brave new world
The NYT ran a very interesting piece on the dirty, open secret of last year's ill-fated Kenyan presidential elections last weekend. Long story short, the International Republican Institute (a U.S. government-supported nonprofit, nonpartisan entity that promotes democracy abroad) ran exit polls during the December 2007 elections. The exit poll results remained a secret. You'll remember the rest of the story; the belief that incumbent President Mwai Kibaki (a member of the Kikuyu ethnic group) stole the election from Raila Odinga (who's a Luo, who (along with the Kikuyu) are one of the largest ethnic groups in Kenya) set off a series of riots and murders across the country that had a distinct ethnic tinge.
The best part about it? Everybody knew what the poll said (that Odinga won), and everybody knew that Mbeki almost certainly stole the election.
Election polling in Africa is a brand new field. (Really. If you're an aspiring, young, Africanist political scientist just starting out, you can make a name for yourself and a have really interesting career by focusing on voting behavior. Just appropriate all the questions and techniques the Americanists use, slap a few questions about ethnicity on your surveys, and start running polls in places that normally don't have a lot of election violence/are covered by the Afrobarometer surveys so you'll have some good data with which to cross compare. The papers that are being written on African voting behavior are, as Nick van de Walle pointed out at ASA last fall, going to be the foundational papers in the field. Get in on the racket while you can.)
Since running election polls in Africa is a pretty new thing, Kenya was supposed to be a good test case for those of us who care about these things. And the people running the studies there were doing really cool things. There were polls being taken via text message, a far more effective way to get a cross-sample of the country's growing middle class than internet or face-to-face surveys.
And it was a good test case. There's no reason to believe that the IRI poll isn't as accurate as any poll, given the conditions.
But the controversy over the poll won't go away. There's an argument that releasing it might have prevented some of the deaths by pressuring Kibaki to settle and form a coalition with Odinga sooner. The U.S. government clearly wanted Kibaki to win, so the ambassador probably really wouldn't have wanted the poll released at the time. And that raises questions about the other dirty secret we never talk about: that Western government actors attempt to manipulate elections in Africa all the time.
I don't mean that Western diplomats are stuffing ballot boxes or anything like that. It's just that there's usually a consensus among the Americans and the British and the French and the Swedes and the Norwegians about which candidate they like best. And the diplomats do all they can to push for the result they believe will be favorable for their country. Which I think makes a pretty good argument as to why foreign government-funding polling should be done and released by university or professional polling organizations or individual scholars.
Meddling happens. That's not going to change. But here's hoping that things will improve as the brave new world of political polling in Africa continues to open to scholars who have no vested interest in electoral outcomes. Everyone is better off when the politicians are kept out of it.