How the numbers work

The Petek does not conduct polls. It collects published seat polls from Israeli pollsters and the outlets that commission them: Midgam, Kantar, Direct Polls, Filber and others, published by N12, Kan 11, i24 News and Channel 14 among others. Each poll is recorded with its pollster, its publisher and its fieldwork date.

The public polling record is maintained collaboratively on Wikipedia’s opinion polling page for this election, which is a useful cross-check against what is shown here.

The average takes the ten most recent pollsby fieldwork date. For each party it takes the plain, unweighted mean of that party’s seat figure across those ten polls, and rounds to the nearest whole seat.

There is no house-effect correction, no recency weighting inside the ten, and no adjustment for sample size or pollster track record. It is a simple mean, chosen because it is easy to check rather than because it is the most sophisticated option available.

Parties that average below 4 seats, the electoral threshold, are shown as not clearing it rather than given a seat count.

A party missing from a poll counts as zero in that poll. Not every pollster prompts for every list. When a pollster omits a party, this average treats it as zero seats rather than excluding that poll from the party’s mean. For small or newly-formed lists that only some pollsters ask about, the figure shown here is therefore lower than the average of the polls that actually measured them. This is the largest known bias in the method and it is worth knowing before citing a small party’s number.

Ten polls is a window, not a trend. When polling is sparse, the ten most recent may stretch back further than the political situation justifies.

Seat projections are not vote shares. Pollsters convert their own vote estimates into seats using their own assumptions about the threshold and surplus-vote agreements. Averaging their seat outputs inherits all of those assumptions.

Polls in Israel have missed before, and a seat or two of movement is well inside normal error. Treat a 61-seat line as a question, not an answer.

New polls are pulled once a day. Pages across the site refresh at least hourly, so the figure you see is never more than an hour behind the stored data.

The current average is built from 10 polls, most recent fieldwork 27 July 2026. See every poll in the average.

Party platforms and candidate profiles are editorial: they are written here, from the public record, not reproduced from party material. Each profile links its sources at the bottom of the page.

The six-issue scale used on party and candidate pages is an editorial reading of stated positions, not a score derived from voting records. Where a candidate’s own position differs from their party’s, it is marked personal position so you can see the disagreement rather than have it averaged away.

Placement on that scale is a judgement call, and reasonable people would place some candidates differently. If you think something is wrong, tell us and it gets fixed.

The seat average and the profiles on this site are free to quote, cite and link, for any purpose including commercial reporting, with attribution to The Petek and a link to the page you took it from.

If you are writing on deadline and need something clarified, get in touch. Press queries are answered first.