Opinion Polling for the 2015 Israeli Legislative Election

The 2015 election was called after the Netanyahu government dissolved the Knesset early, with a vote scheduled for March 17. Polls taken in November and December 2014 forecast a right-wing bloc of 70–75 Knesset seats under Netanyahu, and INSS said he was apparently influenced by those forecasts when deciding to dissolve the government and call the election.[1] Those forecasts proved significantly optimistic for the right: the right-wing and ultra-Orthodox bloc ultimately won 57 seats, down from 61 in the previous Knesset.[1]

Snap polls published when the Knesset agreed to hold an election projected Likud at 22 seats and Jewish Home at 17, with Jewish Home drawing early expectations of at least 17 seats and Yisrael Beitenu as many as 15.[2][3] The main contest that emerged as the campaign developed was between Likud, led by Prime Minister Netanyahu, and the Zionist Union, the center-left alliance of Labor and Ha'Tnua whose leaders, Yitzhak Herzog and Tzipi Livni, had agreed to serve rotating terms as prime minister if their list led the next government.[4]

The race was described before the election as a close contest. A consolidated polling chart credited to the Britain Israel Communications & Research Centre was based on a basket of selected opinion polls each week, although the accessible text did not reproduce the underlying poll-by-poll data, individual dates, or party-by-party numerical observations.[4]

A Rafi Smith Institute poll for Globes, reported on January 1, 2015, projected 22 Knesset seats for Likud and 24 for Labor-Hatenuah, which would later run as the Zionist Union, along with 16 for Habayit Hayehudi, 10 for Moshe Kahlon's prospective party, and 9 for Yesh Atid.[5] January 2015 polls also projected Kulanu winning between 8 and 12 seats; the party ultimately won 10, within that forecast range.[6]

After Labor and Hatnua joined forces, a Channel 2 poll of 500 respondents conducted by Midgam projected 24 seats for the joint list, one more than Likud's 23, and estimated that the alliance would receive 20 percent of the vote.[7] That poll gave 15 seats to Jewish Home, 9 each to Kulanu and Shas, 8 each to United Torah Judaism, Yisrael Beytenu, and Yesh Atid, 5 to Meretz, and 11 to the Arab parties. It reported Netanyahu as the preferred prime minister for 36 percent of respondents, with Herzog and Livni together trailing by three percentage points and 31 percent declining to answer or unsure; the margin of error was 4.5 percent.[7]

A survey conducted February 1–3, 2015, questioned 824 adults who identified as very or somewhat likely to vote, using landlines, mobile phones, and an online panel. Of the respondents, 10.1 percent were interviewed in Arabic and 10.9 percent in Russian, and the reported margin of error was ±3.41 percent at a 95 percent confidence level.[8] The survey found 24 percent of likely voters undecided, compared with 31 percent at a similar point before the 2013 election. After distributing undecided voters by ideology and perceptions of Netanyahu, it projected 27 seats for the Zionist Union and 23 for Likud, with 11 each for the United Arab List, Jewish Home, Yesh Atid, and Kulanu.[8] Netanyahu's unfavorable rating stood at 54 percent; 67 percent rated his performance as prime minister fair or poor, and among undecided voters that figure rose to 71 percent. The findings were interpreted as suggesting that declining views of Netanyahu among centrists, rather than Herzog's own standing, were driving the Zionist Union's projected lead.[8]

By February 19, polling was described as relatively stagnant, with little movement between the main blocs.[2] Likud was estimated at about 20 percent of the vote, or 25 seats, while Jewish Home had fallen to 11 seats, well below the early campaign expectation of at least 17, and Yisrael Beitenu to 6, against an earlier expectation of as many as 15.[2]

Polls taken just days before the election put the Zionist Union three to four seats ahead of Likud.[9] A poll cited from that Friday's Yediot Aharonot found 55.2 percent of respondents naming socioeconomic issues as the main factor determining their vote, compared with 28.4 percent who cited security or foreign policy.[9] The Labor-Hatnua alliance had produced a ten-seat increase in its polling standing, which it maintained throughout the campaign, along with a gain of "a couple of seats" in the preceding week.[9]

