The results of the Israel’s second round of general elections on September 17 have plunged the country into its second political stalemate this year, with the ruling Likud party winning 32 Knesset seats to challenger Blue and White’s 33, leaving the rival blocs virtually tied at 55-54.
The complexity of current Israeli politics may be partially responsible for the continued impasse. In the run-up to the election, even as parties vied to poach voters from their rivals, it was unusually difficult for them to analyze voter migration patterns — and thus to ascertain whether or not their tactics were actually working.
Trying to make sense of Israeli voting patterns, a number of high-tech, cyber and data processing experts have recently developed sophisticated research methods that allow for what they claim are detailed and accurate analysis.
One method developed straight out of Startup Israel has already found surprising results: that many of those who voted for the ruling Likud party in the April 2019 elections simply did not head to the ballot box this time — and cost the party three seats.
This method was developed by Arick Goomanovsky, the co-founder of computer and network security firm Ermetic, and is based on a logical-mathematical model developed by data scientist and algorithm developer Itamar Mushkin, of software startup Precognize. Mushkin previously published an analysis of voter migration patterns from the 1990s to 2015. The method has not been academically substantiated.
Goomanovsky takes a micro view of Israel and analyzes voting patterns at various polling stations by comparing their results to previous voting patterns in the same places. Mushkin’s method delves into voting patterns of communities and compares them to the patterns found in the same communities in previous elections.
“Itamar’s method analyzes the changes in voting patterns of all voters in Israel, that’s its advantage. My method focuses on polling stations and not communities, so it has a smaller, more homogeneous and focused test group,” Goomanovsky told Zman Yisrael, the Hebrew sister site of The Times of Israel.
“It’s important to note that the brilliant idea on which both methods are based was Itamar’s,” he stressed.
As for the margin of error, Goomanovsky explained that his analysis is not based on the Probability Model, which makes it difficult to quantify. “But I would say that the likelihood of errors in the model, which corresponds to the standard deviation in probability-based surveys, is similar to or close to 5 percent.”
“In any case, both methods work, as we can see from the fact that their results are quite consistent,” he added, noting that both his and Mushkin’s results show, for example, that Likud voters remained loyal to the party, while there was a shift between Labor and Meretz voters.
“Most polling stations in Israel show certain voting patterns that make changes easy to spot,” Goomanovsky explained. “Say, for example, that a certain polling station shows that in April’s elections 32% voted for the ‘Purple’ party, 38% voted for the ‘Blue’ party, and 22% voted for the ‘Grey’ party. In September, we see that at the same polling station, only 29% voted ‘Purple’ and 25% voted ‘Grey’ but there was no change in the number of ‘Blue’ voters. In that case, we can conclude that 3% migrated from ‘Purple’ to ‘Gray.'”
The algorithm analyzes the changes in each polling station and then calculates the average of voter migration in all polling stations, from which it derives the national average.
The changes in voting patterns in each ballot are easy to detect and they help decipher the complex changes in what has become a chaotic national voting map.
The fact that only five months separated the two election campaigns of 2019 made the analysis easier, Goomanovsky said.
“Very few people died, reached the age of 18, emigrated or changed their address during this period,” he said. “This means that there was little public turnover in polling stations, so if we see a change in a particular ballot between the last two election campaigns, it means people changed their vote. Scrutinizing the results at a particular polling station allows us to see exactly what changed.”
Some of the data is not surprising: 100% of those who voted for the Arab or mostly-Arab parties Balad, Ra’am-Ta’al and Hadash in April, voted for the Joint List — the faction under which they united — in September, and some 97% of those who in April voted for the Jewish Home, Shas and Yisrael Beytenu parties did the same last week.
The Joint List, Shas and Yisrael Beytenu also received a boost from new voters, both those who did not vote for them five months ago and those who did not vote at all. As a result, all three increased their parliamentary power: Yisrael Beytenu, which won five seats in April, secured eight seats in September’s elections, Shas went from eight to nine seats, and the Joint List from 10 to 13 seats.
An analysis of the results concerning the Likud, however, is more surprising. The party dropped from 35 to 32 seats, but the data does not seem to support claims by rival parties that they were able to pull voters away from the ruling party.
Goomanovsky has found that 92% of Likud voters supported the party both in April and in September, while close to 8% — the three mandates the Likud lost — simply didn’t vote.
Ahead of September’s elections, two politicians who left Likud to form their own parties — Kulanu chief Moshe Kahlon and Zehut leader Moshe Feiglin — returned to the party, but the data shows their voters did not widely follow, scattering across the political spectrum instead.
