Nearly a year ago I com­mented on a Slate arti­cle that won­dered why the Inter­na­tional Repub­li­can Insti­tute would with­hold the results of an exit poll that sug­gested Raila Odinga had won the Kenyan pres­i­den­tial elec­tion. My guess:

The answer to Halperin’s ques­tion might there­fore be depress­ingly simple: the IRI won’t release their polling data because they don’t want the wrong guy to win.

Now it looks like my sup­po­si­tion was right. Here are the key sec­tions from Karen Rothmeyer’s “Meddling in Kenya” in The Nation:

The poll…showed that Odinga had bested Kibaki by an eight-​point margin. This was in con­trast to the offi­cial fig­ures released later, amid chaos and alle­ga­tions of rig­ging, that showed Kibaki win­ning by a two-​point margin. Plans called for releas­ing the poll, the only one of its kind…the day after the elec­tion. But instead, IRI’s top Washington-​based offi­cials, claim­ing they had seri­ous doubts about the poll’s valid­ity, refused to make the results public. [IRI coun­try direc­tor Ken] Flottman says he kept press­ing for an answer as to why. “I was even­tu­ally told that it wasn’t in the best inter­est of IRI,” he says.

…

Why did US offi­cials not take note of their own poll? Why did they in fact not just ignore it but sup­press it? “The results were unpalatable,” says one high-​level, non-​American inter­na­tional offi­cial who declines to be iden­ti­fied because of the official’s con­tin­ued work­ing rela­tion­ship with the US government.

See below the fold for the rest of the story.

More from Rothmeyer:

While I had only an outsider’s view of the elec­tion process, I, too, saw things that made me wonder. On Decem­ber 17, little more than a week before the elec­tion, [U.S. Ambas­sador Michael] Ran­neberger gave a curi­ous inter­view to one of the major dailies. Despite having described cor­rup­tion in Kenya as “like a cloud hang­ing over the nation” a few months after his arrival in 2006, and having told a uni­ver­sity audi­ence in mid-2007 that “not enough has been done to bring to jus­tice those respon­si­ble for corruption,” he now claimed that cor­rup­tion wasn’t a big deal after all. “I always point out that we have lots of cor­rup­tion even in the US,” Ran­neberger said breezily, men­tion­ing Enron as an example.

Such inci­dents, in ret­ro­spect, were a por­tent of what was to happen to the exit poll IRI con­ducted on elec­tion day with the aid of a local polling firm, Strate­gic Public Rela­tions and Research, assisted by polling experts from the Uni­ver­sity of Cal­i­for­nia, San Diego. The poll–which asked voters whom they’d voted for–showed that Odinga had bested Kibaki by an eight-​point margin. This was in con­trast to the offi­cial fig­ures released later, amid chaos and alle­ga­tions of rig­ging, that showed Kibaki win­ning by a two-​point margin. Plans called for releas­ing the poll, the only one of its kind (a second exit poll by a dif­fer­ent orga­ni­za­tion was begun but not com­pleted), the day after the elec­tion. But instead, IRI’s top Washington-​based offi­cials, claim­ing they had seri­ous doubts about the poll’s valid­ity, refused to make the results public. [IRI coun­try direc­tor Ken] Flottman says he kept press­ing for an answer as to why. “I was even­tu­ally told that it wasn’t in the best inter­est of IRI,” he says.

Ran­neberger was given the poll results on the evening of elec­tion day, Decem­ber 27. This was three days before the offi­cial announce­ment that Kibaki had won touched off weeks of riot­ing in which more than 1,000 Kenyans died and as many as 350,000 were made home­less. Nonethe­less, Ran­neberger went on to tell the Wash­ing­ton Post on Decem­ber 31 that “the US would accept” the announce­ment that Kibaki had won, and the State Depart­ment con­grat­u­lated Kibaki on his win–a posi­tion that it later retracted after the Euro­pean Union raised con­cerns about elec­tion rigging.

Why did US offi­cials not take note of their own poll? Why did they in fact not just ignore it but sup­press it? “The results were unpalatable,” says one high-​level, non-​American inter­na­tional offi­cial who declines to be iden­ti­fied because of the official’s con­tin­ued work­ing rela­tion­ship with the US gov­ern­ment. The offi­cial adds that out­side experts involved in the elec­tion process were trying to encour­age “transparency and open­ness. Hiding infor­ma­tion doesn’t con­form to those principles.”

An even bigger issue is whether the exit poll’s release might have altered the out­come of the race. Njeri Kabeberi, head of the Center for Mul­ti­party Democ­racy, a non­par­ti­san group that focuses on party build­ing, says that if the results had been made public quickly “it would have made a big, big difference” by deter­ring vote tam­per­ing. That opin­ion is shared by many, but not all, out­siders. Tom Wolf, a US polling expert who lives in Kenya, says, “It would have fueled [Odinga's party's] claims to legit­i­macy, but Kibaki wouldn’t have given up with­out a fight.” Wolf, who con­sults for a rival firm to Strate­gic, also says he has some reser­va­tions about Strategic’s polling methods.

Flottman, for his part, con­tin­ued to press for the poll’s release, even after he left the coun­try at the expi­ra­tion of his con­tract. “The fact is, it’s the only inde­pen­dent source of data” on the elec­tion, he says. And in the end, after the Uni­ver­sity of Cal­i­for­nia polling experts made a public pre­sen­ta­tion on the data and method­ol­ogy in Wash­ing­ton, IRI backed down on its ear­lier posi­tion. After what it described as an “extensive analysis” that reduced Odinga’s margin from eight to six points–still sig­nif­i­cantly larger than the poll’s margin of error of less than two points–it made the poll data public in August.

By then, it was way too late to make any dif­fer­ence: pushed by the inter­na­tional com­mu­nity, Odinga had dropped his insis­tence that he had won, and he and Kibaki had agreed to form a coali­tion gov­ern­ment, with Odinga installed in the newly cre­ated and, as it quickly became appar­ent, rel­a­tively pow­er­less job of prime minister.