Briefly: the top end of the Gikuyu political class is an unaccountable oligarchy; that oligarchy has ill-served the interests of Gikuyu outside it; given its unaccountability, it continues to command support only by appeal to naked ethnic identification; that naked ethnic identification puts Gikuyu lives at risk, and is therefore irrational. Gikuyu outside this elite, therefore, should abandon their support for this oligarchy.

The interesting bit, I take it, is the claim about naked ethnic identification:

Now we have that old political class, dangerously entrenched by a sense of ethnic "entitlement". Kikuyus should realise that this does not bode well for the nation, let alone their community. But this attitude is not just the prerogative of powerful politicians, it has also affected the middle class and ordinary Kikuyu.

As the passage shows, Kiai and Muite wish to attribute this variety of ethnic identification -- I'm going to call it ethnocentrism from here on in -- quite widely: it's not easy to see how one could be Gikuyu and non-ethnocentric on their understanding. In passing, it's worth noting that Kiai and Muite argue that the Gikuyu oligarchs have nothing but contempt for non-oligarch Gikuyu. This combination of views puts some pressure on their argument. It's possible that the oligarchs combine serious ethnocentrism and deep contempt for their co-ethnics - but rather psychologically unlikely. Either the oligarchs are ethnocentric or they're not. If they're not, then that clashes with the strong evidence that richer Kenyans are more ethnocentric than poorer ones. If they are not, we're left with the unlikely claim about murderous contempt for their co-ethnics. Odd consequences either way.

Back to the thrust of the argument. The relevant kind of ethnocentrism for Kiai and Muite's purposes is the sort that drives political behaviour. What they call the ethnic entitlement matters politically only if they can show that it's a reliable driver of political behaviour. The strategy is simple: they examine the performance of the Kibaki administration, conclude that it hasn't produced the goods, and infer that its continued support from the extra-oligarchic Gikuyu classes is explained by the obstinacy of Gikuyu ethnocentrism. Continued support for the Kibaki regime runs directly counter to the interests of those who bear a Gikuyu identity; it follows that Gikuyu supporters of the Kibaki regime are irrational. That argument is bad because obviously faulty; obviously faulty because made in defiance of facts both relevant and easy to discover.

Start with some platitudes about political behavior; or, to keep it nicely specific, voting behavior. Assume, reasonably enough, that people are reasonable. Four things, all of which I'm going to take for granted, follow. Usually, voters will prefer a candidate who can credibly promise to improve their lives in the usual ways; and they'll reward those who live up to these promises and punish those who don't. Second. When considering reasons for voting, people will assign naked ethnic identity a low priority. Third. If voters are rational, then, in explaining political behaviour, explanations on which they do come out rational are to be preferred. Finally. Reasonable voters vote have both positive and negative reasons for their vote: crudely, voters vote either in the hope of gaining or keeping a benefit, or avoiding or mitigating a loss.

Muite and Kiai look only at the benefits that could reasonably have been expected of a Kibaki administration, find that none were realised and move smoothly to the conclusion that Gikuyu voters' support for Kibaki can be explained only by their ethnic identification. They look only at the positive reasons for voting Kibaki, conclude that there were none, and so attribute irrationality to his supporters, on the ground that those voters have fallen foul of the requirement of rational political behavior mentioned in the previous paragraph. Concede, for argument's sake, that the Kibaki administration hasn't realised any of the promised benefits. The point, so banal I hesitate to make it, is that that argument leaves the negative reasons unaccounted for. Even if one assumes that the first Kibaki administration didn't produce sufficient benefits to drive a non-ethnically-motivated vote, Muite and Kiai's argument that Gikuyu supporters of Kibaki are irrational ethnocentrists fails, because Muite and Kiai consider only positive reasons to vote for the Kibaki regime. ODM's 41-against-1 pre-election campaign, and its post-election conduct, give sufficient negative reason: the average Gikuyu voter's expected loss from an ODM victory is sufficient to motivate a vote for Kibaki, given that a Kibaki vote was the only plausible way to prevent an ODM victory.

In Tribalism as a Minimax Regret Strategy, Kimenyi and Gutierrez Romero argue persuasively that ethnic voting in Kenya is defensive: a given voter acts to minimize the probability of what he takes to be the worst outcome (Kimenyi & Gutierrez-Romero 2008: 7-8 for explanation). A defensive vote is distinct from a negative vote. Suppose all possible outcomes are gains. Then, the worst outcome is a gain, so even if a voter votes to minimize the chances of his worst outcome, he will not have voted negatively, because he'll have voted for some gain or other. But that's an unlikely situation: the more common case is a prospective loss which the voter will be voting to mitigate or avoid. Kimenyi and Gutierrez Romero's result -- that the voting wasn't negative, in the sense of negative I've given here -- is achieved under the remarkably unrealistic assumption that Gikuyu voters did not expect to be discriminated against under an Odinga government (Kimenyi & Gutierrez-Romero 2008: 7 n5). The assumption is false:  credible threats of violence are a reasonable predictor of future behaviour; PNU voters were likelier to report threats of electoral violence. Their result ought to be strengthened once the more realistic assumption is in place: all else remaining equal, raising the potential loss raises the probability that the voter will act to mitigate it. 

