By Imbrahim Oruko for the Star
AN exit poll commissioned by American academics indicates that no presidential candidate reached 50 percent in the March 2013 election. Details and analysis of the exit poll are contained in the latest issue of the Journal of East African Studies.
The exit poll on March 4 interviewed 6,258 voters as they left 404 polling stations spread over all 47 counties. The enumerators randomly selected every second voter leaving polling stations. They had an 88 percent response rate on the interviews which were transmitted live on Android phones. The exit poll indicated that Raila Odinga received 40.9 per cent of the vote while Uhuru Kenyatta received 40.6 percent.
The authors of the article were Karen Ferree and Clark Gibson of University of California, and James Long of the University of Washington.
On March 19, the IEBC declared that Uhuru won the election with 6,173,433 votes against Raila Odinga’s 5,340,546. Uhuru therefore surpassed the constitutional threshold of 50% by 4,100 votes giving him 50.07 percent of the votes.
However the American academics wrote that the exit poll data was “highly suggestive of a deeply flawed electoral process and challenge claims that Kenyatta won a majority of votes in the first round.” “Exit polls represent important tools for verifying official vote counts,” the JEAS article states.
“Unlike actual ballots, exit poll results are immune to ballot stuffing, tally manipulation, or technological failures in the counting system. Therefore they can reflect the vote more accurately than certified vote counts in corrupt elections,” states the article.
The American authors noted that the exit poll result was similar to the Ipsos Synovate poll ten days before the election which showed a statistical tie between Uhuru at 44.8 percent and Raila at 44.4 percent.
The authors then interrogated why there was a 9.5 percent gap between their exit poll results and the final IEBC result. They dismissed the argument that “Kenyatta’s victory flowed from the high rate of registration” because the exit poll by definition only questioned those voters who were on the final register.
They pointed out that the “official IEBC results show consistently high turnout in both Kenyatta and Odinga strongholds.” The academics doubted that the discrepancy could have been the result of last minute changes because exit polls “capture all last minute swings in support”.
They noted that 12 percent of voters Refused to Answer (RTA) the exit poll. The academics analysed whether it was “plausible” that Uhuru won the 80 percent or more of the RTAs to get a simple majority on the first round.
The academics stated that “the coefficients on the political controls are most consistent with political neutrality among RTA voters.”
Using a special “multiple imputation method”, the academics said, “our estimates suggest that 40.2 percent of the RTAs went to Odinga, another 40.5 percent to Kenyatta, and 19.3 percent to other candidates.”
“Overall we estimate that Odinga took 45.3 percent of the vote and Kenyatta 45.6 percent, a statistical tie,” the JEAS authors state. The authors also said that 10 percent of the exit poll respondents expected the IEBC to have problems with the count.
They pointed out that the IEBC result was marred by the failure of Electronic Voter Identification Devices, the failure of polling station officials to report by cellphones, and the lack of transparency in the official tallying process.
The JEAS authors also pointed out that the Elections Observation Group (ELOG) observers found that the electronic poll book failed in 55 percent of voting streams, 17 percent of ballot boxes were violated during voting , and 4 percent of ballot boxes were not shown to be empty before voting. The ELOG figures for Parallel Vote Tabulation gave 49.7 percent for Uhuru and 43.3 percent for Raila.
The American academics said that ELOG “perhaps overstates the case” when saying that their tabulation “verified” the IEBC results as it was within their margin of error.