Why This Is the Most Difficult Election to Predict Since Netanyahu's First Victory

Israeli society, the electoral system and frenetic politics make Knesset elections notoriously difficult to poll. With Netanyahu vs. Gantz, top pollster says this has been the hardest to gauge since 1996

People walking by election campaign billboards in Tel Aviv for Likud and Kahol Lavan, April 3, 2019.
Oded Balilty,AP

Polling has set the tempo in Israel’s raucous electoral campaign. The dead heat between Netanyahu's Likud and Benny Gantz’s Kahol Lavan has been the one constant, alongside a languishing Labor Party, the surprise rise of Zehut and the shrinking of all the other right-wing parties — but how accurate are these polls?

The final polls will be released on Friday. Polling will, of course, continue right up to Election Day on April 9 and during it, but Israeli electoral law forbids the publishing of polls in the last three days of the campaign. This is the hiatus that is supposed to allow voters to focus on the issues and candidates, not on who they think may win or which party is in deep trouble.

So far, the polls have shown us that half of the electorate is more or less evenly split between Likud and Kahol Lavan; the rest are split between a dozen parties spanning the ideological and religious spectrum. In addition, the bloc of right-wing and religious parties supporting a fifth Netanyahu government have a small but steady majority. But again, can we trust those polls?

Like in any other election campaign around the world, pollsters today have a wide array of computational and statistical tools at their disposal to create projections. At the same time, though, the disparate means of communication used by different parts of the population make it harder to build a representative sample of voters.

>> Read more: The rabid right can smell victory, but reality will lead Netanyahu to Gantz | Chemi Shalev ■ Israel's last election before it becomes an illiberal democracy | Opinion ■ In 'occupied Scarsdale,' there's only one question: Is far-right right enough? 

As in other Western societies, the older generation is still best sampled via phone calls to their landlines, while younger voters are more accessible via cellphone interviews and text message or internet questionnaires. In Israel, this is made more tricky by the level of web usage among different societal groups: high among secular Jews; lower among Israeli Arabs and ultra-Orthodox Jews. And, just like everywhere else, pollsters in Israel are struggling to work out the best model.

Israel's political map: A rightward shift
Haaretz

The local polling industry has some other problems. In a small country, there simply aren’t enough polling firms and independent pollsters to serve all the parties and news organizations, and many of them end up working for a mixed group of clients. The pollsters will tell you that they maintain “Chinese walls” and strict confidentiality agreements — and some do. But some companies seem to produce results for the media that benefit their political customers.

Elections aren’t the pollsters’ main income. There’s a local or national election, on average, only once every two years, and in the national elections there are no local constituencies and races to poll separately. Most of their business in-between elections comes from consumer research, which focuses mainly on the secular, mainstream Jewish population.

As a result, the ultra-Orthodox and Arab communities — and also elderly Russian speakers who moved to Israel in the early 1990s — are poorly covered by the main polling companies. When they need to be taken into account in a national election, pollsters have a hard time working out their voting patterns and potential turnout. It doesn’t help that members of these communities, especially the ultra-Orthodox, are frequently suspicious of polling and are notorious for trying to mislead with false answers.

Israel’s political culture also makes things less predictable. Relative to other Western countries, Israelis are highly engaged in the political discourse and are voracious consumers of daily news. This leads to a frenetic electorate that quickly switches its voting intentions — and some politicians are particularly adept at taking advantage of this. For example, Prime Minister Benjamin Netanyahu “is an expert at intervening and interacting with voters just before, or right on Election Day,” says Haaretz’s veteran pollster Prof. Camil Fuchs, head of Tel Aviv University’s Dialog Center.

Tribal voting patterns

Haaretz pollster Dr. Camil Fuchs. “In Israel, with so many parties that could be wiped out if they get one or two percentage points less than we predict, it could make all the difference.”
Eyal Toueg

Constant change in the configurations of many Israeli parties, in addition to splits and mergers and the emergence of new parties in every electoral cycle, makes it almost impossible for pollsters to establish voters’ previous voting patterns to improve their models.

