Google Forms cannot subtract a mark. Its quiz grading pays the points you set for a correct answer and nothing for a wrong one, with no field anywhere for a negative value - so on a Forms quiz a wrong answer and an unattempted question score exactly the same. Every negative-marking fix for Forms is therefore a re-grade that happens somewhere else, after the fact: a script, an add-on, or a formula in the response sheet. Forms itself changes, so verify the current behaviour before you build anything on top of it.
This is written for the person building the paper, not sitting it. For the wider picture of where Forms stops being an exam platform, see what Google Forms can and cannot do for an online exam; for the clock, adding a timer to Google Forms. A student asking whether to guess when the penalty is minus one wants the negative marking strategy guide instead.
Why the Score Floors at Zero
A Forms answer key is one quantity per question: what it is worth. Grading compares the response to the key and awards that amount or nothing. There is no second quantity for a wrong pick and no third for a blank, so the distinction a negatively marked paper is built on - a guess costs something, a skip costs nothing - cannot be expressed. The quiz is not failing to apply your penalty. It has nowhere to put it.
Workaround One: A Script on the Response Sheet
Forms writes every submission as a row in a linked spreadsheet. You attach a script that runs when a row arrives, compares the answers against a key you have typed into the script or a hidden sheet, applies your own marks for right and wrong, and writes a corrected total into a new column. It is the most flexible of the three, because you are writing the arithmetic yourself: a different penalty on a different question is one more line of code.
What it costs: you now maintain code, and the answer key lives in two places. Fix a key in Forms, forget the script, and the paper scores two different ways. The script also runs after the student has submitted and gone, so when it fails, it fails to an audience of nobody.
Workaround Two: A Third-Party Add-On
An add-on is the same script with somebody else's name on it and a settings screen instead of an editor. For a teacher who does not want to write code and does not want to build formula columns either, it is the least work to get running. Weigh what installing one means: you are granting a third party access to your students' responses, and its pricing, its continued existence and its ability to keep up with changes in Forms are all outside your control. The duplicated answer key does not go away either - it moves into the add-on's own configuration.
Workaround Three: A Formula in the Sheet, Afterwards
No code at all. In the response sheet you add a column per question that compares the answer against a key row and returns three states, not two: correct, wrong, and blank. Blank has to be its own case. A column that reads 1 for correct and 0 for everything else cannot tell a guess from a skip, so the total charges the penalty to the students who left it alone - the exact distinction the penalty was added to make. Then one more column multiplies the correct count by your positive value and subtracts the wrong count times your penalty. Every step is on screen and you can check a student's row by hand.
What it costs is honesty about when it happens: after the session, by you, once per paper. It is a marking exercise, not a scoring engine. Sensible for one unit test; a chore that recurs forever on a weekly test series.
The Four Costs They Share
All three re-grade outside Forms, and that single fact produces the same four problems no matter which one you pick.
- If scores are released on submit, the student sees the un-penalised number first. The corrected one arrives later and has to argue with the one they already told their friends.
- The answer key exists twice - once in the Forms quiz, once in your script, add-on or sheet - and the two drift the first time you fix a key.
- Varying the rule by section means conditional logic that knows which question numbers belong to which section, and that mapping breaks the moment you reorder a question.
- Nothing shows the penalty to the student before they answer. A marking scheme a candidate discovers from the score report is a marking scheme applied unfairly.
A score the student sees, and then has to be told was wrong, costs more trust than the penalty was ever worth to the paper.
The Other Approach: Marks That Live on the Question
The alternative is not a cleverer workaround. It is a scoring model with somewhere to put the penalty in the first place. On MockSetu a marking rule is three numbers, not one: what a right answer is worth, what a wrong one takes away, and what leaving it blank takes away. You type the penalties as positive numbers and the editor shows them back with a minus sign, and neither can exceed what the question is worth. Four presets cover the common schemes in a tap - plus one with no penalty, plus one with minus 0.25, plus two with minus 0.5, and plus four with minus one, which the panel labels JEE / NEET style.
Those numbers resolve down a chain: a question uses its own rule if it has one, otherwise its section's, otherwise the exam default. That is what lets one paper carry two rules without conditional logic, and no Forms workaround reproduces it without hard-coding question numbers. Totals are not floored at zero either - a student who guesses badly enough finishes below zero, and the result screen shows that figure in red. For the full tour of the panel, including part marks and rounding, read how to add negative marking to an online test.
A Worked Example: SSC MTS in Two Rules
SSC MTS is the paper that breaks every Forms workaround. It runs 90 questions for 270 marks, which divides to three marks a question, across two sessions of 45 minutes each. Session I is qualifying only and carries no negative marking; Session II counts towards merit and takes one mark off a wrong answer. A re-grade script can do this, but only by being taught which question numbers belong to which session. Where marks resolve per question, it is five steps and no arithmetic.
- Create the exam and two sections - Session I and Session II - with 45 minutes each.
- Set the exam default to three marks for right, one off for wrong, nothing for blank. That is the Session II rule, and the one a new question should inherit by accident.
- Override Session I to three for right, zero for wrong, zero for blank.
- Leave Session II with no override at all, so it keeps following the exam default - one fewer rule to maintain and nothing to re-check later.
- Write both rules into the instructions in plain words, and check the current bulletin for whether a candidate may return to a finished session before you set section switching to match. Know what that switch costs: left off, each section runs on its own clock and submitting it closes it; turned on, the whole paper runs on one clock and the 45-minute sectional limits stop being enforced.
What the Student Sees Before Pressing Start
Because the scheme is stored data rather than a note you typed, the instructions page builds the marking card from it: one scheme when the paper has one, a per-section breakdown when sections genuinely differ, and a line flagging that some questions carry their own rule. Above it sits a table of the paper - each section's name, question count and maximum marks - computed from the questions actually in the exam, so it stays right when the prose around it goes stale. Inside the paper each question carries its own plus-and-minus badge, which you can switch off on the exam default.
Two things to know before you publish. Marks are worked out the moment a paper is submitted and stored on that attempt, so correcting a wrong key changes what the next student scores, not what the ones before them scored - nothing re-scores a completed attempt. And publishing enforces exactly one rule about marks: a paper carrying a scheme on only part of itself is blocked, because a half-scored paper silently stops ranking by marks. A paper with no scheme anywhere goes through with a warning, since ranking everything by correct count is at least consistent.
What This Will Not Do for You
A penalty is not an integrity control, and nothing here pretends otherwise. There is no proctoring of any kind - no webcam, no lockdown browser, no tab-switch detection - no question shuffling and no cap on attempts. Published papers are public: anyone with the link may sit them, and there is no paid or private delivery to one batch. There is also no CSV or Excel export of results and no per-student report card, which is the one place a Forms response sheet is genuinely better. If your workflow ends in a spreadsheet you hand to a head of department, weigh that before you move anything.
When Google Forms Is Still the Right Answer
A teacher running one quiz a term does not need to migrate. If the paper is a reading check, a feedback form with a few graded questions, or a class poll, a penalty is not what it is missing, and the spreadsheet formula will cover the one time you want one. The case for moving is narrow: the paper rehearses an exam that charges for a wrong answer, the same scheme runs again next week, and the student needs to see the price of a guess before they take it.
If that is the paper you are building, build it where the rule belongs to the question. Everything a creator can set up here - sections with their own clocks, per-question marking, bilingual papers, a listing in the public library - sits behind one free account that takes no card. If the questions already exist as a PDF, the bulk route is the extraction prompt in the JSON upload guide; the importer leaves numeric, TITA and match-the-column questions for you to add by hand.