per cent and 10 per cent, respectively). Less impactful among this group as a whole was the failure of an admired leader to join (or withdraw from) the presidential race (e.g., Mudavadi, Kalonzo, and a few others, at 8 per cent) as well as an evaluation of opinion polls (4 per cent, with Odinga having a definite advantage here, however small the total figure).
Poleni kwa the ‘Undecideds’: What this contest is really about
Even if this is Ruto’s first attempt at the presidency, the considerable duration of his public career even before his decade-long service as deputy president, means that neither of the two main candidates suffers from any “public profile” deficit. Yet one factor contributing to the higher than usual proportion who have chosen to remain “undecided” in recent surveys may be a higher level of confusion as to what agendas they both represent as compared with those who sought the country’s highest office previously. Such “confusion” may stem from the fact that Odinga and Ruto have largely exchanged their political “clothing”, with the deputy president largely campaigning against his president—and thus the status quo, even if throughout their first term they appeared to be in comfortable lock-step with each other—while the former prime minister usually defends him, even as he, too, tries to offer credible change for an electorate the vast majority of which is currently suffering a plethora of economic, and other, woes. Indeed, in a TIFA survey of June 2021, 50 per cent more respondents identified Ruto rather than Odinga as “the political leader most active in terms of criticizing the Jubilee government and trying to hold it to account”, and in TIFA’s April 2022 survey, some three-quarters of respondents identified Odinga as “Uhuru’s preferred successor”. Such a situation makes it largely impossible for Odinga to assume the anti-government posture he has in the last five elections, notwithstanding his short-lived absorption into Moi’s KANU government and party in 2001.
Finally, in more systemic terms, is the fact that Kenya lacks political parties that exist from one election to the next to which are attached contrasting policies relating to issues that matter to voters and that are fairly consistent over time. In “advanced” countries, such parties are often organized around policies related to such issues as taxation and its application to inequality, or the level of government penetration into private lives. In Kenya, though, the “choice” of candidates, even when belonging to different parties, tends to fall back onto perceptions of their personalities and identities (including but not limited to their ethnic groups), as well as onto voters’ feelings about the status quo and the conviction (or hope!) that one or the other candidate will do more to change things for the better.
The impact of education on the inclination or ability to make a decision about one’s presidential vote is significant.
At the same time (and as is often the case in “established” democracies), some voters will seek to “punish” incumbents as a way of expressing their unhappiness with the status quo, rather than expressing confidence in any particular set of would-be leaders.
Given the above, with less than an entirely clear picture as to just “what” either William Ruto or Raila Odinga represents in terms of what Kenya would be like after five years of having either of them at State House, it seems of little surprise that the “undecideds”—and those who admit to having changed their voting intentions within just the last seven months—are as numerous as the research has revealed.
But with only a few hours remaining until “judgment day”, the “mystery” of just how they will use their votes will soon be resolved, with this group of voters very likely to make the difference—assuming they actually get to their polling stations on 9 August. For as has been pointed out, try as they might, they will not find “undecided” as a vote option in the list of presidential options, nor in those for any of the other five offices to be filled.
Let us now consider some additional issues.
Five + two poll comparison: Third horse exit and some integrity questions
To no surprise, there has been a flurry of surveys conducted and released in the last several weeks, with two firms—TIFA and Infotrak—doing this twice during this period, as shown in this table:
|Firms||Sample Size: Margin of Error||Survey Dates/ Method||Ruto||Odinga||Wajackoyah/ Waihiga||Undecided/
|TIFA||2,056: +/-2.2%||21-26 July Household||44%||47%||2%||7%|
|Nation (Infotrak)||2,400: +/-2%||30 July-1 Aug.
|Radio Africa||3,000: +/-1.8%||20-27 July
|Ipsos||6,105: +/-1.8%||23-26 July
|Infotrak*||6,000: +/-1.3%||1-2 Aug.
|TIFA||2,268: +/-2.1%||31 July-1 Aug.
|5 Survey Average|
*In releasing its previous Radio Africa poll a month ago, the Star announced that it would be releasing additional polls “every week between now and the election”, but only the one listed here has appeared. The fact that this poll is the only one putting Ruto ahead of Odinga in the last two months also raised questions in some quarters<
** In its report, Infotrak shows the margin of error in this survey to be only +/-1%.
Several takeaways can be drawn from this “gaggle” of surveys. First, Odinga’s margin over Ruto has increased significantly, even if the latter’s rating has also risen, both benefiting from a substantial increase of those stating they are “undecided” or simply refusing to answer the question, so that Odinga’s gain is not dependent on any Ruto loss. Second, the drop in about half of expressed support for Wajakoyah (together with the failure of Waihiga to register any significant support) means that the prospect of a second round, run-off contest, has ebbed markedly. Indeed, whereas Odinga would just barely cross the first round win-threshold according to TIFA’s late July survey with 50.2 per cent (if the 7 per cent “undecideds” and “no response” are removed), barely a week later in the most recent of all these surveys, this figure rises to 53 per cent, with Ruto at 45 per cent. (The TIFA media Release for this most recent survey of the lot was the only one which included this calculation.) Third, of the six surveys, only the Radio Africa one puts Ruto ahead, but only by 1 per cent, and thus within the survey’s margin of error; nevertheless, this still constitutes its tagging as an “outlier”. Fourth, even if CATI surveys may have a slight anti-Ruto bias (though that by Radio Africa appears to contradict this), it is clear that larger samples do not yield significantly different results, as discussed in a previous article in this series, even if larger samples allow for more “precision” at the sub-national level (as pointed out by David Ndii, see below).