A segmentation of the results showed that only 16% of those who voted for Kulanu in April voted for the Likud in September, while 40% voted Blue and White. Seven percent voted for Shas, 12% for Labor-Gesher, 20% for Yisrael Beytenu, 3% for the far-right Otzma Yehudit party, and 2% voted for the Tzomet party.
Neither Otzma Yehudit nor Tzomet, headed by former — and highly controversial — Likud MK Oren Hazan passed the 3.25% electoral threshold, which is roughly translated into four Knesset seats.
Among those who voted for Zehut in April, only 4% voted for the Likud last week.
Goomanovsky’s data shows that the highest number of former Zehut supporters — 27% — voted for Shas, followed by Otzma Yehudit at 26%, Yisrael Beytenu at 20%, Yamina at 10%, Labor-Gesher at 9%, and 3% voted for United Torah Judaism. The Blue and White party received only 1% of Zehut supporters’ votes.
An analysis of Blue and White’s voting patterns shows that 91% of those who voted for the party in April did so again last week. Still, 3.5% of its voters migrated to Yisrael Beytenu, 3.5% voted for Labor-Gesher, and 2% voted for Meretz.
All shook up
Meanwhile, the turmoil that shook up the smaller center-left parties between the two election campaigns changed their would-be electorates.
Running independently in April, Labor won six Knesset seats and Meretz secured four, while Gesher did not pass the electoral threshold. September’s elections, however, saw Labor and Gesher join forces, and Meretz team up with former prime minister Ehud Barak’s Israel Democratic Party and breakaway Labor MK Stav Shaffir to form the Democratic Camp.
As a result, only 48% of those who voted for Labor in April voted for it again last week. Forty-eight percent of April’s Labor voters gave their ballots to Meretz, while the Likud, Yisrael Beytenu, United Torah Judaism and Tzomet each received 1% of the remaining votes.
Meretz did not fare better, as only 52% of those who voted for it in April did the same in September. Fifteen percent migrated to Labor-Gesher, 14% supported the Joint List, and 4% chose Blue and White. According to Goomanovsky’s data, 15% of Meretz voters did not vote at all last week.
Gesher also saw almost half of its voters migrate elsewhere, as 56% voted for it again, 15% voted for the Likud, 9% for Blue and White, 7% for Yisrael Beytenu, 4% for Tzomet, 3% for Shas and Yamina each, 2% for Otzma Yehudit and 1% voted for United Torah Judaism.
In their September reincarnation, Labor-Gesher won six Knesset seats and the Democratic Camp secured five.
Goomanovsky’s analysis showed that while right-wing voters tended to remain within the bloc, their migration still created substantial shifts.
During April’s elections the small right-wing parties offered voters a choice between the New Right, formed by former Jewish Home party leader Naftali Bennett and Ayelet Shaked, and the Union of Right-Wing Parties (URWP), an amalgamation comprising Jewish Home, Otzma Yehudit and the National Union party.
The URWP won five seats, but the New Right fell short of making it into parliament.
Seeking to better their position ahead of September’s vote, the New Right, Jewish Home and National Union joined forces to form Yamina, which secured seven Knesset seats, while Otzma Yehudit did not pass the electoral threshold.
The data shows that 69% of URWP voters lent their support to Yamina, 24% voted for Otzma Yehudit, 3% for United Torah Judaism, and 2% for Shas. The Likud and Tzomet each won 1% of URWP supporters’ votes.
Interestingly, despite situations in which party members did not vote en masse in September, Central Election Committee figures indicate that voter turnout in the recent elections stood at 69.4%, higher than April’s 67.9%.
Goomanovsky’s analysis of the result concluded that 89% of those who didn’t vote in April again didn’t vote on September. Eight percent cast their vote in favor of the Joint List — reflecting the rise in voter turnout in the Arab sector — 2% voted for Yisrael Beytenu and 1% for Shas.
Still, Israel’s political parties have shown little interest in using Goomanovsky and Mushkin’s methods to potentially improve on future results.
“We developed [the algorithms] and published the information voluntarily — this type of information should be public,” Goomanovsky said. “After Itamar published his analysis, I contacted people in parties from across the political spectrum and turned their attention to the results. Some expressed interest and they may use these methods. Others didn’t really understand what I was talking about.”
This article first appeared in Hebrew on The Times of Israel’s sister site, Zman Yisrael.