“In other countries I’ve worked in, we usually ask voters about their previous voting as well,” says Dr. Dahlia Scheindlin, a pollster and political strategist. “In Israel, there are some parties that have tribal voting patterns. Likud has it, and so have the ultra-Orthodox and Arab parties. But for wide swaths of the political scene, it doesn’t really exist. Labor, for example — which was once a party with tribal voters — has totally lost them.”

In many countries, parties try to use polls to their benefit by overemphasizing leads, real or manufactured, to create a bandwagon effect. But in recent election campaigns, some Israeli parties have used the polls in the opposite way: Highlighting the fact that they may be in danger of losing power, as Likud did in 2015, or even dropping beneath the electoral threshold, as Meretz did when it sent out distress signals in 2013 and 2015. These so-called gevalt campaigns have worked well — though in many cases with small parties, they don’t want voters worrying that they may drop beneath the 3.25 percent threshold, thus sparking fears that a vote for them would “go to waste.”

Info

It is this proliferation of small parties — dozens of which have no chance of getting into the Knesset, while others hover just above or below the threshold — that makes an Israeli election particularly hard to predict.

In 2017, “in the second round of the French presidential election, there was a 6 percent variation between the different polling companies on the result. But they all still got the winner right,” explains Fuchs. “In the Israeli election, with so many parties that could be wiped out if they get one or two percentage points less than we predict, it could make all the difference.”

The 3.25 percent threshold; the proportional representation system; the surplus vote agreements some of the parties sign and the intricacies of the Bader-Ofer method, which calculates how “leftover” seats are allocated when parties don’t pass the threshold — all of these factors contribute to labyrinthine calculations where the tiniest fraction of a decimal point can determine the fate of an entire governing coalition.

Info Poll

It’s a pollsters’ nightmare, one Scheindlin says she prefers to solve by “focusing more on the main blocs, in which voters tend to be more constant, than on the parties.”

The factors making life difficult for Israeli pollsters are all exacerbated this time around. Over 40 parties are running in total, and even though nearly three-quarters of them will almost certainly not make it over the electoral threshold, their votes still have to be reckoned with in polling calculations.

Israeli Arab voter turnout over the past 20 years
Haaretz

Fourteen parties are polling at over 1 percent. Of these, four are completely new: Kahol Lavan; Zehut; Hayamin Hehadash; and Gesher. Four are running in different configurations: Labor, which ran together with Tzipi Livni’s Hatnuah in 2015, is now standing alone; Hadash-Ta’al is running separately from Balad-United Arab List, with which it ran as the Joint List in 2015; and Habayit Hayehudi has now linked up with National Union and Otzma Yehudit in the Union of Right-Wing Parties. Only six major parties — Likud, Meretz, Kulanu, Yisrael Beiteinu, Shas and United Torah Judaism — are running in the same form as they did four years ago.

Around half of voting-age Israelis are casting ballots for parties that are either new or whose composition has changed significantly since 2015. To make matters worse, only Kahol Lavan and Likud are polling in double figures — the rest are set to get between four and eight seats. Trying to work out their final tally and which of them may or may not cross the electoral threshold is mind-boggling.

Another issue making this election particularly difficult to call is the question of turnout among the Israeli Arab community. In past decades, it was usually significantly lower than the general turnout — usually just over 50 percent. In 2015, partly due to the four main Arab parties running together in the Joint List, it jumped to 63 percent (still lower than the general turnout that year of 72 percent). So far, the polling of expected Arab turnout in 2019 remains inconclusive: Some polls have indicated turnout will be similar to 2015; others predict a slump due to the Joint List splitting and anger over passage of the nation-state law last year. A 10-percent drop in Arab turnout could mean two seats lost by the left bloc and Balad-United Arab List not even crossing the threshold. This would almost certainly clear Netanyahu’s path to victory on Tuesday.

The emergence of Zehut — the far-right, libertarian party that calls for the legalization of marijuana — is also a disruptive element. Many of its supporters are first-time voters, and the party’s appeal seems to exist across the board: Religious, secular, right-wing, centrists and even some on the left.

“The last time we were so unsure was in 1996, when just a fraction of a percentage point divided Netanyahu and Shimon Peres — and we got it wrong,” says Fuchs, recalling Israel’s first direct election for prime minister (a format subsequently abandoned in 2003). He adds: “This is the most difficult election for me as a pollster in 23 years.”