Methodology
Where the data comes from, how the match is calculated and how to challenge a position.
What is here at the moment
- parties
- 17
- questions
- 14
- positions
- 238/238
- with a source link
- 85%
Of 238 positions, 59 (25%) are marked “position not stated”. This is not a gap in our work but the state of the public record: many parties have simply said nothing on some of the questions, and some parties published no manifesto at all. By confidence: high — 70, medium — 68, low — 100. Data updated 27 July 2026.
How the match is calculated
No machine learning and no hidden coefficients. The whole of the arithmetic fits into four steps, and you can redo the result on paper.
1. The party's position becomes a number
In favour = +1, tends in favour = +0.5, neutral = 0, tends against = −0.5, against = −1.
2. Your answer, on the same scale
Agree = +1, tend to agree = +0.5, don't know = 0, tend to disagree = −0.5, disagree = −1.
3. The match on a question
1 − |your number − the party's number| / 2. A full match gives 1, complete opposition gives 0, and the intermediate values are 0.25, 0.5 and 0.75. If you marked the question as important, it enters the calculation with a weight of 2 instead of 1.
4. The total for a party
The sum of the matches multiplied by their weights, divided by the sum of the weights. The result is converted into a percentage and rounded to a whole number.
What is excluded from the calculation, and why
This is the most important part of the methodology, and it is exactly the part that is usually hidden.
The party has stated no position
The question is removed from the denominator for that party only. Different parties therefore have different numbers of questions counted, and we show that number on every card. We do not substitute “neutral” for an unknown position: that would look like moderation where in fact there is silence.
You skipped a question
The question is excluded for every party at once — otherwise a skip would quietly penalise the parties that do have a position on it.
This has an awkward consequence, which we decided not to hide. A party that has said nothing on ten questions out of twelve gets a percentage based on two — and if it happens to match you on those, it can end up above a party that matched you on eleven. The arithmetic here is formally correct; the conclusion is false.
That is why such parties do not enter the main ranking. If a party's positions cover less than 60% of the questions you answered — or fewer than 4 questions in absolute terms — it is moved to a separate list below the main one, with an explanation of what is missing.
The threshold is a proportion rather than a fixed number, and that matters. If you answered every question and a party has spoken on four, its percentage is calculated from a third of the data. But if you answered only five yourself, those same four are already eighty per cent, and there is nothing to object to.
We do not remove such parties from the results: deciding what to show you and what not to show you is not our job. But nor can they be mixed into the ranking — their place in it would be a matter of chance. Where percentages are equal, the party with more data is placed higher.
Where the positions come from
In descending order of reliability: official party manifestos, recorded votes in the Knesset, public statements by leaders, media reports. Every position carries a source type, a link and a date of verification — all of it visible if you open the “why this percentage” panel on the result page.
- Confidence: high
- The position is set out directly in the party's manifesto, in a recorded Knesset vote or in an unambiguous statement by its leader. A source link is mandatory — without one the site build fails.
- Confidence: medium
- The position is supported by media reports or by statements from individual members of the Knesset, but not by a programmatic document.
- Confidence: low
- The position is inferred indirectly: from related votes, from previous Knesset terms, from statements open to more than one reading. We show such positions, but we label them.
What we deliberately do not do
- We do not show only the first three parties. The list is complete, including the parties we disagree with and those that fall below the electoral threshold.
- We do not infer a position from ideological logic. If a party has said nothing, we write that it has said nothing, even when it “goes without saying”.
- We do not tune question weights in anyone's favour. All 14 questions count equally, until you mark one of them as important yourself.
- We do not collect data about you. One anonymised record of the answers goes from the result page to us — no IP address, no timestamp, nothing that points to you. Subscriptions are stored separately and are not linked to answers in any way. In more detail, and without the evasive wording.
How to challenge a position
If you believe we have got a party's position wrong, write to us and attach a link to a source. We will correct it and give the date of the correction. Every position is stored as a separate record with its source, source type and date of verification, and edits go through version control. If you need the change history for a particular position, ask and we will send it.
Who is doing this
This survey was made by Naum Kogan and Mikhail Pellivert. At the time of publication the project is not affiliated with any party and is funded by no one. The Israeli Elections (Modes of Propaganda) Law, 1959 requires campaign material to state on whose behalf it is distributed. If an affiliation arises, it will be visible in the footer of every page — including this one.
Limits of what this can tell you
Party lists are submitted before the beginning of September, and the configuration of the parties changes until then: mergers, break-ups and new lists are all possible. Parties with the status “merger not completed” or “participation in doubt” are marked on the result card itself. Twelve questions do not cover all of politics: the economy, healthcare, housing and education are outside the quiz — we were unable to gather comparable data on them for all 17 parties. The quiz is a way to work things out, not a recommendation on who to vote for.
Questions in the quiz: 14. Take the quiz