How did the polls in Israel get it so wrong?

How did the pollsters get it so wrong?

After Israeli Prime Minister Benjamin Netanyahu confounded predictions he could go down to defeat and instead scored a solid election win, some of those who conducted the opinion polls sought Wednesday to explain the dynamics involved in the bad calls.

The upshot: It’s complicated.

A week ago, opinion polls began to show Netanyahu’s conservative Likud lagging behind the center-left opposition party of Isaac Herzog. Two days later, in the final surveys allowed under election rules to be published before the Tuesday vote, the gap appeared to have widened, with Herzog’s Zionist Union expected to best Likud by five parliamentary seats.

But pollsters noted that those late-in-the-campaign polls were clouded by large numbers of undecided voters – as many as 20% of those surveyed, by some counts. Netanyahu, who until last week had done almost no direct campaigning, engaged in a blitz of interviews and appearances in the days before the vote, apparently managing to draw off support from far-right parties with tough talk about the Palestinians and external threats to Israel.

As soon as balloting ended Tuesday night, exit polls commissioned by the major Israeli television channels were released. All pegged the race as a virtual tie – a result proved seriously off-base after the all-night vote count.

Pollster Avi Degani, the founder of the Geocartography polling agency, said he was surprised by the degree of divergence between prediction and reality, but at the same time recognized it beforehand as “a possibility.”

An overnight swing during the vote-counting is not unprecedented, including in races involving Netanyahu: In 1996, when he was running against veteran Labor Party leader Shimon Peres, the electorate – as a Hebrew-language catchphrase goes -- “went to sleep with Peres and woke up with Bibi.”

How people vote can sometimes be more a matter of sentiment than science, said pollster Degani.

“In elections, everything is very emotional, particularly in Israel where you don’t have two parties like in the U.S., but many parties, some that are here today and gone tomorrow, but still get votes,” he said. “At the last moment, many votes get lost and the remaining ones get redistributed.”

One important factor: The public is not always truthful with pollsters, especially in a crowded field of contenders in which people might have an attachment to one party, but vote for another for pragmatic reasons.

Statistician Mano Geva of the Midgam polling organization said in a television interview that published polls can end up affecting results, because people make strategic decisions based on them. Some voters, for example, might decide to support a major faction that could have more clout than a small one in danger of oblivion in a system that requires a party to win 3.25% of the vote to gain seats in parliament.

Netanyahu’s last-minute campaigning, featuring an array of strident accusations, may have deepened the rifts in Israeli society, but it also galvanized his traditional base.

In the last days and hours before the vote, Degani said, many long-term but wavering Likud supporters “got scared the left would win, and came back home.”

Sobelman is a special correspondent and King a staff writer.

Follow @laurakingLAT on Twitter for news out of the Middle East