‘Rogue’ or ‘fake’ pollsters
One issue that arose during a recent workshop in Naivasha on “The Media and Opinion Polls” organized by the Media Council of Kenya and that also came up at the IEBC pre-conference forum on the same subject held at KICC a few weeks later, was that of the appropriate response to the results released by “unknown”/unproven (or so-called “rogue”) pollsters. Unfortunately, no attempt was made to define what these terms mean. On occasion (and as I have mentioned in a previous Elephant piece), a pollster may appear “out of nowhere”, as was the case with the British firm, RealField, that in January released the results of a survey conducted the previous month, but has not been heard from since (even if some reports indicate it has been conducting internal polls for Azimio). Yet being a “newcomer” and thus lacking any (at least Kenyan) track-record does not necessarily mean it lacks integrity. This latter category could more genuinely include firms that release results wildly at odds with at least several other known ones with a history of fairly reliable results. At the same time, the term would more accurately apply to any firm that releases results from an actual survey that have been “doctored”, let alone those based on no survey at all. Unfortunately, for whatever reasons, the professional body to which most “reputable”/established firms belong, the Market Survey Association of Kenya—MSRA—appears to make at most minimal efforts to internally query significantly contrasting results of its members. It is thus largely left to the media to decide whether the results from any particular firm warrant publication or broadcast.
The results from two firms that raised some eyebrows after their results were released recently are the following:
|Firms||Sample Size/ Margin-of-Error||Survey Dates/ Method||Ruto||Odinga||Wajackoyah/ Waihiga||Undecided/
|21-26 July Household||45%||52%||2%||7%|
In this connection, at the second of the above workshops, I expressed my disagreement with several other panellists with regard to whether survey results from such “rogue”—or unknown/”out-of-nowhere”—pollsters should reach the public through the mainstream media, however much they deviate from those of the “mainstream” firms. Whereas others felt that they should be ignored entirely, I suggested that this was a mistake, based on faith (or hope!) that scrutiny would do more to limit their influence than silence, given the increasing use of social media to spread such “misinformation” far and wide. The key issue for me is just how they should be covered. In general, there are three options. One is neutrality, so that their results and methodological details are simply treated as “news”. Another is to interrogate them, in part by comparing them with other “mainstream” findings, so as to (indirectly but clearly) raise questions about their credibility (assuming they deviate substantially from the latter—which they invariably do). And a third option is to do this, but with an added “pinch” of interrogation (or even ridicule) about the possible motive(s) behind them, including questioning whether any interviews were actually conducted. Such questioning could be accompanied by any information gleaned from communication with those involved in such polls. A place to start with such interrogation is to (at least) report their level of compliance with the “Publication of Electoral Opinion Polls Act” of 2012. Taking these two “outsider” polls, they violate the Act in several respects.
Such “confusion” may stem from the fact that Odinga and Ruto have largely exchanged their political “clothing”.
To begin with, the media release from the Centre for African Progress included no physical address (although their company profile can be located via the Bloomberg News website). It also claims that the survey “is funded by an endowment fund that it shares with other similar entities around the world”, but the actual identity of this “fund” is not revealed. (The document also claims to be “the only researcher that correctly predicted Kenya’s polls in 2017”, which is false; Ipsos’ final 2017 survey showed Uhuru-Ruto defeated Odinga-Musyoka 52 per cent to 48 per cent.)
As for Intel Research Solutions poll, the most obvious violations are the failure to identify “the sponsor of the poll” and its address, as well as the address of the organization itself although it does have a website. (According to one journalist who attended the public release of the survey at the Panafric Hotel, when asked who paid for it, IRS’s CEO, Karen Mwangi, said that it was funded by “EU donors”, but a source at the EU subsequently assured me that the EU “had nothing to do with it.”) Further, the release is silent on “the education level of the participants” even though it is still unclear why the Act requires declaration of this specific demographic variable and not others. In addition, and while not a violation per se, a closer look at the distribution of the poll’s massive sample reveals some strange anomalies. For example, a fairly populous county such as Embu with over 309,000 registered voters had a sample of 608, while two other counties with far fewer registered voters—Taita Taveta and Isiolo, with around 156,000 and 75,000, respectively—had far more respondents (679 and 682, respectively). The heavily populated Kiambu with nearly 1.3 million voters also had only 605 respondents. Such mismatches would inevitably lead to incomparable results in terms of the margins of errors for such counties.
Whether anyone would bother to seek “legal redress” for such violations seems very unlikely, but they could (or rather, should) be included in any media coverage—having given the Management of those firms every opportunity to explain these failings, of course.
Achieving 50 per cent + 1: Misinterpreting the undecideds
One common error in media reporting of recent survey results relates to the impact of the “undecideds”. As shown above, none of the mainstream pollsters (TIFA, Radio Africa, Infotrak and “out of nowhere” —after more than three years of silence—Ipsos), show any candidate with more than the 50 per cent + 1 votes needed for a clear win on 9 August, the most recent total average (of the four firms shown above) being 48 per cent. The most visible error is to (evidently) assume that the remaining percentages for “undecided” together with “no response” won’t “vanish” when ballots are actually cast, since these two options will appear nowhere on the presidential “menu”.
For example, reporting the results of Radio Africa’s final pre-election survey in the Star at the start of last week, Bosco Marita stated that with Ruto at 45 per cent and Raila at 44 per cent, “neither of two presidential frontrunners would win on the first round on August 9”. Likewise, Infotrak’s CEO, Angela Ambitho, added some confusion when she offered that “If most of the undecided vote for Raila he would win on the first round, but realistically I do not think they will all move to one candidate.”
Yet a closer look at the Radio Africa poll reveals that such a first round win by either of the top two candidates is actually very likely. This involves first removing 270 respondents who claimed to be “undecided” from the calculation. When their numbers are then re-calculated, Ruto has 49.5 per cent, Raila has 48.4 per cent, and Wajackoyah has 2.2 per cent, and with a margin of error of +/-1.9 per cent (slightly higher than the +/-1.8 per cent reported by the Star, since the calculation is based on a smaller number of respondents who named a candidate: 2,730 rather than the total sample of 3,000). Moreover, it seems highly unlikely that among all the “undecideds” who eventually do vote, their votes would be split equally between Ruto and Raila, thus giving one or the other of them the margin of victory. As noted above, the DP, at least according to the polls, has been winning over more those who had initially been within the category.
