Proportional Representation: STV vs STAR-PR¶
One line: with multiple seats, a proportional method gives a coalition seats in proportion to its size — not all the seats to whoever leads first-choices. STV does this with ranked ballots + vote transfers; STAR-PR does it with scored ballots + ballot reweighting. They're different machinery, same goal.
→ Glossary: STV, Proportional STAR · Curriculum: 301.1 (proportional STAR) and 301.2 (STV) — this page is the bridge between the two rungs · Level: Voting 301 · STAR-PR methods explained: STAR_PR · STV's own door: 06_Other/STV
The shared idea: a quota¶
To win one of k seats you need a Droop quota = votes / (k+1). For 3 seats and 100 voters that's 25 — and a hand count uses the next whole ballot above it, 26, so that no four candidates can all reach it. Either way the answer to "how big is a safe share?" is just over 25%, which is why the video says so for three seats. Any candidate at quota is elected; surplus or last-place support is redistributed until the seats are filled. (The two forms of the quota are one vote apart and both standard — fork 1.)
The demo electorate (constructed)¶
100 voters, 7 candidates, two clusters:
- Progressive — Housing, Schools, Parks, Transit — 58% of first-choices
- Business — SmallBiz, BigBiz, TaxCuts — 42%
Proportionally, 58% / 42% over 3 seats ≈ 2 progressive + 1 business. (The "Better elections are possible" video gives only first-choice % and STV outcomes, not ballots — so this is an illustrative reconstruction, not a transcription.)
→ Run them: 03a_stv_3seats.yaml (ranked, STV) · 03b_star_pr_3seats.yaml (scored, STAR-PR)
Results — proportional methods agree; majoritarian doesn't¶
| Method | Type | Seats elected |
|---|---|---|
| STV | proportional (ranked) | Housing, Schools, SmallBiz |
| STAR-PR — SSS | proportional (score) | Housing, Schools, SmallBiz |
| STAR-PR — Allocated Score | proportional (score) | Housing, Schools, SmallBiz |
| STAR-PR — RRV | proportional (score) | Housing, Schools, SmallBiz |
| Bloc STAR | majoritarian (score) | Housing, Schools, Parks |
Two things to read off this (verified on the engine):
- STV and all three STAR-PR methods land on the identical slate — 2 progressive
- 1 business. Different mechanics (transfers vs reweighting), same proportional result. The business minority (42%) earns its one seat.
- Bloc STAR gives the third seat to Parks, not SmallBiz — the 58% progressive majority sweeps all three seats and the business cluster gets nothing. Bloc STAR is majoritarian by design; it's the multi-winner method to avoid when you want proportionality.
Also note the proportional twist that matches the video: BigBiz (17 first-choices) wins no seat, while Schools (15) does — breadth of support beats raw first-place count once seats are shared.
So which proportional method is best?¶
Notice that all four proportional methods above elect the identical slate. That's typical, not a fluke — and it raises the obvious question: if they agree, how would you ever rank them?
The honest answer is that there's no settled way to measure it. Single-winner methods have a standard yardstick — VSE for accuracy, PVSI for strategy resistance. Multi-winner has no accepted equivalent.
The most serious attempt is AVEC (Average Voter Effective Choice) by Jameson Quinn — the same person who built VSE. It multiplies two factors: AVE (how much of the electorate's voting power actually elected someone — votes wasted below a party threshold drag it down) and AVC (how much real choice voters had). Scored examples: Israel's April 2019 election ≈ 84% (91.5% AVE × 92% AVC — 8.5% of votes fell below the threshold and were simply discarded), versus single-seat plurality districts at roughly 44%.
But Quinn shelved it: he marked the work "[not ongoing]," left several sections unwritten, and conceded it stacks "5 different levels/types of approximation" on top of a one-dimensional-ideology assumption. Treat AVEC as a serious sketch of the right question, not a number to quote.
What follows for this library. When someone asks "is STAR-PR better than STV?", the defensible comparisons are mechanical and administrative — summability, ballot expressiveness (scores capture strength, ranks capture order), how easily a voter can check the count, and whether the jurisdiction can run it — not a satisfaction score. Anyone quoting a precise "proportional accuracy" figure is ahead of the evidence. That applies to STAR-PR exactly as much as to STV.
STV is not IRV¶
STV = the proportional, multi-winner cousin of IRV. Same ranked ballot and transfer idea, but it fills several seats against a quota instead of one. Reserve "IRV" for the single-winner case. (See Tips — Terminology: RCV vs IRV vs RCV-IRV (and friends).)
Engine notes¶
- STV runs on the vendored
pyrankvote(single_transferable_vote); the RCV-IRV wrapper now dispatches to it whenevernum_winners > 1. No new library was needed — setvoting_method: STVandnum_winners: k. - STAR-PR runs on the STAR engine:
voting_method: sss | allocated | rrv(proportional) orbloc(majoritarian contrast), withnum_winners: k. Each method is explained in STAR_PR.
Source: Equal Vote — "Better elections are possible. Here's how." (video).