It isn't news that the Standard isn't a neutral paper. Two instances - their reporting of Balala's remark, and their reporting of the International Crisis Group's report on Kenya - serve to establish that claim. If you followed the Kriegler Commission's hearings, and you should have, you ought not to have followed them via the Standard, because some of the commentary on the hearing was egregiously biased. I'm referring, as you might have guessed, to Sarah Elderkin's pieces on the 17th and 24 August.

I. Elderkin on the 17th August. 

Elderkin's articles were neither of them very good. Start with the 17th August one. She argues that Throup isn't a polls analyst, but a historian. Therefore, he oughtn't have made a presentation on the matter to the Commission. We must assume that she means to say that none may make a submission to the Commission on this matter save poll specialists.

That criterion disqualifies all previous submissions made on the matter by NGOs, ODMers, and associated parties who weren't themselves poll analysts. In particular, it disqualifies Professor Clark Gibson, on whose expertise she hinges her case in her piece of the 24th, because he isn't a poll specialist either, as a quick look at his CV demonstrates. Her argument is self-refuting.

Indeed, Throup is the leading historian of Kenyan politics; he also has a Masters degree in Political Science from the LSE. He's the authority on Kenyan elections, and has quite literally written the book (Multi-Party Politics in Kenya: The Kenyatta & Moi States & the Triumph of the System in the 1992 Election). If that sort of expertise and experience doesn't qualify him to testify on the matter, it is unclear what could.

She next argues that Throup was grossly mistaken in his taking to task the EU observer mission.  He argued that the star EU cases, the Molo and Kieni results, weren't proof of rigging.

(a) Molo first. In previous elections, turnout there was significantly higher than that given in the EU figures; second, if the EU figures were correct, Molo had a significantly lower turnout than neighbouring Kuresoi and Narok North. Throup backed that argument up with stats. Given that there had been serious violence in Kuresoi before the election, and significantly less violence in Molo, and none in Narok North; given that there was record turnout for this election; given historically high turnout in Gikuyu areas; and given prior ethnic polarization, the finding that Molo had a lower turnout than both those constituencies is incredible. He had he better of that argument.

(b) Kieni next. On EU figures, Kibaki had approximately 65% of registered voters in the Presidential election. That is a fundamentally implausible figure. It is wildly out of line with the number total votes cast for Parliamentary candidates in the same constituency, as well as the trend in surrounding constituencies. Further, Kieni is in the President's home district, where had he had had over 90% of the vote in 1992, and approximately the same this time. Everyone admits that election turnout this year was the highest ever. One doesn't get to say this often, but the ECK figures are the only ones consonant with commonsense and history here.

Finally, she argues that "Perhaps his many contradictions were best exemplified in one amazing assertion about his work: 'I've assumed the [electoral] register is perfect, and that is the basis of the statistical analysis. I know it's not.'"

I'm surprised by her surprise. That assumption is an entirely defensible idealization, familiar to all those who were taught physics by the sort of Kenyan teacher who instructs that friction be ignored in answering exam questions.

 More importantly, it's an idealization relied upon by both parties to the dispute. When ODMers argued that the election was rigged, they relied on discrepancies between the number of voters registered and turnout, or argued that turnout was unusually high, or argued that registered voters were not allowed to vote. All those arguments depend essentially on the premiss that the electoral register is sound.

II. Elderkin on the 24th of August.

This was essentially a puff piece for Long poll report. Insofar as Elderkin has an argument here, it goes like this: after (1) a competent electoral commission's counting, and (2) a parallel vote tabulation, exit polls are next best way of reliably determining the outcome of an election.

Now consider the exit poll released by the IED on the night of the 27th December. It showed Kibaki in the lead by 47.4 to 42.7. At the time, and since, ODMers argued that Kenyan exit polls were inherently unreliable because voters were too fearful of the power of the state.

That argument is entirely ridiculous. Dercon demonstrated that likely PNU voters were more worried by violence than likely ODM voters, and that ODMers engaged in more violence and intimidation. PNU agents in Nyanza were murdered on polling day and three AP were killed just before. Far more incidents of ethnic hate speech were recorded in ODM strongholds than elsewhere; indeed, RVP alone accounted for over a third of recorded incidents (cf. KHRC's Violating the Vote).

Now that we have an exit poll with congenial results, ODM's public relations branch - for that is Sarah Elderkin's role - assures us that exit polls are reliable enough to determine that they won. Her argument relies on a premiss previously disavowed by the organization in whose name she speaks has. There's a fundamental problem of plausibility here.

It would be useful to compare the Long and IED exit polls for their underlying method and assumptions. Any satisfactory vindication of either poll will have to explain why the other got it so badly wrong. (And perhaps also why they're both out of line with pre-election polls.)

__________________________________