Some voters will seek to “punish” incumbents as a way of expressing their unhappiness with the status quo.
Indeed, among the final polls released from the four (mainstream) polling agencies, only TIFA offered such a calculation (i.e., with the “undecideds” removed). It showed that while without removing the 7 per cent of those “undecided” and those who did not respond to the “no responses”, Raila, despite enjoying an 8-point lead, fails to get over the 50 per cent + 1 hurdle (48 per cent to 41 per cent); when that 7 per cent is removed, he does easily, with 53 per cent as against Ruto’s 45 per cent.
But as explained above, any effort to explore these “undecided” respondents is to question the “honesty” of at least some of them. Further, and where I do agree with Angela Ambitho, is that among those who do turn out to vote, they are unlikely to all move in the same direction, even if some of them who do may indeed deliberately spoil their presidential ballots or even leave them blank (being insufficiently enamoured of any of the four candidates and their running mates), thereby raising the percentage totals among those who do vote by reducing the “denominator” against which such calculations are made. (And recall here one of the 2013 Supreme Court judgments that “invalid votes” must be excluded from the total number used for such calculations.)
Media coverage and the act
Another aspect of polling during this electoral cycle has been the impact of the Publication of Electoral Opinion Polls Act (2012). As noted above, several firms have released survey results during this twelve-month period prior to an election during which the Act applies, apparently without any criminal repercussions. (Conviction under the Act invites punishment of about to Sh1 million and one year in prison.) The most relevant part of the Act for this discussion is the ban it imposes on the publication of any election-related results within the five-day period preceding an election, meaning that the last day for doing so was Wednesday, 3 August, although only TIFA released such results on this final day. While they were covered (with interpretation) in an article in the Standard (which included several of TIFA’s graphics) and were included in midday and evening news bulletins by (at least) KBC and NTV, no stories about them appeared in the Nation or the Star on Thursday. While it is unclear whether this was due to an interpretation (correct or otherwise) that the Act prohibits the reporting of such polls within the proscribed five-day period even if they had been released prior to it, a source at Citizen-TV informed me that, based on “legal advice”, such content could not be included in their Sunday evening news “State of the Race” segment which hitherto had provided details of all polls released in the preceding week. Presumably, such caution is based on an assumption that at least some viewers (however few in number) would be learning of such results for the first time, even if they had been in the public domain earlier. On other hand, political editors at NTV and the Standard indicated that their “legal advice” was of a contrary nature, evidently based on an interpretation that the law applies only to a poll’s initial release. Whether such differences of interpretation will eventually invite an effort to clarify the Act remains to be seen.
The importance of the election by elective office
Another finding relevant to voters’ choice as they enter their polling stations was obtained in TIFA’s final (CATI) pre-election survey. The question (apparently never previously included in any Kenyan survey) asked respondents to identify, among the six positions to be filled on 9 August, which one they consider the “most” and “least” important. Given the media’s overwhelming attention on the presidential race, it came as a surprise that more respondents did not attribute more importance to it. As shown, it is in a statistical tie with the position of governor, and nearly twice as many respondents ranked their vote for MCA as “most important” as did those who indicated the same for their MP.
While such ratings may support the argument that devolution has at least partly diluted the “do or die” competition for the presidency, they certainly raise the question as to how much these non-presidential races affect the motivation to vote altogether.
A Pre-post-mortem on the polls: Shoddy sampling, late-hour “bandwagons”, and other factors
One question asked increasingly as election day approaches is: “How will you pollsters defend yourselves if the actual (presidential) results prove you wrong?”
To begin with, it is important to correct Steve Otieno writing in last Wednesday’s Nation, when he began by asserting “The pollsters largely got it wrong in the past two elections when they showed ODM leader Raila Odinga in the lead only for President Kenyatta to be declared winner.” In the final pre-election survey released by Ipsos (where I was then working), Kenyatta led Odinga 52 per cent to 48 per cent. So even if the official result was 54 per cent to 45 per cent, the Ipsos results were certainly not “all wrong”. He also states that “a similar event happened in 2013”, when “in none of the polls did Mr Kenyatta’s popularity cross Mr Odinga’s even though both of them never passed the 50 per cent hallmark”. In fact, the final Ipsos poll had UhuRuto at 44.8 per cent and Odinga/Kalonzo at 44.4 per cent. True, it showed that neither could reach 50 per cent, but it should be recalled that some 800 pages of ODM’s evidence meant to accompany its Supreme Court petition-challenge were excluded due to “late filing”. Moreover, a national exit poll (implemented by a group of Kenyan and expatriate academics and published the following year) confirmed the findings of Ipsos: that neither had reached 50 per cent. So perhaps Otieno should do more fact-checking before trashing pollsters.
Whatever the case, can pollsters sometimes “get it wrong”? Of course, such an argument would first have to precisely define what “wrong” means. Specifically, is it mainly a question of who is announced as the winner, or how great the margin between the polls and the official numbers (assuming they are credible) is regardless of who wins?
In a TV discussion about the accuracy of such polls last week, David Ndii argued that a key weakness in the work of most Kenyan pollsters is sampling, even if it is based on IEBC’s distribution of registered voters, since it does not sufficiently match (the eventual) voter turnout, which is certain to vary across the country. This is a valid argument, although he would have to agree that given the absence of a (main) Kikuyu presidential candidate, and the presence of members of this community as running mates, previous turnout variations may not be a very precise guide for this election. So just which turnout projections should be used?
Whether anyone would bother to seek “legal redress” for such violations seems very unlikely.
