Saturday, January 5, 2008

Exit poll projects Saakashvili win

The group of four local organizations, commissioned by Rustavi 2 and other TV networks, have released preliminary exit poll results:

See our earlier post for why this could be a problem.

Misha Saakashvili: 52.5%
Levan Gachechiladze: 28.5%
Badri Patarkatsishvili: 6%
Shalva Natelashvili: 5.5%
Davit Gamkrelidze: 3.6%
Gia Maisashvili: 0.8%
Irina Sarishvili: 0.4%

Margin of error is 2%, and the non-response rate was 23%.

The non-response rate could be a disputed factor; Gutbrod of CRRC suggests Saakashvili's campaign may be premature in declaring a victory tonight.

Exit polls could 'undermine trust in election'

In under an hour, the first of two separate groups running exit polls will release their results.

The first exit poll results will come from a Ukrainian group of uncertain provenance; the second, at 10 p.m. Tbilisi time, from a group of four reputable NGOs and public institutes in Georgia.

Opposition candidates have urged the international community to preemptively declare the exit polling unacceptable. The group of local organizations was commissioned by the country’s leading television networks (except pro-opposition Imedi TV, which has suspended its broadcasting), all of which have been accused of a pro-government bias; opposition campaigners allege the government indirectly commissioned the exit poll, and will manipulate the results to justify a Saakashvili win.

There is no direct evidence that the exit polling will be manipulated, but even the best-intentioned exit polls could be “a potential source of conflict,” according to CRRC’s Hans Gutbrod, who has observed five elections in Georgia.

“Exit polls are reasonable tools, but end up often deviating from the actual results,” Gutbrod says. He pointed to exit polls in America’s 2000 presidential election, which incorrectly called the race for George W. Bush in Florida. And in 2004, exit polls showed John Kerry winning in Ohio, but he ultimately lost the race there to Bush.

“One reason the exit poll could be off,” Gutbrod explains, “is because voters who exit the polling station may have reasons not to be entirely candid about who they’re voting for, particularly in this type of context [where there is alleged pressure on some voters to support Saakashvili].”

In 1992, for example, exit polls showed John Major losing in Britain’s elections; it turned out it was less socially acceptable to say you voted for the Conservatives.

There are also methodological challenges in conducting exit polls in Georgia, he adds, where no one has much practice in holding (or judging) democratic elections.

The exit polling we see tonight is as likely to cloud the situation as give a clear indication of the result. Recent polling released by Saakashvili’s campaign predicts 52 percent of the vote for him today, making it entirely possible that exit polls and the official results could each show Saakashvili on opposite sides of the 50 percent he needs to avoid a runoff.

On top of that, there are two separate exit polls being run today, both by groups which have had their credibility questioned.

Gutbrod says it’s fairly likely the two exit polls won’t be in close agreement, and politicians will naturally grab onto whatever numbers favor them, regardless of the veracity of the result.

“Either way, the exit polls were going to be politicized,” he concludes. “[They] can actually undermine trust in the election, rather than confirm—even if the election has been going perfectly well.”

Thursday, May 22, 2008

How do they spot it?

Hans at CRRC has an interesting post which sheds a little light on OSCE/ODIHR's election observation methodology.