Polling reported on March 10 showed a similar picture. A Knesset Channel survey of just over 1,000 respondents projected 24 seats for the Zionist Union and 21 for Likud.[10] A Channel 2 poll of 1,003 respondents, with a 3 percent reported margin of error, projected 25 for the Zionist Union and 21 for Likud, the widest Zionist Union lead recorded in surveys to that point. It also projected 13 seats each for the Joint Arab List and Jewish Home, 12 for Yesh Atid, 8 for Kulanu, 7 for Shas, 6 each for Yisrael Beytenu and United Torah Judaism, and 5 each for Meretz and Yachad.[10] Despite that seat projection, Channel 2 respondents still preferred Netanyahu as prime minister over Herzog by 49 to 36 percent, with 15 percent unsure.[10]

A Rafi Smith Institute poll published March 12 projected 24 seats for the Zionist Union and 20 for Likud, a four-seat lead, with Likud down from 22 in the previous Globes poll and the Zionist Union up from 23.[11] The last Channel 2 poll before the election, conducted by Midgam's Mina Tzemach and Manu Geva, projected 26 seats for the Zionist Union and 22 for Likud, with 13 for the Joint List, 12 for Yesh Atid, 11 for Habayit Hayehudi, 8 for Kulanu, 7 each for Shas and United Torah Judaism, and 5 each for Meretz, Yisrael Beitenu, and Yahad.[12] Channel 10's last poll also gave the Zionist Union a four-seat lead, projecting 24 to Likud's 20, and reported Netanyahu as the preferred prime minister for 43 percent of respondents, compared with 37 percent for Herzog and 22 percent undecided.[12]

Polls during the week before the election put the Zionist Union slightly ahead of Likud, forecasting 25 or 26 seats for the Zionist Union and 21 or 22 for Likud.[13] Some analyses suggested that undecided voters, whose views of Netanyahu appeared to be worsening, could tip the balance toward the Zionist Union; a cited February 2015 report was titled "1 in 4 Israelis still undecided, but more of them leaning toward Herzog."[4]

Two Statnet polls of 500 respondents each were conducted on February 18 and March 14, 2015, using samples representing different geographical regions, religions, and family ties among Arab citizens.[14] The February poll estimated Arab turnout at 62.4 percent and the March poll at 63.4 percent; actual turnout among Arab citizens was 63.5 percent, a close match.[14]

The share of Arab respondents planning to vote for the Joint Arab List rose from 78 percent in February to 81.4 percent in March, against an actual result of 82.4 percent.[14] The March poll forecast 13.6 mandates for the Joint List; the list won 13 seats.[14][15] By comparison, the Zionist Camp was forecast at 6.5 percent in February and 4.4 percent in March among Arab voters, against an actual 4.9 percent; the corresponding figures for Meretz were 5, 3, and 2.6 percent, and for Likud 2.8, 2.7, and 1.5 percent.[14] Among Arab respondents who had not previously voted but said they would vote in 2015, or who were undecided, 58 percent cited the formation of the Joint Arab List as their reason for participating.[14]

Exit polls published as voting closed on the evening of March 17 projected a near-tie between Likud and the Zionist Union.[15] The Channel 1 exit poll gave each party 27 seats; Channel 10 also showed the two parties level, while Channel 2 gave Likud a narrow lead of 28 to 27.[16] Despite the near-parity in seat projections, the exit polls suggested an advantage for the right-wing bloc overall and indicated that Likud was better positioned to form a coalition. The same reporting gave a preliminary turnout indication of 71.8 percent.[16]

Pollster Mina Tzemach's intraday tracking added further detail. Her election-day surveys showed the Zionist Union leading Likud by three to four seats at noon.[17] After Netanyahu posted a warning that Arab voters were turning out in large numbers and called on supporters to vote, Tzemach reported that the parties were tied by 5 p.m. and that Likud led 28 to 26.5 seats by 8 p.m. Her 10 p.m. exit poll projected 28 seats for Likud and 27 for the Zionist Union; a survey finalized at 11:30 p.m. showed Likud at 29 and the Zionist Union at 23.[17]