At the same time, it is strange that Ndii appeared so uncomfortable with these polls showing Odinga leading, since he also claimed that “our internal polls show the opposite”. Unless he, too, believes in this “bandwagon” effect, it seems such “incorrect” polls would help Kenya Kwanza win by giving Azimio a false sense of confidence which would likely lead to reduced voter turnout among Azimio supporters, either because of “laziness” in mobilizing pro-Azimio voters on 9 August, or insufficient motivation among such voters due to overconfidence (both of which determined the result in the Msambweni by-election, as recalled below).
(One frustrating aspect of the entire discussion was that at no point was any attention given to the challenges the media sometimes faces in interpreting survey results, and even deciding whether to publish/broadcast them, as indicated above. Nor was the question raised at to the contribution, or otherwise, of these polls to Kenya’s democracy, whether they be internal/confidential ones conducted and analysed by campaign teams or those made public so as to allow voters to make more realistic choices, or be influenced by them.)
Another potential factor—that everyone says they hope will not occur—is “flawed” results, although in order for it to be taken into account in explaining a significant variance from the polls, it would be necessary to verify that such “rigging” did indeed take place and at what magnitude—not an easy task.
Still another would be any significant movement in voting intentions between the final polls and election day. While it is generally agreed that nearly all voters have usually made up their minds by this time, last-minute “waves” can occur, nonetheless. For example, Peter Kagwanja, speaking on another TV panel show a few days ago, said that Odinga’s “numbers” in Mt. Kenya will increase dramatically in the final days, since “what matters most to our people, especially in this situation where we don’t have our own presidential candidate, is to be with the winner so that we will be as close as possible to Government”. (This also constitutes an interesting hypothesis that once elected, whoever controls the Executive branch will pay more attention to the needs of those areas/communities that voted for the winning pair than to those of other parts of the country.) In this regard, he noted that nearly all recent (credible) polls showed Odinga-Karua opening up a greater gap between them and Ruto-Gachagua, so that this “bandwagon” effect is bound to materialize, thus making even such final polls look inaccurate.
One recent survey finding by TIFA (in its late July household survey, and published exclusively in the Standard) is that a significant proportion of respondents (13 per cent, as noted, all claiming to be registered voters) reported that they had changed their voting intentions since the start of the year. The following chart shows how these respondents changed.
Taking just those who moved from being “undecided” to supporting any candidate together with those who changed from one candidate to another (but excluding those who changed from supporting any candidate to being “undecided” now) —who comprise two-thirds of all those who reported some voting intention changes—just a handful (4 per cent) said they did so due to opinion polls, as compared with nearly half who cited policy-promises/manifesto content and another quarter who mentioned either “influence of other people” or “choice of running mate”:
True, among this “handful”, near twice as many say they now intend to vote for Odinga than for Ruto, but this is hardly a sufficient number to talk about any significant “bandwagon” effect, at least from polls.
Yet a closer look at the Radio Africa poll reveals that such a first round win by either of the top two candidates is actually very likely.
Still, the belief in this effect is clearly very widespread. Ever since the Deputy President was overtaken in most of the polls by the former Prime Minister, he and his associates have bashed them, often by alleging they were part of the “System’s” efforts to ensure an Odinga win. For his part, Senator Kipchumba Murkomen asserted in a recent TV discussion that “Odinga was leading in an Ipsos poll just a few months before the 2017 election, yet he lost”, although a review of the record reveals that at no time before that election did any Ipsos survey report any such thing. Whether Murkomen’s statement was deliberate or a consequence of a faulty memory is unclear, but the point is that polls—especially those from the more “established” firms that show one’s preferred candidate trailing, invariably cause discomfort. On the opposite side, Karanja Kibicho, CS for Internal Security, was shown in a 2 June news broadcast saying that Azimio had the support of 60 per cent of Kenyans, according to an unnamed survey, although Ruto was then shown refuting this, claiming that he still has personal access to such National Intelligence data, which showed that he was leading “by 8 per cent”. He also dismissed other unnamed polls not showing him in the lead as falsified.
(It is also said that Kenya Kwanza/UDA had been conducting its own surveys, under the “expert guidance” of the former Energy CS, Davis Chirchir. However, some time ago, a UDA candidate approached TIFA to ask for a quote for a survey in his constituency, although he subsequently reported that he was told “not engage any local firms”, but was not given any alternative.)
At the same time, I received an e-mail message from a UDA stalwart a few weeks ago asking for my opinion about survey results that had just been released by one of these “non-mainstream” survey firms showing the Deputy President with a comfortable lead, as she wanted to know whether I consider their work reliable. When I told her I did not, and gave my reasons, she replied,
“I suspected as much; it seemed that they are just a random pollster trying to make a name by saying Ruto will win. A bit opportunistic. And none of the main media houses carried it, so I guess it flunked. These are the ones who need to be called out.”
At least in this case then, and whatever politicians say in public, there is clearly a sober need to know which poll results can be trusted, at least in terms of not being driven by any partisan or personal agenda.
However influential (or otherwise) such a “bandwagon” effect, some have argued that these polls have substantially underestimated Ruto’s eventual numbers, given the current president’s overt opposition to his ambitions. That is, they say that a significant number of survey respondents will be too “shy” to reveal their intentions to vote for the de facto “opposition”, by saying either that they are “undecided”, or even that they will vote for Odinga. This argument is somewhat similar to that offered by many analysts in the US as to why Donald Trump’s numbers were higher than what he received from pollsters in both of the last two elections, but especially in 2016. But in this case, it seems the main weakness was that too few of his voters were interviewed in the first place, since Trump himself had frequently trashed them, as well as the motives of those conducting them. In addition, post-2016 research revealed that a significant proportion of women who were interviewed live (in phone calls) were reluctant to reveal their pro-Trump intentions due to his history of sexual molestation and rhetoric, whereas in automated (non-live) survey calls, more women with the same demographic profile named Trump as the recipient of their forthcoming votes. Since all polls in Kenya are conducted “live” by phone or in person (aside from a few SMS polls that have their own problems), whether this kind of reluctance to reveal one’s choice (generally known as “the Bradley effect”) will be evident when the official results can be compared with those of the final polls remains to be seen, but a surprise Ruto win—perhaps at least in part based on a further move of a majority of the “undecideds” in his direction (and shown above)—would certainly invite scrutiny of this nature.
