NAIROBI, Kenya — Six months after a deeply flawed election triggered a wave of ethnic killings in Kenya, a U.S. government-funded exit poll finds that the wrong candidate was declared the winner.
President Mwai Kibaki, whom official results credited with a 2-point margin of victory in the December vote, finished nearly 6 points behind in the exit poll, which was released Tuesday by researchers from the University of California, San Diego.
Opposition leader Raila Odinga scored "a clear win outside the margin of error" according to surveys of voters as they left polling places on Election Day, the poll's author said.
The incumbent Kibaki was sworn in for a second term despite major irregularities in vote-counting, sparking tribal attacks that killed more than 1,000 people, the worst violence in this East African nation in nearly two decades.
The exit poll, whose existence McClatchy first reported in January, was financed by the Washington-based International Republican Institute, a nonpartisan democracy-building organization, with a grant from the U.S. Agency for International Development, the foreign-aid arm of the State Department.
Amid a worldwide furor over the election results, the institute decided not to release the poll, citing concerns about its validity. But the poll's authors and the former head of the institute's program in Kenya stand by the research, which the authors presented Tuesday in Washington at the Center for Strategic and International Studies, an independent research center.
In the exit poll, Odinga had 46.07 percent of the vote to 40.17 percent for Kibaki, a difference well outside the poll's margin of error of 1.32 percentage points. The official results gave Kibaki 46.42 percent of the vote to 44.07 percent for Odinga.
"The results of the exit poll do show a clear win for Raila," said the poll's author, James D. Long, a Ph.D. candidate in political science at UCSD. His co-author is Clark C. Gibson, the chair of the university's political science department.
Their research offers more evidence of serious fraud in Kenya's government election commission. Independent observers have alleged that partisans of both candidates altered or invented results in the vote-counting center and that top officials in Kibaki's government pressured election officials to announce incomplete or disputed tallies.
The pollsters contracted Strategic Research, a veteran Kenyan public-opinion firm, which surveyed voters in each of Kenya's eight provinces and in 179 out of 210 electoral constituencies. According to their projections, Odinga, who also led Kibaki in pre-election polls, should have received about 58,000 more votes than he was credited with. Kibaki should have received about 356,000 fewer votes.
In four provinces where Odinga had strong support in the exit poll — the northeastern, western and coastal areas of the country as well as the capital, Nairobi — he received substantially fewer votes in the official tally. Kibaki's vote totals in those areas were surprisingly high.
Kibaki's official margin of victory was a slim 231,728 votes out of nearly 10 million cast.
"Our results cannot definitively prove fraud," Long said. "However, they do highlight important discrepancies in the official vote count that suggest both candidates may have engaged in the artificial production or subtraction of votes."
It's not clear what impact the poll will have now. The International Republican Institute maintained the rights to the poll for six months, ending June 26, after which the researchers could publish it themselves.
In that time, Kibaki and Odinga, under serious international pressure, formed a fragile coalition government, with Odinga as the prime minister. To top officials in both camps, the election and its bloody aftermath appear to be distant memories.
While jockeying already has begun for the next presidential contest, in 2012, there's been little effort to identify or prosecute the perpetrators of post-election crimes or to heal the tribal rivalries that the disputed vote ripped open. Tens of thousands of Kenyans who fled their homes still live in squalid displacement camps.
Ken Flottman, who served as the head of the International Republican Institute's Kenya office until his term expired in May, said that the poll's findings could serve as a catalyst for electoral reform.
"I hope that (Kenya's) parliament will take note of the important research from the University of California in understanding how Kenyans actually voted and in addressing the problems with the conduct of the election," Flottman said.
The institute, which also paid for exit polls in Kenya's two previous national elections, in 2002 and 2005, has tried to distance itself from this poll. Officials initially cited concerns about the data, saying that many survey forms from far-flung areas were returned to pollsters in Nairobi several days after the election due to the violence, and that some forms may have gone missing.
Long and Gibson said those problems didn't compromise the results. Surveys that were returned late showed no signs of being tampered with, and the polling firm eventually received all the allocated questionnaires, the authors said.
The authors also found that, despite the ethnic nature of the violence, most Kenyan voters chose presidential candidates for a variety of reasons, not just tribe.
Voters who were richer and who saw the economy as the most important issue in the election favored Kibaki, who's presided over strong macroeconomic growth in Kenya. Poorer voters who felt shut out from such growth tended to vote for Odinga, the poll found.