How it works
The quiz asks you 30 statements and tells you how closely each of the eight biggest parties matches your answers. This page explains who is covered, where the statements and the party positions come from, how the match is worked out, and what happens to your answers.
Edition: 2026 local government elections. Dataset version 0.3-draft, 5 October 2026. Figures as at October 2026.
Which parties are covered
Coverage follows a written rule, not editorial choice, so anyone can check it. A party is covered in this edition if it meets any one of these tests on the day the dataset is frozen:
- National vote test. It won at least 1.0% of the national ballot in the most recent national election (29 May 2024, IEC final results).
- Metro test. It holds the executive mayor's office in any of the eight metropolitan municipalities, or it won at least 5.0% of the proportional vote in any metro at the most recent local election (1 November 2021).
- Polling test. Its average support across all published national polls in the six months before the freeze date is at least 3.0%.
Eight parties pass, all on the first test: ANC (40.18% of the 2024 national ballot), DA (21.81%), MK Party (14.58%), EFF (9.52%), IFP (3.85%), Patriotic Alliance (2.06%), Freedom Front Plus (1.36%) and ActionSA (1.20%). The closest party outside the rule is the ACDP at 0.60%. Independent candidates are not covered in this edition because they contest single wards and have no party-wide positions to code. Vote shares are the IEC's final figures at results.elections.org.za. A party that is in the edition is not removed mid-edition; the rule is re-run for each election.
How the statements were chosen
About 80 statements were drafted across the issues that decide local votes (water, electricity, housing and land, safety, jobs, councils and government) and the national questions that define the parties. The 30 kept are the ones the parties actually split on; a statement everyone agrees with tells you nothing. Every statement follows the same rules:
- One issue per statement, in plain language, with no cost, benefit or "even if" clause inside it. The trade-off sits in the explainer, which ends with one line giving each side's case at equal strength.
- Under every statement you see what agreeing and what disagreeing mean, so nobody answers the wrong way round.
- Statements are worded positively where possible, and the same term is used for the same thing everywhere.
- At least one covered party sits on each side of every statement.
- Roughly as many statements have "agree" as the state-led answer as have it as the market-led answer, and the share of each party's positions on the agree side sits between 45% and 55%. This stops the quiz from favouring any party simply because people agree with statements more often than they disagree.
The wording went through three independent review rounds (plain readers, a hostile reading from each party's point of view, and a check of every coded position) before this version. A pilot with real voters follows.
Where the party positions come from
Each party's position on each statement is coded as agree, in between (a mixed or conditional position) or disagree, from the party's own published documents: its 2026 local government manifesto, its 2024 national manifesto, policy documents and conference resolutions, recorded parliamentary votes, and dated public statements by its leader, MPs or official spokespersons. Every position shown on the site carries its source: the document, the page or section, and the date.
A position is shown only when it was read in the party's own document, or comes from a dated statement by a leader, MP or official spokesperson that answers the statement directly. A remark on a neighbouring topic, practice under the current law, or a vote on a whole Act with no statement on the clause is not enough; those readings stay in the research files and the site shows "no published position found". Where a party's documents say nothing, the site says so rather than guessing, and that statement is left out of your match with that party.
The same rules were applied to every party at the same depth. Parties were not asked to confirm their placements; what you see is what the documents say. If you think a position is wrong, send the document that shows it and it will be checked and corrected: corrections address to be published with the site's domain.
How the match is scored
For each statement you answered, a party gets 2 points if its position is the same as yours, 1 point if it is one step away (for example you agree and the party is neutral), and 0 if it is the opposite. After the 30 statements you can mark the ones that matter most to you; those count double. Your match with a party is its points as a share of the maximum it could have scored on the statements you both have a position on. Statements you skipped, and statements where a party has no published position, are left out for that party, which is why each result says "based on N of 30 statements". All eight parties are always shown, ranked.
For the statements you marked as mattering most, each party also shows where its position differs from yours, or where it has no published position, so you can see the conflicts without any party disappearing from the list.
If you tell the quiz which party you expect to match, it shows you how that compared with your result. The answer is not used in the scoring.
The dials and the map
Five dials summarise groups of statements: economy (state-led to market-led), services (pay for use to free and relief), redistribution (less to more), order and borders (less strict to stricter enforcement), and where power sits (national and concentrated to local and dispersed). Your position on a dial is the average of your answers on its statements, each counted in the direction the dial runs; a party's position is the same average over its shown positions. A dial needs at least two statements to show a position. A sixth dial, social values, is in the data but is not shown in this edition because it rests on too few statements. The map puts two dials together; you can change the up-down dial. The dial names describe the two ends of a scale, not a label for any party.
Your answers and your privacy
Everything is worked out on your phone or computer. Your answers are never stored or sent anywhere. The result link you can share contains your answers so that the result can be rebuilt by whoever opens it; share it only if you are happy with that. The site uses no cookies. Site analytics, when enabled, count page views and share-button clicks without identifying anyone (Cloudflare Web Analytics, which is cookieless). The only personal data the site will ever hold is an email address you choose to give for updates, once that signup exists.
Open data
The full set of shown positions, with justifications and sources, is downloadable as CSV or JSON. Proposed licence: Creative Commons Attribution 4.0. Positions that rest on a weak reading appear in the download as "no published position found".
Figures in the statements
Every rand amount and date is as at October 2026 and is checked each year: minimum wage R30.23 an hour (from March 2026); SRD grant R370 a month (unchanged in the 2026 Budget); top income tax bracket 45% from R1,878,601 (2026/27); free-house income limit R3,500 a month and First Home Finance band to R22,000 a month; NSFAS household income limit R350,000 a year; first planned coal station closures from about 2030; 1% council threshold in the government's bill.
Who built this
Which Party is an independent, non-partisan site. Its only interest is that the quiz is fair to every party, which is why every rule above is written down and every position carries a source.