The Proof of the pudding: Voter turnout, last minute surprises, and election integrity
However, the most widely recognized and probable cause for such contrasts between final polls and the official results is voter turnout. As suggested above, making any such predictions based on past elections is challenging. However, it seems that turnout in Kenya is based on a combination of individual motivation and community (and/or political party/campaign team) mobilization. The latter factor raises the question as to which side will be more successful in doing so. Here, as discussed in my third piece for The Elephant, two surveys conducted in Msambweni before the December 2020 parliamentary by-election there (by Radio Africa and TIFA) both found the ODM candidate, Omar Boga, leading the independent (but UDA-backed) candidate, Feisal Bader, by around 20 per cent, yet he lost by nearly that much. With only 40 per cent turnout, it was clear that the Bader campaign was far more effective in getting his supporters to the polls. With turnout in Kenya’s general elections usually above 70 per cent (although the exact figure has varied considerably), such a dramatic “turnout advantage” is far less likely, but given the high numbers of registered voters in certain areas—starting, perhaps, around Mt. Kenya with some 5 million out of a national total of 22.1 million—this factor could well compensate for the current Odinga-Ruto margin, especially if combined with some level of reticence on the part of the latter’s supporters to reveal their voting intentions, as noted above.
Any effort to explore these “undecided” respondents is to question the “honesty” of at least some of them.
One additional factor that could “pollute” analyses that compare these final pre-election polls with the official results is the anticipated judgment on Monday—just a day before the election – that will determine whether the case filed by “a group of activists” at the Constitutional and Human Rights Division of the High Court a week ago seeking to prevent Rigathi Gachagua, Johnson Sakaja and twelve others from being sworn in should they win their respective contests (excluding out-going Governor Okoth Obado) on the grounds that they are facing criminal investigation for various forms of corruption, with reference to the integrity provisions of Chapter Six of the Constitution. Assuming the case is allowed to proceed, it is not possible to predict—or perhaps even gauge after the affect—its impact on the presidential outcome, but it seems inevitable that it would have some.
Finally, with specific regard to individual motivation, it might be asked: will Ruto gain more from the distress so many Kenyans have felt about current economic conditions for most of which the Kenya Kwanza “brigade” blame the Uhuru-Raila “handshake”, combined with a determination to “punish” the president for opposing his deputy, without whose assistance as a running-mate he probably could never have ascended to power? Or will Raila gain more from a combination of a “bandwagon” effect, combined with an admiration of his past struggles and suffering for the sake of better “governance”, with such credentials evidently buttressed by the choice of Karua as his would-be deputy?
The actual results of the election cannot themselves answer such questions fully, but together with sufficient post-election research of both a qualitative and quantitative nature—all elements of psephology (the study of elections) —they should provide an understanding of their salience and interaction than we have now, but again, depending upon whether the official results are without any (verifiable) credibility deficit, a scenario that both sides in this contest have vociferously championed.
“How Could William Ruto Contain RVP?”: Echoes of 2007 in Kenya Kwanza Rhetoric
Wherever there is violence, it seems Ruto’s name features. It began with the infamous YK’92. Then came Kiambaa and Ruto’s trip to The Hague. Soon, witnesses started disappearing or being found murdered. Ruto’s name keeps cropping up in other cases.
It came as no surprise to some of us, who have lived through it all before, when William Ruto and his sidekick Rigathi Gachagua began speaking the language of violence and daily inventing scenarios designed to strike fear and anxiety into the general public.
Equally unsurprising was the reported discovery in Rift Valley of leaflets threatening extreme consequences to anyone not voting a certain way. I am sure many people had a sense of déjà vu.
Ruto is heir to the concept of the Nandi Hills Declaration of 1969, which laid claim on behalf of Nandis to all the settlement land in Nandi District and opposed the incursion of non-Kalenjins, especially Kikuyus, into the area.
And it is this attitude that has been at the root of the problems Kenya has had with election violence throughout the past 30 years – all of it with some kind of link to Ruto, all of it divisive and exclusionary, and all of it starting in Rift Valley, from Molo and Burnt Forest in 1992 via Kiambaa in 2007 and now threatening again this year.
So-called ‘ethnic clashes’ began in earnest in 1991, when the push for multi-partyism was gaining ground. Kanu was caught out by a wave of support for the nascent opposition and some of its young followers established a group that was determined to ensure the Independence party was not defeated in the 1992 elections. Youth for Kanu (YK) ’92 was birthed.
Ruto, apparently possessed of the appropriate temperament, quickly became a leading light in an outfit that struck dread into the populace. Well-funded and well-armed, YK’92 was brutal in the terror it inflicted on anyone not toeing the anti-multi-party line. Any such ‘undesirables’ were termed ‘madoadoa’.
The 1998 Akiwumi Commission of Inquiry into the tribal clashes of 1991-2 reported a witness as saying the non-Kalenjin in Rift Valley were threatened with dire consequences if they even talked about multi-partyism. YK’92 operatives were deeply involved in the policing of this.
The report noted that “Paul Kipkemei Murei, a Kalenjin himself, told us that, in or about 1991-11, he heard that the Luo, the Kisii, and the Kikuyu, who were the ‘madoadoa’ because they were perceived to be supporters of multi-partyism or its sympathisers, would be driven away.”
Nearly two decades later, very little had changed. The Waki Commission of Inquiry into the post-election violence of 2007/8 said that several witnesses narrated how the pre-election campaigns in Rift Valley were characterised by tension, with the “Kalenjin saying that, on election-day, they did not want to see ‘madoadoa’.”
