Auto-grade short answers — a missing letter is not a wrong answer.
Most quiz tools compare a written answer to a string and mark anything else zero. Quizoma lets you auto-grade short answers with five ways to decide what "correct" means for each question — or turn the marking over to yourself entirely.
The problem this solves
A student who knows the answer types Pythagoras as Pythagorus.
A trainee writes seven where the answer key says 7. Someone gives a
perfectly good three-sentence explanation that happens not to contain the exact phrase you
wrote down at 11pm the night before.
In a string-comparison quiz tool, all three score zero. Then you spend the evening un-doing the marking, which is worse than not having auto-marking at all — you now have to check every answer and distrust the ones it got right.
The five matching rules
Each written question carries its own rule, and typo tolerance is the one it starts with. You can also give a question several rules at once, and the answer is accepted if any of them is satisfied. Fill-in-the-blank questions work differently — see below.
1. Exact match
Letter for letter, after case and punctuation are normalised away. Use it when the answer is a single unambiguous token and you want no interpretation at all.
2. Case-sensitive exact
The strict one: capitalisation and punctuation are both preserved and both must
match. This is the rule for a spelling test, a code identifier, or anything where
don't and dont are genuinely different answers.
3. Smart match (typo-tolerant)
Compares the answer word by word and scores how close it is, then accepts it if the similarity clears a threshold you set on that question. Loosen it on a history quiz where the spelling of a name is not the point; tighten it where it is.
Below the accept threshold but still recognisably in the region of right, the answer is recorded as a partial match. What that is worth depends on the question's scoring setting: all or nothing gives it zero, half for close gives it half the marks, and by similarity scales the marks to how close it was. All or nothing is the default, so a near answer scores zero until you pick one of the other two.
4. Keyword coverage
For answers that are explanations rather than facts. You list the ideas that have to be present, and set how many of them are enough:
- All — every keyword must appear.
- Most — 75% of them.
- Half — 50%.
A participant who covers the required share is accepted; one who covers noticeably less is recorded as a partial match with the coverage percentage attached, so when you open the result you can see at a glance what they missed.
5. Numeric equivalence
Numbers are compared as numbers, not as text. Digits, number words and different decimal spellings of the same value are all the same answer.
Accepted answers and aliases
Any question can carry a list of acceptable answers rather than one. H2O and
water can both be right without you marking forty scripts by hand. On fill in
the blanks, each blank keeps its own list, so a sentence with three gaps is
really three small questions with three sets of accepted answers — see the
fill in the blanks quiz maker for a
worked example.
Partial credit
Where it makes sense, a near answer earns a proportion of the marks instead of all or nothing. Multiple-correct questions score by how many of the right options were found. On written answers it is the question's scoring setting that decides — all or nothing, half for close, or scaled by similarity — and all or nothing is what you get until you change it, so some questions deserve partial credit and some genuinely don't. Fill in the blanks is the exception: every blank has to be right for the question to score.
When you would rather mark it yourself
Manual marking is a setting on the quiz, not something that happens on its own. Switch it on and nothing in that quiz is graded automatically — every answer, in every question type except polls, waits for you to read and score it. Leave it off and the rules above decide, including deciding that an answer nothing matched is wrong.
So the choice is yours per quiz. On an auto-graded quiz, objective questions — multiple choice, true/false, multiple-correct — are marked instantly, and written answers are marked against their rule with the similarity or coverage score recorded next to them. You can still open any submission afterwards and change a mark or leave a comment.
Same results in the app and the browser
The grading engine runs in both the Android app and the web quiz-taking page, and the two implementations are kept in lockstep and tested against each other. A written answer scores identically whether it was typed in the app or at quizoma.com/take. Nobody gets a different mark for using a different device.