About
On the method, on the data, and on what can and cannot be concluded from them.
What the site does
KnessetLens shows what Members of Knesset and parties actually did: how they voted on bills in the plenum and what they said in committee discussions. Scores on three axes are calculated from this data, and the site shows them next to the data behind them.
The site does not urge anyone to vote, or not to vote, for any list, and does not rank parties or candidates against one another. The questionnaire on the site places you on the same three axes. Its result is information only, and is not a recommendation or a forecast.
How far the data goes
The data on the site stops at the dates in the table. The site does not yet include what happened after them.
| Data | Knesset terms | Data through | Score calculated on |
|---|---|---|---|
| Votes on bills in the plenum | 20th to 25th Knesset | ||
| Remarks by Members of Knesset in committee discussions | 25th Knesset |
The source of the votes, the bills and the committee transcripts is the Knesset's open data.
How the site measures
Each candidate's page and each list's page shows up to three scores on each axis: from votes, from committee remarks and combined (provisional). Here is how each one is determined.
Three axes
Every score is a number between minus 10 and plus 10 on one of three axes. The axes and the topics that make them up are based on the CHES-Israel expert survey, as analyzed by Zur & Bakker (2023). The site is not affiliated with those who conducted the survey and does not act on their behalf.
| Axis | Minus 10 | Plus 10 | Topics such as |
|---|---|---|---|
| The security axis | Universalist | Particularist | Relations with the Palestinians, a Palestinian state, settlements, immigration and asylum seekers, the Jewish and democratic character of the state, relations with the Arab world |
| The economic axis | Economic left | Economic right | Redistribution of wealth, government spending versus taxes |
| Conservative-Liberal | Liberal | Conservative | Abortion, gender equality and LGBT rights, environmental protection |
Bills
A language model (artificial intelligence) reads the text of the bill and rates it on every topic it touches. From these ratings, a score for the bill is calculated on each axis, together with the bill's degree of relevance to the axis and the level of confidence in the score. A bill with no available text is not rated, and a bill that touches no topic is not included in the calculation.
From votes
A Member of Knesset's votes on bills are turned into a score as follows:
- A vote for a bill counts as support for its direction on the axis, and a vote against counts as opposition to it.
- Only votes on the bill itself are counted, in the preliminary, first, second and third readings. Votes on reservations and procedural votes are not counted.
- Abstentions and absences are not counted. If someone voted for a bill in one vote and against it in another vote on the same bill, that bill is not counted in their score.
- The more relevant a bill is to the axis, and the higher the confidence in its score, the more weight it carries in the Member of Knesset's score.
From committee remarks
Separately, the language model reads what each Member of Knesset said in committee discussions and places that member on the three axes. Only substantive remarks in the discussion are counted, not procedural comments and interruptions. The score is an average across the member's sessions: the more speaking turns the member took in a session, and the higher the confidence in that session's score, the more weight the session carries.
Lists
A list's score is the average of the scores of its candidates who were measured, and a candidate who voted on more bills, or who spoke in committees in more speaking turns, carries more weight. A list's score may rest on a single candidate: each list's page shows how many candidates it is based on.
Combined (provisional)
The combined (provisional) score joins the score from votes and the score from committee remarks according to the amount of evidence behind each. It is provisional because the two scales have not been calibrated to each other, so it is also worth looking at each of them separately.
What cannot be concluded
- A language model gave the scores of the bills and of the remarks in committees, and it can make mistakes.
- A vote shows what was done, not why.
- A score is not a fact about a person's views and is not a forecast of what a party or candidate will do.
- A candidate who did not serve in the 25th Knesset is not measured and has no score, even if they served in earlier Knessets.
- Three axes are a simplification: not every position fits into them.
- The official election details, such as the candidate lists and the ballot letters, are published by the Central Elections Committee. Where there is a gap between them and the site, the Committee's information is the binding one.
Who runs the site
KnessetLens is run by Dave Ben Dor and Adam Simonov. To get in touch or report a mistake: knessetlens@gmail.com.
See also
Terms of use (section 11 explains how to report a mistake or ask for content to be removed), Privacy policy, Accessibility statement and Image credits.