In both instances, the threats were curtain-raisers to the extermination of thousands of people and the permanent displacement of hundreds of thousands more. And it is no surprise that a similar scenario is occurring today. For ‘multi-partyism’ in 1992, perhaps substitute ‘Raila Odinga’ in 2022.
In 2007, I was working with Raila Odinga, leader of the Orange Democratic Movement (ODM) and then MP for Kibera, on his autobiography. This meant I was taking nearly verbatim notes at every meeting he attended, including with the ‘Pentagon’, the group he formed with his four closest allies at the time – Joseph Nyagah, Najib Balala, Musalia Mudavadi and William Ruto, with Charity Ngilu also later invited to join.
The Pentagon operated out of offices in a house off Ole Dume Road in Kilimani, Nairobi. It was a buzzing place – leaders and MPs and party officials in and out, meetings and discussions all day every day, young activists planning support activities, IT experts setting up a parallel vote-counting system, diplomats and African leaders visiting, local and international media continually banging at the gate – a million things going on. People were upbeat and confident of an Odinga victory.
That was before December 30, 2007, when, as the chairman of the then Electoral Commission, Samuel Kivuitu, later told his church congregation, “the devil stepped in” and people woke up to find Kibaki had somehow overtaken Raila’s nearly one-million-vote lead in the presidential count and was then secretly sworn in as president under cover of darkness, less than half-an-hour after the fraudulent result had been announced.
Kivuitu later admitted that he had been pressured to announce this result and that he did not really know who had won. Those outraged at the theft took to the streets to vent their displeasure. Violence spread, with the result we all know.
At the Pentagon, a dismayed team sat silently glued to TVs as the ongoing violence was reported by brave news crews. Meeting after meeting was held to discuss how to stop people fighting. Press statement after press statement was drafted at Raila’s direction, condemning the violence and calling for peaceful protest. His colleagues still came in and out, but they came in battered and beaten from visiting the front line of the violence, returning to report to the others what they had heard and seen.
Kivuitu later admitted that he had been pressured to announce this result and that he did not really know who had won. Those outraged at the theft took to the streets to vent their displeasure.
On January 4, French ambassador Elisabeth Barbier arrived to offer her support. She was met by Raila, Ruto and Charity, and the head of the European Union Commission also joined the group. I took notes.
Raila told Barbier that “without fear of contradiction, our stand is unequivocal – we are for dialogue. It is the only way out”. He told her he had confirmed to Gordon Brown (then UK prime minister), Condoleezza Rice (then US secretary of state) and the German and Canadian foreign ministers that he would talk to Kibaki. Charity added, “We need to partner with Kibaki’s people in a true way and we shall see the difference.”
Ruto was apparently not so keen. He had some background he wanted to share. He said, “Rift Valley Province is a very interesting scenario. It has seen the biggest backlash of the outcome. Let me give you a little of the background. For the past five years, the majority have felt this government has worked against them … That is why the people of RVP voted with passion – for him [indicating Raila] and against Kibaki.
“When Raila Odinga became a symbol against Kibaki, there was a lot of passion. Moi could not believe what happened …. The Kibaki group really wanted RVP. He was willing to close his eyes to his differences with Moi. He didn’t succeed.”
The same day, a group from the Pentagon met civil society leaders, who included Muthoni Wanyeki, Mugambi Kiai, Millie Odhiambo and Njeri Kabeberi. The ODM team had Ruto, Charity, Omingo Magara and Oburu Oginga, with Raila joining later. I took notes.
Ruto boastfully repeated to the group what he had told Barbier, with added details: “In RVP, it is an interesting phenomenon. It didn’t start with the announcement of the election. It’s strange. It had built up over time. Kibaki got worked against.
“… Six of us [who have been] elected have cases in court, but there is anger against Kibaki and people said they were going to elect [us] anyway. [The other side thought] you cannot compare William Ruto to Moi. [But] if the people of RVP defied Moi and cornered him [Kibaki], they could not stand in the way of what RVP wanted.
“How could William Ruto contain RVP? [Kibaki’s side thought]. [We went] to Kass FM. It was a phenomenon. They [Kibaki’s side] could not understand [it]. [We] cleared [out] everybody associated with Kibaki. It was more about Kibaki than Raila Odinga.”
These comments by Ruto came two days after the burning of the Kenya Assemblies of God Pentecostal Church in Kiambaa, with 30 women and children, mostly Kikuyus and likely Kibaki supporters, incinerated inside.
Before the fire, rumours of an impending attack on local homes and shops had sent women and children to the church as a place of refuge, while their menfolk remained outside to defend them.
But in an explosion of ethnic violence, hundreds of people, many of them Kalenjin neighbours known to the Kikuyus they now sought to kill, arrived with bows and arrows and sharpened sticks, overwhelming the men trying to protect their families.
When Raila Odinga became a symbol against Kibaki, there was a lot of passion. Moi could not believe what happened …. The Kibaki group really wanted RVP
The mob pelted the church with rocks and then blocked the exits with petrol-soaked mattresses, piled on dried maize leaves from the nearby fields and turned the place into an inferno, pushing back in anyone who tried to escape.
They slashed with machetes the men desperately trying to rescue their families, and chased others into neighbouring fields, where they hacked them to death and chopped them into pieces.
After the church fire had died down, relatives went in to search for their people. None of the blackened corpses, grandmothers and mothers who died holding their children close, was recognisable.
A week later, distraught women were still searching the fields for parts of their husbands, a wife perhaps recognising a dismembered leg by the trousers her husband had been wearing. Police officers had the grim job of picking up the slashed and decomposing human remains and taking them to the mortuary in Eldoret.
The Kiambaa killings led to revenge attacks and no community escaped unscathed, including Kalenjins. Kenyans of different ethnic backgrounds who had lived peaceably side-by-side for generations found themselves at odds with each other. It was the result of their having been incited by people who cared nothing for those forced to kill or be killed, and the incident led directly to Ruto’s being indicted at the International Criminal Court at The Hague.
