Polling Place by Polling Place

How the live projection was built, and why it failed on the night, is in Station by Station.

A national result is an average, and an average hides things moving in opposite directions. On a projection from the count so far, Israel’s election on 27 October moved the Netanyahu bloc (Likud, Otzma Yehudit, Religious Zionism, Shas and United Torah Judaism) from 64 seats to about 54, seven short of the 61 a government needs in the 120-seat Knesset. That is the headline. Underneath it, some places moved much further than the country did, some moved the other way, and in one group of towns turnout did not move at all while it rose everywhere else.

These findings are preliminary. They come from the Central Elections Committee’s results file of 10:02, Israel time, on the morning of the 28th, with 3.76 million valid votes counted. The envelope ballots, cast by soldiers, diplomats, prisoners and hospital patients, are still to come. When the count is final I will update the findings here, with the date of the update. The seat figures are a projection from the same file, not the result.

Arab turnout jumped by more than twenty-five points

The committee put national turnout at 77.7% when the polls closed, against 70.6% for the whole of 2022. By far the largest rise came from one community. In polling places where the Arab lists won a majority in 2022, turnout went from 50.6% to 76.9%. The polling-place figures here cover only the places matched across the two elections, and none of the envelope ballots, so they run below the committee’s national figure and compare only with each other.

It rose most in the Triangle, the band of Arab towns along the old Green Line in the centre of the country, and among the Bedouin of the Negev: 42 points in Umm al-Fahm, 39 in Fureidis, 38 in Hura, 37 in Jatt. It rose least in Galilee towns with large Christian and Druze populations, 14 points in Mi’elya and Kafr Yasif, and in the Arab neighbourhoods of Haifa, 12.

Turnout was half the story. The other half is that the Arab parties ran together. A list needs 3.25% of the national vote to win any seats. In 2022 Balad, one of the Arab parties, ran alone, took 2.9%, and elected no one. This time the Joint List, the alliance of the other Arab parties, won 60% of the vote in Arab towns, and Ra’am, the United Arab List, most of the rest. The Arab lists already have 525,000 votes counted, more than the 511,000 they won in the whole of 2022, and the projection gives them 18 seats against 10.

Haredi turnout stood still, and Shas lost ground

In Haredi (ultra-Orthodox) polling places turnout did not rise at all: 72.8% in 2022, 72.2% now. In Beit Shemesh’s Haredi neighbourhoods it fell six points. Those who did vote turned away from Shas, whose share fell in every Haredi town with enough polling places to compare: by 10.9 points in Beit Shemesh, 8.0 in Betar Illit, and 7.0 in Modi’in Illit and Tiberias. Nationally Shas is projected at seven seats, against eleven in 2022.

Some of that vote went to Itamar Ben Gvir’s Otzma Yehudit, which took 10.4% in El’ad and 7.0% in Betar Illit. In 2022 it ran on a joint list with Religious Zionism, so there is no separate figure to compare. Taken together, the two parties gained most in development towns, the small towns built for immigrants in the 1950s: 10.9 points in Kiryat Malakhi, 7.2 in Netivot.

The settlements lost votes below the threshold

The Netanyahu bloc fell furthest in some West Bank settlements, but not because their voters crossed over. Their votes went to small lists that will not enter the Knesset. Amcha Yisrael, Ofer Winter’s list, took 11.9% in Kfar Adumim and 9.1% in Efrat. In Eli, Amcha Yisrael and Noam together took 14.7%. Amcha Yisrael has 62,000 votes nationally, about 1.7%, short of the threshold. Those votes elected no one.

Druze towns went the other way

While the country moved away from Likud, Druze towns moved towards it. Likud rose 7 points in Daliyat al-Karmel and 6 in Hurfeish. Turnout rose too, by 22 points in Isifya, 19 in Mughar and 13 in Yirka, far more than in the Jewish towns around them. Kisra-Sumei shows even larger changes, but there it is 2022 that was unusual: turnout fell to 32% that year and National Unity came first. In 2026 the town voted much as it did in 2021.

In the mixed cities, two stories at once

Lod, Akko and Ramla, mixed cities with Jewish and Arab neighbourhoods, look at first like the Arab turnout story. In their Arab neighbourhoods turnout rose 16 to 21 points and the Arab lists gained 7 to 9. But the Netanyahu bloc also fell in the cities’ other neighbourhoods: by 12.4 points in Lod, 10.5 in Akko and 8.1 in Ramla, nearly two to nearly three times its fall elsewhere in the country. A city’s total mixes the two. Only the polling places separate them.

Elsewhere the change was more even. Likud fell most in Modi’in (6.5 points), Ariel (6.3) and Shoham (5.5), and held or gained in Kiryat Gat, Netanya and Ashdod. In the big cities the bloc as a whole fell by between 2.3 points (Netanya) and 7.3 (Rishon LeZion).

The new parties sit where the old ones did

bechirot.fyi draws a political map from the votes alone: places that vote alike sit close together, and each party sits where its voters are. Drawn from polling places rather than towns, the map puts Gadi Eisenkot’s Yashar! and Naftali Bennett’s Together almost on top of each other, where National Unity sat in 2022, and not where Yesh Atid did. Eisenkot was third on National Unity’s list in 2022.

How these numbers were made

Each comparison is a polling place against itself: the same place, in the same neighbourhood, in 2022 and in 2026. The committee renumbers its ballot boxes between elections, splitting and merging them, but the whole number of a ballot box (the 20 in 20.3) identifies the polling place, and that carries over. Matched that way, 3,813 polling places in 859 towns line up, holding 2.02 million of the votes counted so far.

As a check, I compared how far each matched place leaned away from the rest of its town in 2022 with how far it leaned in 2026. The correlation is 0.97 for the Haredi parties, 0.97 for the Arab parties and 0.91 for Likud. Pairing each place with the next number in the same town instead gives 0.19, 0.35 and 0.26, so the matches are the same neighbourhoods and not chance. Places that do not match, mostly new or renumbered ones, are left out, and so are the envelope ballots. The 431 places in the turnout chart are the polling places the Arab lists won in 2022. Station by Station counts something else, the 1,606 ballot boxes that had an Arab majority in 2022, so the two numbers do not compare.

How the night itself went, and how the live projection was built and rebuilt, is in Station by Station.