Quest Ladder
TeachFORGE.ai
Quest Ladder / Answer policy
Answer policy: how an answer is accepted
The rules that decide whether what a learner typed counts as right.
Most unfair marks come from this tab, not from the question.
What it governs
The Answer policy tab is headed Correct answer & acceptance. It sets how answers are graded for every instance of the type, and an individual question can still override it where it needs to.
This is the tab to check first whenever a learner insists they were right. Nine times in ten they were, and a setting here was stricter than you meant.
Answer value type
Answer value type says what kind of thing the answer is. Left at Infer from format preferences it follows what you chose elsewhere, which is usually correct.
Worth knowing before you go looking: this dropdown lists the raw internal identifiers rather than the friendly names, so you will see values like numeric_with_units where the format-preferences selector says Number with units. They are the same list.
Match mode and evaluation mode
Two settings, easily confused. Text match mode is the broad choice:
- Lenient (case/space-insensitive + algebraic equivalence) — the sensible default for almost everything.
- Strict (exact match, case-sensitive) — only when the exact characters are the point.
Evaluation mode is the finer choice underneath it: Case-insensitive text match, Exact match (case-sensitive), Numeric equivalence or Symbolic / algebraic equivalence. Alongside sits a tickbox reading Treat algebraically equivalent forms as equal, with the example x+2 and 1x+2.
Choose the mode that matches what the question is testing. Numeric equivalence on a question about notation will accept the thing you were testing for; exact matching on a question about a concept will fail a learner over a capital letter.
Tolerance and range
Numeric tolerance with Tolerance type set to Absolute or Percentage accepts answers near the correct value. Acceptable range — minimum and maximum accept a band instead.
The two interact in one specific way, and the panel says so: if a range is set it is used for grading, and the tolerance around the correct value is ignored. Setting both and expecting the tolerance to apply is a common and invisible mistake.
Alternatives and keywords
Other acceptable answers (one per line) is the list of forms you will also accept. For algebra, equivalent forms are accepted anyway when symbolic evaluation is on, so the list is for the cases the engine cannot infer: a synonym, a common abbreviation, an alternative spelling.
Optional keywords (any one may count) is looser again: a match on any one of them counts. It suits a short answer where several different words would each show the understanding.
Every field here has a tooltip in the editor. Where this article and the tooltip disagree, the tooltip is describing the build you are looking at.