Against this never-to-be-forgotten background of terrible events, the kind of rhetoric about killing that has been recklessly engaged in by Ruto and Gachagua in the run-up to the 2022 elections is nothing if not chilling. It is, in fact, the very definition of incitement.
As everyone knows, the 2007 election dispute was eventually solved through the formation of a coalition government after the intervention of Kofi Annan. ODM’s first meeting with Annan took place at the Serena Hotel on January 23, 2008. At the end of the meeting, Annan said he would meet Kibaki the following day and hopefully get the two leaders to the table for talks as soon as possible.
In the debriefing afterwards, ideas were exchanged among the ODM team, especially over whether now was the time to call off the mass action ODM had supported. Tinderet MP Henry Kosgey reported that people were still angry, especially because Kibaki, despite the ongoing negotiations, had already taken his place in parliament.
In fact, reported Kosgey, “People are going to Sudan to look for guns.” Ruto was unequivocal. “No choice,” he said. “We have to go for guns.”
Wherever there is violence, it seems Ruto’s name features. It began with the infamous YK’92. Then came Kiambaa and Ruto’s trip to The Hague. Soon, witnesses started disappearing or being found murdered. Ruto’s name keeps cropping up in other cases – I won’t name them. And now here he is again, along with his running-mate, making wild accusations and telling implausible stories that can only have an ill intent.
We cannot afford for this kind of violence to continue. Even leaving aside whatever has gone before, those who cannot accomplish their stated aims without stooping so low as to begin setting the stage once again for inter-ethnic violence and killing clearly have no right, ‘God-given’ or otherwise, to be in the business of national leadership.
Editor’s Note: The views expressed in this article are the author’s own and do not necessarily reflect The Elephant’s editorial stance. The Elephant has offered the Kenya Kwanza campaign an opportunity for response but received none.
Can the IEBC Be Trusted to Deliver a Free and Fair Election?
The IEBC delivered a flawed election marred by irregularities in 2017. As reports emerge of strange goings-on at the commission, what hope is there for a free and fair election in 2022?
It is exactly five years since Chris Msando, the Independent Electoral and Boundaries Commission (IEBC)’s ICT manager, was brutally killed in mysterious circumstances. Msando was murdered after giving media interviews days before the August 2017 election in which he spelt out in detail the various measures he and his ICT team were taking to prevent technical glitches and rigging during the election.
Then there was a lot of hue and cry for justice, but efforts to investigate and prosecute Msando’s murderers have come to naught; they have yet to be found or prosecuted. Chances are they will never be found.
At that time, there were attempts to sully Msando’s reputation and cast doubts about his personal integrity. His body was found alongside that of a girlfriend, whose parents are wondering to this day what warranted the death of their 21-year-old daughter. The killings did, however, send a chilling message.
A history of flawed elections
Kenya has held few elections that have been completely free and fair, and which do not carry the threat of violence. Almost every election in the country has resulted in some form of violence, from the so-called “ethnic clashes” in the Rift Valley in the 1990s to the widespread violence of 2007/2008 that left more than a thousand people dead and hundreds of thousands displaced.
After Mwai Kibaki was hurriedly sworn in as president on the evening of 30 December 2007, the late chairman of the now defunct Electoral Commission of Kenya, Samuel Kivuitu, went on record saying that he did not know who really won that election. According to Ken Flottman, an American lawyer who had the opportunity to observe the 2007 elections first-hand, two things contributed significantly to the violence that followed that election: the ban on live broadcasting ordered by the then Internal Security Minister John Michuki and the government’s decision to pull down the media’s reporting of the results.
Since the events of 2007/2008, Kenyans have lived with the fear of a stolen or rigged election resulting in large-scale violence. This fear has made us wary of scrutinising election-related irregularities too closely. We assume that any dispute regarding irregularities, legal or otherwise, will now be resolved in the Supreme Court that was formed after the 2010 constitution was promulgated. The constitution and the various commissions and bodies that it established, including the Supreme Court, are seen as a bulwark against flawed electoral processes.
But are they? In March 2013 the Supreme Court legitimised what many instinctively believed was an election that was ethically, constitutionally and technically flawed, not least because IEBC officials had the gall to announce sometime around midnight during vote-counting that that they would be shutting down the tallying centre because they needed to sleep, only to announce the election results in the early hours of the next morning.
Perhaps the most disturbing aspect of that election was that the presidential candidates Uhuru Kenyatta and William Ruto had cases to answer at the International Criminal Court. But this did not deter Kenyans from voting for them. On the contrary, a large number of Kenyans decided that the election would be a “referendum against the ICC”. They even remained silent when the IEBC’s BVR kits failed and manual registers appeared at polling stations. A “green register”, which no one had heard of before, also miraculously made an appearance.
The media at that time had been effectively “silenced” for the sake of “peace” and so did not ask hard questions (such as why the commission was shutting down the tallying centre in the middle of vote-counting). Although the Supreme Court ruled that the election was free and fair, scandals that emerged later regarding the procurement of non-functioning BVR kits and the “Chickengate” scandal involving kickbacks given by a British printing firm (whose directors were prosecuted and jailed in the UK) did raise suspicions about whether the IEBC’s former office-bearers were corrupt. Later, the auditing firm KPMG revealed that the voters’ register may have contained the names of as many as one million dead voters.
Most disturbingly, the High Court had earlier ruled that it had no jurisdiction to determine the suitability of candidates vying for the presidency as this was the job of the IEBC. So, Chapter Six of the Constitution on integrity and leadership was essentially swept under the carpet. This allowed all manner of shady characters, including Mike Sonko, who was the Jubilee candidate and who was eventually hounded out of office by his own party, to vie for office.
The auditing firm KPMG revealed that the voters’ register may have contained the names of as many as one million dead voters.
