Diary

Keep This In Mind When Reading Election Polls

Israeli newspaper Haaretz:

Israel’s pollsters have admitted their failure to correctly predict Wednesday’s Kadima primary results.

The polls made public ahead of the vote promised a resounding victory to Foreign Minister Tzipi Livni, by a margin of over 10 percent, but her actual win was extremely narrow. Livni secured 43.1 percent of the vote while Transportation Minister Shaul Mofaz fetched 42 percent – a difference of only 1.1 percent.

But from conversations with the heads of polling organizations it seems most agree that the main reason for the failure was the large number of voters who misled pollsters, telling them they intended to vote for a different candidate than the one they actually chose.

“We will find out where we went wrong but there will be very little to be done even when we do find out,” Mano Geva, CEO of Midgam Research and Consulting, said . “We can’t tell the public off. We have no way of dealing with people who interfere with the polls’ integrity. It’s a shame.”

“There is a huge problem here. We are living in a polling dictatorship,” Channel 1 reporter and political commentator Ayala Hasson said. “Polls have become a reality- and opinion-shaping tool that doesn’t necessarily reflect reality.”

Hasson said that rumors of a landslide for Livni, based on the first ballot boxes to be counted, began spreading in the early evening Wednesday, when voting was still in full swing. “Who knows what effect this has had on actual voting,” Hasson said. “We all built castles of commentaries based on virtual reality,” Hasson said with respect to the press.

Some reporters accused polling firms of “fixing” their results with their colleagues so that none would “crash too hard.” Pollsters denied these charges.

Channel 10 pollster Prof. Camil Fuchs said, “We rightly predicted the identity of the winner, as well as the fact that victory will already be secured in the first round.”

However, Fuchs added, “the statistical error was massive, much greater than any incidental margin of error. There was clearly a divergence” between people’s reported voting and their actual voting. Fuchs added that one reason people deliberately lie to pollsters is because “they are sick of answering surveys.”

If you think it can’t happen here, think again. This article also helps put polls in perspective.