Pollsters reconsider primary exit polls

"We can't help it if voters lie."

By SHELLY PAZ
livni votes 224 88 (photo credit: AP [file])
livni votes 224 88
(photo credit: AP [file])
Media exit polls proved widely off the mark Wednesday night, as Foreign Minister Tzipi Livni indeed emerged as Kadima's new leader, but by a margin much smaller than the surveys of voters had predicted. Far from the 10 percent to 12% advantage attributed to Livni, the final results Thursday morning showed that she had beaten Transportation Minister Shaul Mofaz by only 1.1% - 43.1% to Mofaz's 42%. Pollsters hastened to explain that party primary elections have their own dynamics that cannot always be accounted for in exit polls, in which voters are questioned immediately after they have cast their vote and exited the polling station. Veteran pollster Dr. Mina Zemach of the Dahaf Research Institute, which had predicted a 48% victory for Livni and had given Mofaz 37% in an exit poll for Channel 2, said she might not conduct any more exit polls for party primaries. "Primaries in general have many problems," Zemach explained. "Many people who are members of a certain party do not really identify with the party because they were recruited to be members only for the primaries. Often these voters have no motivation to answer the pollsters' questions truthfully, and sometimes they give misleading answers." She added that her research firm tries to overcome this problem by asking several other questions, in an effort to gauge the voter's real preference, "but that isn't enough," she conceded. A party like Kadima poses a special problem, she added, since it is a new party that had never held a primary before and whose core constituency was not well defined. "Usually we can build a forecast on the basis of a party's profile and its traditional supporters' behavior," she said. Mano Geva is the general manager of the Midgam Consulting and Research Company, which had predicted a 47.2% victory for Livni and given Mofaz 37.1% of the vote in its poll for Channel 1. He said he might also stop doing exit polls for party primaries. "Certain sections of Israeli society don't cooperate, and even deceive the pollsters," he said. "There is no problem with the method we use to conduct these exit polls," he continued. "We have shown accurate results in past exit polls and [in] the last one, for the Labor Party [primary], we even predicted a 51% victory for Ehud Barak and were off by only 0.3%. We depend on the voters' willingness to cooperate with us," Geva said. Calling close races via exit polls, however, is a problem that is not unique to primaries. On Election Day in 1996, Israelis went to sleep believing Shimon Peres was their next prime minister, but woke up to find that in fact the winner had been Binyamin Netanyahu. Shvakim Panorama was the only group whose poll - which was not an exit poll but was released a week ago by Israel Radio - was remotely accurate. It predicted a close race and said Livni would win with 39.6% of the vote while Mofaz would get 35.3%. In a head-to-head race, the poll said, Livni would receive 44.6% and Mofaz 41.7%. "The more the candidate is unpopular and enjoys support mainly from certain sectors, the candidate's supporters tend not to reveal their real choice to a pollster for various personal reasons," said Shvakim Panorama general manager Yossi Vadana. "Many of them refuse to answer [or] they choose the 'I don't know' option. Others simply try to mislead the pollster and quote the name of the most popular candidate in the race," he said. Vadana claims to have found a way around this obstacle. "Mofaz was not hallucinating when he predicted he would get 43.7% of the votes. He used [the same] formula I used, which involves conducting a poll among groups that typically support one candidate, for example Mofaz, and according to that percentage and by breaking down their answers we can calculate the margin of error and add it to the result we get in a comprehensive poll of all candidates' supporters," Vadana said.