In 2017, the public’s trust in the IEBC’s ability to deliver a free, fair and credible election was tested again. Signs that the 2017 election was not going to be free and fair began emerging even before a single vote was cast. The mysterious murder of Msando a week before the election was an ominous sign that things were not going smoothly and transparently. On election day, many of the critical forms 34A and 34B that were used to tally the vote seemed to be missing or had not been transmitted electronically.
The perception that the IEBC had been compromised or was just plain incompetent was strengthened by the IEBC’s own commissioner Roselyn Akombe, who cited various anomalies in how the commission conducted its business after she fled the country and resigned. Some reports also suggested that local and foreign firms that were contracted to manage the electronic transmission of results had dubious reputations. Then, as now, there were many questions being raised about the IT company hired to provide voting technology and services for the elections, and whether the IEBC was really up to the job.
Pegging hopes on the judiciary
The nullification of the August 2017 presidential election results by a majority on the Supreme Court bench renewed hope that the country could resolve electoral disputes and deliver free and fair elections peacefully through the judiciary. The Supreme Court’s decision stunned Kenyans and the world. The Economist called it “an astonishing decision” while the New York Times noted that “the ruling offered a potent display of judicial independence on a continent where courts come under intense pressure from political leaders”. Western and other nations, whose election observers were quick to declare this election free and fair, were caught with their pants down.
The Supreme Court sent an important message to the country’s citizens – that no one, not even the president, is above the law and the constitution. As the then Chief Justice David Maraga stated, “The greatness of any nation lies in its fidelity to the constitution and to the rule of law.” Countries in Africa and elsewhere that had become accustomed to electoral fraud and violent elections could now look to Kenya for inspiration.
The Supreme Court sent an important message to the country’s citizens – that no one, not even the president, is above the law and the constitution.
But the subsequent 26 October 2017 repeat election could hardly pass the test of being free or fair because only one leading presidential candidate – Uhuru Kenyatta – was running. The opposition leader Raila Odinga had urged his supporters to boycott the election (which they did) because the IEBC had still not resolved many of the issues raised in the Supreme Court ruling. It was essentially a one-horse race that led to the election of Uhuru Kenyatta and his deputy William Ruto. Many international observers said that postponing the election with a view to making the electoral process more credible would have been a wiser option, which is what Ghana did prior to its 2016 election.
Lessons from Ghana
Ghana managed to avert a looming crisis by significantly improving its electoral processes. In the months preceding the 2016 elections, violence broke out during various electoral processes and politicians began using hate speech in their campaigns. However, Ghana managed to have a peaceful election because its Electoral Commission, political leaders and civil society took steps to ensure that the elections were credible.
First, the Electoral Commission took highly visible steps to improve the credibility of the voters’ register by cleaning it up and publishing the list of names online. Second, it made it easier for people to vote; Ghanaians could change their original polling station to one that was near the place where they lived or worked. Third, the National Collation Centre, where the election results were tallied, was made more accessible to the media, civil society and party supporters. Local observers stationed in each of the country’s 275 constituencies could also record the election results. Because the polling station data had become so accessible and transparent, Ghanaians knew the results of the election long before they were announced by the commission. Has the IEBC ensured these processes? The IEBC has allowed the media to set up parallel tallying centres at polling stations but it is not clear whether this is enough to ensure transparency. On the contrary, conflicting figures from the IEBC and the media might ignite tensions in a country where there is already so much mistrust of the electoral body.
Ghana managed to have a peaceful election because its Electoral Commission, political leaders and civil society took steps to ensure that the elections were credible.
Many Kenyans, including Raila Odinga, who has the support of the outgoing Uhuru government, thanks to the famous 2018 “handshake” between him and the president, have stated that they are not convinced that the IEBC can be trusted to conduct a free and fair election this month. A KPMG audit report has revealed weak protections against hacking of the voter database and other lapses and irregularities, including tbe registration of 246,465 dead voters. The audit shows that up to 2 million voters on the 2022 voters’ register may not qualify to vote either because they have invalid IDs or because their details do not match existing records. Nearly 5,000 voters have registered more than once using either their Kenyan IDs or passports. These are astoundingly large numbers that could be subject to manipulation and vote rigging in an electoral contest where there is no overwhelming support for just one candidate. Even more alarming is the revelation that there are 14 “ghost” IEBC officials who are not returning officers but who have the authority to transfer, delete, truncate or update the voters’ register. One of these mysterious officials has super access to the register and can change it at will.
As various reports emerge of strange goings-on at the commission, including the mysterious arrival in the country of Venezuelans carrying IEBC materials, hopes of a free and fair election are fading fast. The loud and boisterous defence of the IEBC and its commissioners by the leading presidential candidate William Ruto despite such anomalies has led many to suspect that perhaps some commissioners are partisan and might already have been compromised, or that Ruto has information that the rest of us don’t. Moreover, IEBC commissioners whose terms ended or who resigned have not been replaced, and the current chairman, Wafula Chebukati, oversaw a flawed 2017 election inundated with irregularities. Can he be trusted to not repeat the mistakes of 2017? There is also the troubling question of why the IEBC has cleared so many candidates of objectionable or dubious backgrounds. As for the technology, it has failed us twice before. Will it fail again?
The 9 August election will likely be another test for the IEBC, the judiciary and Kenya’s democracy.
Politics2 weeks ago
Three Critical Questions Will Determine the Kenyan Election
Politics2 weeks ago
Towards the Home Stretch: Poll Update in an Evolving Electoral Context
Politics2 weeks ago
Invoking Empire in the Face of Collapse: A Dangerous Nostalgia
Op-Eds2 weeks ago
Road to 9/8: Two Sides of the Same Coin?
Op-Eds2 weeks ago
Without Political Reforms Rwandans Will Continue to Seek Refuge Abroad
Op-Eds2 weeks ago
Kenya: Fair Taxation for Economic Justice
Politics6 days ago
Smartmatic: The Election Company and their Role in the Upcoming Elections
Op-Eds2 weeks ago
Back to the Future: Imperialism’s Strange Bedfellows