Analysts pointed to features of Israel's multiparty system as reasons that polling in Israel is structurally more difficult than in two-party systems. The Foundation for Defense of Democracies argued that polling did not capture the full picture of multiparty competition, noting that pollsters commonly asked both which party a respondent would vote for and whom the respondent wanted as prime minister, and that the answers did not always match.[18] Strategic voting and coalition calculations meant that a party preference alone did not straightforwardly predict the government a voter's ballot would ultimately help produce.[18] An estimated 10 percent of Israeli voters were still undecided on election day, and voters could remain uncertain about the government their ballot would produce even after casting it.[18]

The electoral threshold further complicated forecasting. Parties near the threshold could either enter the Knesset or fail to win representation, materially changing the distribution of seats and the possible coalition arithmetic.[9] Turnout levels, undecided voters, parties close to the threshold, and last-minute events all introduced uncertainty into the final result.[9]

The 2017 academic volume The Elections in Israel 2015, edited by Michal Shamir and Gideon Rahat, included a chapter by Camil Fuchs titled "Surveys and Election Forecasts in a World of Social Media and Party Dealignment" and a chapter by Yariv Tsfati on media attitudes and changes in voting intentions, indicating that the 2015 polling experience generated sustained scholarly attention.[19]

As counting progressed on election night, a report based on 99 percent of votes counted put Likud at 30 seats and the Zionist Union at 24 and cited a turnout figure of 71.8 percent.[20] A later official statistical table records Likud with 985,408 votes, 23.4 percent of valid votes, and 30 seats, against the Zionist Union's 786,313 votes, 18.7 percent, and 24 seats. It records nationwide turnout at 72.3 percent among 5,881,696 eligible voters.[21]

The gap between forecast and result was substantial. Final pre-election polls put the Zionist Union four seats ahead of Likud, the television exit polls showed a near-tie, and the final result gave Likud a six-seat margin.[22][15] A post-election survey commissioned by the New Majority, covering 1,133 voters, found that more than a quarter of the electorate decided very late: 15 percent made their decision on the way to the polling station on March 17, and 13 percent decided in the final days of the campaign.[23] Among these last-minute decision-makers, 62 percent voted for right-wing parties, 16 percent for left-wing parties, and 21 percent for centrist parties.[23] The survey also found that 66 percent viewed Netanyahu as a strong leader, compared with 24 percent for Herzog, and that 79 percent of voters who ranked security as their top concern voted for right-wing parties.[23]

Retrospective accounts offered several, sometimes competing, explanations for why polls and exit polls underestimated Likud's result. Pollsters Mina Tzemach and Joseph Sarid attributed Likud's late gain largely to movement within the right-wing bloc, especially Jewish Home voters shifting to Likud, rather than a major change between the right- and left-wing blocs as a whole.[24] They pointed to a large right-wing rally shortly before the election and to Netanyahu's election-day appeal about Arab voter turnout as factors in that intra-bloc shift. Tzemach noted that exit polls closed at 8 p.m., leaving the final two hours of voting unmeasured, and described the exit-poll findings as precise for the time at which they were taken. Sarid said the polling industry's methods and assumptions required reassessment.[24]

Pollster Kamil Fuchs later discussed possible nonresponse bias and respondents giving false answers as contributors to the systematic underestimate of Likud, which pre-election polls had placed at 22–24 seats.[25] A separate commentary attributed an estimated 5.5 seats' worth of additional Likud votes to previously inactive supporters who were not visible to polls and who emerged on election day. It cited turnout of 67.78 percent in 2013 and 72.36 percent in 2015, characterizing the change as an increase of approximately 194,803 voters.[26] These were retrospective interpretations rather than established causal findings.

Likud's campaign pollster John McLaughlin said the campaign's internal polling on the Sunday night before the election had put Likud at 23 percent of the vote, which he described as close to its eventual share.[27] Isaac Herzog, by contrast, stated that his party's own polls had shown him five seats ahead as late as noon on election day.[27] Many commentators attributed Likud's victory at least partly to Netanyahu's final-days campaign statements urging right-leaning voters to support Likud rather than smaller parties in order to prevent the Zionist Union from taking power.[28]