Counts from the public record

What these are for

Every set here comes from a document anyone can open: the verbatim record of proceedings, the government's own clarifications site, a town council's annual report. Nothing is surveyed, modelled or estimated.

What the site adds is the tally and the rule. The rule is the part people skip, and it is the part that decides whether two figures can be compared at all — oral questions against all questions, actions against falsehoods, a financial year against a calendar year. Each page states its rule in the table rather than in a footnote, because a figure quoted without its rule travels further than the correction ever does.

When a count is made again

Every page here carries the day the count was made, and so does the first line of every file. That date is not decoration: the records these come from are added to. Hansard gains a sitting, the clarifications record gains an action, a council publishes another annual report. A figure read in one year and a figure read in the next are different readings of a growing document, and a citation that cannot say which reading it used cannot be checked.

A set is counted again when its source publishes again, and the date on the page moves with it. Where a recount changes a figure, the page says what changed rather than replacing the number quietly — a corrected total that arrives without a note is indistinguishable from a total that was always wrong.

Reuse

The figures may be reused with attribution. Each page carries a citation line naming the original source, the window, the day of the count and this page, and the CSV repeats the same rows so nothing is lost between the table and the file.

Where a row-level dataset exists it lives at the original source, and the page says so. This site republishes what it counted; it does not reconstruct records it has not checked line by line.

Why the rule matters more than the number

Three of the most-quoted figures about Singapore's Parliament are wrong in the same way: not because the arithmetic is bad, but because the rule behind them travelled without them. A count of oral questions is quoted as a count of questions. A count of enforcement actions is quoted as a count of falsehoods. A March-to-March financial total is set beside a calendar year and the gap of one quarter goes unmentioned.

None of those is a mistake anyone made deliberately. They happen because a number is easy to carry and a rule is not, and by the time the figure reaches its third citation the qualifier has fallen off. The only defence is to publish the rule where the number is, in the table rather than in a note underneath, and to repeat it in the file so that it survives the download.

What would be added next

The useful sets are the ones a record supports and nobody has tallied: divisions adjudicated without a contest at each election, the composition of the chamber by class of member across parliaments, the interval between a boundaries report and the dissolution that followed it. Each is a matter of reading published documents in order and writing down what they say.

What would not be added is an estimate. A figure that requires modelling, weighting or a survey belongs to somebody else's method, and mixing it in here would cost the whole set the one property that makes it worth citing.