Text size

Perhaps there has never been an election campaign in which so many predicted the failure of the opinion polls. Politicians keep saying that the widespread support for Kadima predicted by the polls is not reflected in the public. The pollsters reply that they can only go according to the polls they have, which predict 35 to 40 Knesset seats for Kadima.

"I can't say I have a feeling Kadima will win fewer votes," says Professor Camil Fuchs, Haaretz's poll expert. "A statistician has no feelings."

Polling firms are concerned by three main anomalies that could skew predictions - low voter turnout, the high level of floating votes and the large numbers of youngsters on cellphones whom they can't poll by land line.

The absentee vote. If the forecast that Kadima will win 35 to 40 Knesset seats is proved false, says Dori Shadmon of Teleseker, "it will derive only from the voter turnout problem. This may be the greatest difficulty in these elections, and then everyone will blame the pollsters for getting it wrong."

"The voter turnout is the most difficult issue," says Rafi Smith of Smith Institute. Although pollsters ask people if they're sure they'll vote, Smith says most people are reluctant to say they won't vote. "I walk on the street in Ramat Gan and there isn't one poster. There is no feeling of elections in the air," he says.

Another unexpected factor that affects the voter turnout and complicates matters is the weather.

The pollsters all predict that Yisrael Beitenu will be the surprise of the elections and win ten or more Knesset seats. However they agree that if the voter turnout is low in the Russian community, it will get less. A low voter turnout among the youngsters, the religious right and the Arabs may also be seen.

The floating vote. Most pollsters said the number of floating votes is unusually high these elections, reaching the equivalent of 20-25 Knesset seats. This week the floating votes in Haaretz's poll dropped to 18 Knesset seats - still high. "It's crazy trying to crack 25 Knesset seats," says Smith. "I'm not sure we know how to do it right."

The cellular phone vote. In the past when everyone had land lines the polls were very accurate. These elections, says Smith, there is a large group of voters, especially young ones, who only have cellphones and cannot be reached by land line. Fuchs says that if there are not enough youngsters in the polling sample, this could be overcome by giving greater weight to the young voters. However, if those with cellphones have different voting patterns than the other youngsters, the poll results may be off. The mobile phones will change future polling methods. One possible direction could be polls combining mobile phones and emails.

The flexible vote. "Sometimes up to ten percent vote differently from what they said in the poll. They're not lying, They change their mind," says Shadmon. In Haaretz's poll, people were asked to say whether they will change their vote. Hard to believe, but 17 percent say they might change their mind, whereas 3 percent are sure they will.

While pollsters emphasize that the polls are right for a given moment, the public sees them as an election forecast. The last polls are held 36 hours before the elections. During that time people can change their decision, as probably happened in the 1966 elections, when polls predicted a narrow victory for Peres, but Netanyahu won.

The refusenik vote. Only 30 percent agree to answer the pollsters.

The misleading vote: Some people deliberately mislead the pollsters, others are not comfortable telling the truth. The pollsters ask how people voted in previous elections, but many tend to "abandon" unpopular parties and "correct" their vote retroactively. In the 1999 elections Shas won 17 Knesset seats and the Center Party six, but in the 2003 polls, many of these voters said they had voted for the Likud. This may be why the pollsters failed to predict the magnitude of the Likud's victory in those elections.