The data behind the seat projections

Every seat number in the media is the output of a calculation, not a quantity that was measured. Who counts in the base, what was done with the undecided, and which lists respondents were offered in the first place — all of these are decisions taken before the number reaches the screen. During an election period, and only then, the intermediate step is visible.

Section 16E of the Elections (Means of Propaganda) Law, 5719–1959 requires a polling institute to file with the Central Elections Committee the raw support percentage for every list — including lists below the electoral threshold, and without any adjustment to the manner of seat allocation. That duty is why this data exists at all. This archive publishes what was filed, and states what each filing does and does not contain.

Raw filed support, by list and by institute

One column per institute or channel, holding its most recent survey — dated by the day the survey was run, not the day it reached the Committee. There is no time window: an institute that has not polled lately is still shown, with the age of its survey beside it. Columns and rows come from the data, so a new institute, a new list or a merger appears on its own.

What we do and do not claim

Every figure names the page, table, row and cell it was read from. A value that cannot be traced to a cell is recorded as absent, never estimated, and anything computed here rather than read is labelled as derived.
Gaps are shown, not hidden. A missing particular describes a published document, never an institute breaching a duty — the Committee publishes filings as sent and does not check them, and annexes may not be visible in the public PDF.
Columns are not adjusted to agree and no average is published. Each filing constitutes its base differently, so two figures are not comparable until they share one — which is why every column states what its percentages are of.

To the Hebrew archive →