Scores and ranks — the two kinds of ballot data¶
The single most consequential distinction in ballot design: a rank records order ("I prefer A to B"), a score records strength ("A is a 5, B is a 1"). Everything else in the library — why methods disagree, what converts cleanly and what doesn't — builds on getting this pair straight.
Start here → Scores vs. Ranks — Don't Confuse Ranks and Ratings
The pages¶
- Scores vs. Ranks — the core distinction: relative vs. absolute preference, and why it has real consequences.
- Preference vs. Support — the vivid special case:
1,0,1,0and5,4,5,4have the same preference but opposite support, and as ranks they're identical. Which do you like more, vs. how much you like each. - The Ranked Ballot · The Score Ballot — one anatomy page per ballot type.
- Strict ranks · Weak ranks · Strict vs. Weak Ranks — not all ranked ballots are the same: whether ties are allowed changes what the ballot can express.
- The Fidelity Ladder — converting between scores and ranks: which direction loses information, and how much.
- Scale granularity can flip the winner — a 301 case: the same opinions on a 0–5 vs. a wider scale can elect different candidates.
- What the ballot can and cannot say — expressiveness, measured — a 301 deep dive that counts what each paper can record and then varies the ballot and the counting rule independently: past six candidates the two ballots express disjoint sets of orderings, a real (capped) ranked ballot is the least expressive of all, and RCV-IRV can't spend the resolution either way.
- Grading as a rival primitive — Balinski & Laraki's stronger claim: not that ranks are lossy, but that "how good is this one?" is the more basic question and ranking was the wrong model to build a century of theory on.
- electowiki's "Cardinal voting," claim-checked — a 301 audit of the best short map of the score-ballot family: what to borrow, two logical errors to skip, and the article's own 51/49 argument run through the engine.
- ABIF — the all-in-one ballot format — the election-methods interchange format that writes ranks and scores in one grammar (
Allie/5 =Billy/5 >Candace/4), decoded and weighed honestly against this library's YAML grid.
Related¶
- Alternate ballot styles — one voter, three ballots — the same opinion marked on a ranking, Yes/No, and 0–5 ballot, side by side.