How to fill up the expected losses column? Paul Collier has an interesting, and somewhat counterintuitive, argument: in African elections, violence is a weapon of oppositions; incumbents tend to have the advantage in money and vote-fiddling, so less need than the opposition to resort to beating up voters. Party difference in the potential for violence is your decisive negative reason: the rational Gikuyu voter had good reason to think that there would be anti-Gikuyu violence during election season. Even better, the differential is sufficiently strong to appear in pre-election surveys: Stefan Dercon found that likely PNU voters were more afraid of violence than likely ODM voters, and by a substantial margin (Dercon 2008). Clearly, then, ODM's asymmetry in the threat of force was already generally perceptible well before the violence.

Collier arrived in Kenya in 2007, expecting violence; we didn't disappoint. His research remains unpublished; the findings are reported inWars, Guns and Votes (Collier 2009). Three are immediately relevant: that "Violence against the Kikuyu was a deliberate strategy electoral strategy of Raila Odinga" (Collier 2009: 72); that "Raila Odinga ran a campaign that was tantamount to promising ethnic cleansing" (Collier 2009: 70); and that "...by far the main culprit [in ethnic mobilization] was the opposition leader, Raila Odinga. Recall that the incumbent has the advantage in respect of bribery and miscounting, so the opposition is indeed more likely to resort to the cheaper strategy of playing on ethnic identity" (Collier 2009: 70).

Collier's findings, if sound, would be overkill. Those findings are supported by a lush variety of independent sources.

The obvious case is the RVP. There's compelling evidence of incitement for violence both early and late. Lynch (Courting the Kalenjin, 564) dates the 'virulent' anti-Gikuyu language in the RVP to the immediate post-referendum period. KHRC's Violating the Vote (Table 5.1 on p. 23) has a useful table laying out recorded incidence of hate speech just before polling day. More than a third of all recorded incidents of ethnic hate speech -- more than anywhere else in the country -- occurred in the RVP. 

Immediate pre-election planning for violence is also lavishly attested. The Human Rights Watch report has eyewitness testimony of a pre-election meeting near Turbo at which an ODM campaign official arranged anti-Gikuyu violence; it details similar pre-election meetings around Eldoret at which ODM officials strove quite openly to prepare their community for violence. The International Crisis Group's report corroborates HRW's findings: they also have local politicians in RVP planning anti-Gikuyu violence before the election.

Waki has persuasive evidence of planning too. We're told that a Kalenjin employee of the Eldoret Armoury gave firearms training to youth in Sigowet Forest and in the Kerio Valley (Waki 2008: 71). A Kalenjin eyewitness interviewed by the Waki Commission saw the co-ordination of roadblocks in outlying areas of Eldoret (Waki 2008: 68); another submitted videos to the Commission as evidence of deliberate incitement; a resident of Soy division detailed various meetings in Eldoret beginning on 1 December 2007 and continuing through the violence, as well as the activities of an elder whom he accused of rounding up youth to participate in the violence (Waki 2008: 69-70). The elder, it is safe to assume, is Jackson Kibor; the fact that the Prime Minister chose to be photographed with him at Orange House in March 2008 rather makes Paul Collier's point; making Collier's point is sufficient to establish expectation of a large enough loss to justify a vote for Kibaki.

One might argue that the prospective Kibaki voter couldn't have known all this before the vote, and so it can't have been expected to enter their calculations. That would be to ignore the result from the Dercon survey, the widespread pre-election knowledge of hate speech, and basic historical context: violence in the Rift Valley has been quite deliberately stoked by the Kalenjin political class more or less continuously since 1991. A rational Gikuyu voter living in the RVP would let that influence his decision, even in the absence of clear evidence of preparation for violence this time. But of course, there was good, if non-specific, reason to expect violence this time.

A more interesting objection begins by asking why there's so little variation by class. The reasoning underlying the Kiai piece goes like this (compare the bits about 'speaking with one voice'). Suppose Gikuyu voters are ethnocentric. Then, they'll all vote the same way: any pair of Gikuyu voters, however different in class, are equally likely to vote for a given candidate. A uniform choice among Gikuyu voters demonstrates ethnocentricity. Kibaki had an overwhelming majority of the Gikuyu vote; therefore Gikuyu voting was ethnocentric, since, even if there were negative reasons, they would not have had equal weight across classes.

