Loading…
Loading…
Aggregates over the canonical record. Every figure carries the population it was measured over, and the outcome figures carry how many cases have no recorded outcome at all. There are no percentages taken over unknowns on this page.
A level filter necessarily excludes events with no recorded legislature: those have no level at all.
132
events
the unit is the event, not the person
23 of 132
meet the publication standard
94 of 132
carry at least one cited source
62 of 132
dated to the day
168
participation legs — the population every figure below is measured against
Not a count of anything that happened. One person in one event is one leg, and the split below says which act each one was.
130
distinct people
Two different things were being counted under one number. They are counted apart here, each against the same population, and they are never added together.
The unit here is the participation leg — one person, in one event. It is not the event count used elsewhere on this site, and the two must not be compared without converting first.
89 of 168
Carried to another party
A seat given to one party at an election, sitting under another party’s banner without an election in between. This is the floor-crossing figure, and it is the only figure that may be published under that name.
49 of 168
Left a caucus for the crossbench
The member left a caucus and sat as an independent. No party gained the seat. This figure has its own headline and is never added to the one beside it — adding the two is the single error this whole classification exists to prevent.
24 of 168
Joined a party from the crossbench
Counted toward neither headline above. The outbound leg of the same episode is already in the crossbench figure, so counting the return as well would count one person twice. Whether the voters were asked in between is the interesting question, and it is answered separately below.
These two figures are never summed. A seat carried to another party and a member who left a caucus for the crossbench are different acts, and a total that merges them answers no question anybody asked.
0 of 168
Structural — no seat moved
Mergers and renames, relabelling at a general election, an appointment being onboarded, a Senate group reshuffle, a crossbench designation changing while the member did not. Nobody carried a seat anywhere.
6 of 168
On a superseded duplicate record
Legs on a record superseded by another record holding the same person. Removed before the class is even considered, so a duplicate is a duplicate whatever it was classified as.
0 of 168
No ruling made yet
Legs on which nobody has made a ruling. Published as its own figure rather than folded into a larger one: a crossing count that does not say how many legs it could not classify is not a denominated figure.
0 of 168
Class the counting rule does not recognise
Legs carrying a class the counting rule does not know. Expected to be none; shown anyway, so that a vocabulary widened without widening the rule cannot hide inside another bucket.
The seven buckets above partition the participation record exactly: every leg sits in one of them, none sits in two, and they sum to the population each was measured against. That is checked as this page renders, so the figures audit each other rather than asking to be trusted.
A member who leaves a caucus, faces a general election, and then sits under a new banner has been given a mandate for it. One who does the same without an election in between has performed a crossing with a pause in the middle. The record can often show that an election fell in the gap; it can much more rarely show that none did, because an election missing from an incomplete series is indistinguishable from an election that never happened.
The record cannot sayunknown | 24 of 24 |
|---|
Two of the four verdicts are refusals to answer and must not be read as “no election”. Together they are usually the larger half of this table.
7 of 24 Of those, returned to the party they had left
A third case again, overlapping the four verdicts rather than sitting beside them: the seat went back under the banner the voters originally chose. Neither a crossing nor a defection. Never add this to the rows above.
The table below counts EVENTS by the kind of change each record says it is — the `event_type` column, which is a description of a record. The split above counts PARTICIPATION LEGS by which act each one was, ruled per person from the two affiliation labels that leg carries. The two are close enough in wording to be mistaken for each other and are not the same measurement: one row here can hold ten legs of three different classes. If you want the floor-crossing figure, take it from the split above; the row below labelled “changed party affiliation” is not it.
The second column is how many of those events are dated to the day. A decade where it is zero is a decade that arrived by bulk import.
| Jurisdiction | Events | Publishable | Span |
|---|---|---|---|
| British Columbia | 57 | 6 | 1952–2026 |
| Quebec | 22 | 13 | 1879–2026 |
| Alberta | 12 | 0 | 2004–2026 |
| Saskatchewan | 11 | 0 | 1925–2026 |
| Ontario | 9 | 1 | 1986–2025 |
| Newfoundland and Labrador | 7 | 0 | 1955–2024 |
| New Brunswick | 5 |
No rate is published for this measure. With so few cases recorded either way, a percentage would be a lie with a decimal point.
Measured over 168 participant legs.
Measured over 168 participant legs.
Literal domains recorded against an outcome. This is not a score and not a causal conclusion. The earlier site published a 0–10 “benefit score” with no method and no source behind it; that field is gone.
| 0 |
| 1866–2013 |
| Nova Scotia | 5 | 3 | 2011–2026 |
| Manitoba | 3 | 0 | 1950–2026 |
| Prince Edward Island | 1 | 0 | 2013–2013 |