“A Chaotic Kenya Vote and a Secret US Exit Poll” New York Times, Jan. 31, 2009

Kenya’s 2007 Election and the US Exit Poll

The Kenya program I led for IRI in 2007-2008 included a polling program and an Election Observation Mission for the 2007 Kenyan election and coincided with the “Post Election Violence” in Kenya.  The Exit Poll for the election was the subject of investigative reporting by Karen Rothmyer in the Nation magazine of December 10, 2008 and on the front page of the New York Times by Mike McIntire and Jeffrey Gettleman on January 30, 2009, as well as earlier “real time” reporting by Shashank Bengali for the McClatchy papers and by Alex Halperin for Slate magazine.

Unfortunately, I had something of a difference of opinion with at least part of the leadership of IRI’s Washington office in the aftermath of the Kenyan election over what to say and disclose publicly about the exit poll which was funded by USAID and the University of California, San Diego (“UCSD”).  After a contractual six month embargo on discussing the results imposed by IRI, Dr. Clark Gibson and James Long of UCSD, the researchers who were the primary authors of the poll and its methodology, released it to the public at the Center for Strategic and International Studies (“CSIS”) in Washington on July 8, 2008 with media coverage in the US and Kenya.   The poll showed  a roughly six point victory in the presidential vote for the leading opposition candidate Raila Odinga whose ODM party had gained a large margin over the incumbent president Mwai Kibaki’s PNU coalition in parliamentary seats, but who was announced by Kenya’s electoral commission on December 30, 2007 as losing to Kibaki by roughly 200,000 votes amid great controversy and ultimately significant violence involving government paramilitary forces, demonstrators, and in some areas armed militias and gangs.

After release of the poll in July 2008 USAID ran a story in its in-house magazine reflecting that the USAID-funded exit poll indicated that the wrong candidate had been named winner by the Kenyan electoral commission.  Subsequently IRI itself released the poll results on its website in August the day before Gibson and Long were to testify about the poll before the Kreigler Commission in Nairobi which had been appointed to investigate the conduct of the election under the February 28 settlement agreement brokered between Kibaki and Odinga by Kofi Annan.

Strategic Public Relations and Research, a Kenyan polling firm, conducted the exit poll in the field under contract to IRI with direct training and supervision by UCSD. At USAID’s request, Strategic gathered preliminary results for the presidential race through a system of cell phone verification with its field supervisors. Strategic reported these results on election day directly to USAID, which reported them to the US Ambassador.  The Daily Nation, Kenya’s leading circulation newspaper, had a story on the exit poll plans, including an interview with Strategic.  It was widely understood in the Kenyan media that IRI was doing another exit poll, having conducted them in the 2002 general election and the 2005 national referendum.  Nonetheless, IRI said nothing at all about the exit poll when it issued a “preliminary statement” on the election on December 28, or its statement on post-election violence on January 2, 2008.  Internally I dissented from this “silent approach” which was asserted to me to be based on concerns about the violence, rather than on any concerns about the poll itself.

Going forward, as instability and uncertainty continued in Kenya, IRI in Washington issued on its website a statement on January 15, 2008 asserting that the poll was not being released at that time due to concerns that it was “likely invalid”. On February 7, 2008 Senator Feingold in hearings of his Africa Subcommittee of the Senate Foreign Relations Committee insisted that Asst. Sec. of State Jendayi Frazer and Asst. Admin. Kathleen Almquist of USAID submit an explanation as to why the poll had not been released.  That evening IRI posted a statement concluding that the poll was invalid and essentially denigrating the work of those who had conducted it.  When IRI issued its final election observation report in July, which as much as concludes that the election was stolen, nothing was said about the exit polling.  This is where IRI left it until August 14.

Because I felt IRI’s February 7, 2008 disparagement of the poll was substantially unjustified by the facts as I knew them, when I was contacted shortly thereafter by a reporter for the Nairobi Star newspaper who had the leaked preliminary results of the exit poll, although I wouldn’t comment on the results, I declined to endorse or adopt the statement from Washington as reflecting my own opinion.  When the paper published its story on the front page under a headline that Odinga rather than Kibaki had won the election, with extensive discussion of the poll, it reported that the statement from Washington “did not reflect” my personal opinion.  My relationship with Washington was strained after that.

I turned the original survey questionnaires over to my successor as East Africa Director in Nairobi on May 4, 2008, more than four months after the election.  No hint was given by IRI Washington that there was any flexibility on reconsidering or retracting the February 7 conclusion that the poll was invalid during those months.  The poll was conducted solely on election day and its validity was determined by the effort done on that day.  All the original data, more than 5000 questionnaires, was back at Strategic in Nairobi by January 15 and all the voluminous data entry covering all 2000+ parliamentary candidates from dozens of parties as well as many more municipal candidates and issue and demographic questions was coded by Strategic by January 30, in spite of the ongoing violence.

This is history now.  There is not and has not been since August 2008, if there ever was, any disagreement between myself and IRI about the basic validity of the exit poll and its results.

In reacting to the reporting in the New York Times early in 2009, IRI has published on its website some material that is apparently intended to be a defense of their conduct regarding the poll, as well as material in other fora.  Part of this is an attempt to suggest that I agreed with their decision to withhold the poll based on a quote picked out of context from an e-mail chain that was in fact available in full in IRI and to the New York Times all along, along with some specific factual misrepresentations including an initial statement that I did not advocate release of the poll until February 8 after IRI’s February 7 statement. I am confident that any fair-minded person with access to the record would find the Times reporting rather than IRI’s after-the-fact attack on the Times to reflect the actual record.   Ultimately, it is clear that IRI’s February 7, 2008 publication of a statement that the poll was “invalid” was wrong.   To the extent that you, dear reader, wish to make further evaluation of my personal credibility in this matter, I will be happy to have you contact me for documents and records and/or further questions at africommons [at] gmail.com.

While there may be people who dislike the fact that I was not willing to actively support the Washington “party line” during the interim between the polling on December 27, 2007 and IRI’s release of the results on August 14, 2008, or my willingness to answer directly and truthfully those questions put to me about the subject after completion of my service with IRI, it is my hope that the organization will obtain some institutional learning from the experience that will make it more effectively helpful to the integrity of elections going forward.   I worked for IRI because the job was important, my previous experiences were positive, I felt they had done good work in some difficult situations in the past so far as I knew and I respected people with whom I was acquainted there.  While I think they made a consequential mistake in Kenya on my watch, I certainly expect that this decision was driven by government rather than IRI initiative.  I do wish them success in effectively supporting democrats in the future.

It is the statutory foreign policy of the United States to engage in “the promotion of good governance through combating corruption and improving transparency and accountability.” We will be better promoting those things when we practice them.

On the Election Observation, please see my post “Election Observation: Diplomacy or Assistance?”.

Update–September 3, 2011: Please note the additional evidence about the Ambassador’s approach to the exit poll reflected in my series of posts from State Department cables regarding the exit poll released to me under the Freedom of Information Act in August 2011. In particular Part One and the summary at Part Three. March 2013–And especially Part 12.

Related articles

• Exit Polls and Orange Revolutions; Eastern Europe and East Africa (africommons.com)