The richer a Kenyan, the likelier she is to vote on ethnic lines, as Gutierrez Romero and Kimenyi's September 2008 brief reports. And the more educated the Kenyan, the likelier he is to identify with his ethnicity (Bannon et al . 2004).  If ethnocentricity matters, then the strength of one's attachment to one's ethnicity matters. If ethnocentricity were the decisive explanation for voting behaviour, then richer and better Kenyans would be likelier to vote their ethnicity than their poorer co-ethnics. The research was done in Kenya; I'm happy to apply that conclusion to Gikuyu voters too. If the strength of the Gikuyu voter's attachment to his ethnicity were the decisive explanation for his voting behaviour, then richer and better-educated Gikuyu would be likelier to vote their ethnicity, than their poorer coethnics. It turns out that what Kiai and Muite thought was evidence for their claim -- the uniformity of the pro-Kibaki vote -- is actually evidence against. Their move from uniformity to ethnocentricity is unsafe.

That still isn't a non-ethnocentric explanation of the Kibaki vote among poorer Gikuyu voters. Catherine Boone's intriguing paper on the politicization of land rights (in much of Africa) might do the trick: she argues, if I understand her aright, that where land rights are granted directly by the state, their stability remains directly dependent on it. African states have grown (comparatively) weaker and more decentralised the longer they've been exposed to liberal economics. The weakening of the state makes it sensible for regional barons to consolidate their local control; land is an easy way to do so; restricting access to it by ethnicity is, from their point of view, a sensible strategy. In Kenya, the tendency tends to spread from land to other forms of property: we had Luo tenants refusing to pay rent to Gikuyu landlords on the ground that soon they would be in State House. Voters have property in their right to residence in a particular place; a credible threat of eviction from it, as in the RVP, should suffice to generate a large enough potential loss to justify a Kibaki vote even for those who have no land. The importance of central government for property rights, even very far from the heart of the state, may suffice to explain the willingness of the poor to vote with Kibaki, provided a suitably broad view of property is taken.

Suppose all this is sound. Then Kiai and Muite's explanation of Gikuyu voting behaviour by ethnocentrism is fatally flawed. Remember that, for the sake of argument, I conceded that the Kibaki regime had not produced, and could not be expected to produce, any benefit for its voters. Even with that deeply unrealistic premiss in place, ODM's campaign generated, for the average Gikuyu voter, an expected loss large enough justify a Kibaki vote.

What of the future? Begin by considering the aftermath in the RVP (Ashforth 2008: 16, 18-19):

None of the Kalenjins we spoke with in the Rift Valley evinced any sympathy for his or her missing neighbors. During my previous visit, in the northern summer of 2007, relations between the two groups in the district had not seemed particularly troubled. Indeed, my Kalenjin in-laws were not alone in having close friends, lovers, and co-parents from the Kikuyu community. On my visit in March, however, I found the ruined houses of people I had thought of as family friends along with those of many other neighbors. The consensus after the 2007 election was that Kikuyu did not belong in this part of the world. "This is Kalenjin territory," was a steady refrain. One woman  insisted to us that the Kikuyu actually belonged in Congo, whence they had been imported by white farmers in the days of colonialism.

...As for the conflicts that involved ethnic cleansing in the Rift Valley, however, the prospects for reconciliation are slim, and the implications for the future of the country are grim. The refugee camps established during the fighting, housing some three hundred thousand people, are now being dismantled, forcing refugees either to return home or to seek shelter elsewhere. Few are returning to the contested zones of the Rift Valley.

In some districts of the Rift Valley virtually no Kikuyu residents are left. The ethnic cleansing of Kikuyus from Kalenjin territory is seen almost without exception, by Kalenjin, as a good thing. In conversations with Kalenjin in the Nakuru district in March 2008, I met only one person who expressed discontent with the course of events, and she was a local politician in the town of Mogotio who had been forced into hiding because of her unpopular views. Even those not actively engaged in perpetrating or supporting the violence have little sympathy for the people whose relatives were killed and whose houses were burned and properties ransacked. The displaced will not be welcomed home.

The post-election period confirmed expectations of violence. Those voters also know what Ashforth points out: that they're unlikely to regain their property; that they're unlikely to see the perpetrators of the violence held accountable; and that a repeat of the violence is quite likely, given wholly credible reports of militia rearming in the RVP. The expected losses column will continue to look very grim.

________________________