Sign up free

The question with the wrong answer key: how to spot it

Four signs of a miskeyed quiz question, the innocent cause behind each, a five-step check, what exam boards do when a key is wrong, and how to tell the class.

The EdProcto editorial teamWritten and checked against the research it cites. Last checked . How the guides are made

Guides · Reading the results · · 13 min read

How do I tell if a quiz question was marked against the wrong answer key?

A question with the wrong answer key, a miskeyed question, usually shows four signs together: most of the class marked wrong, the strongest students wrong more often than the weakest, one wrong option chosen by more students than the key, and a negative discrimination figure. Each sign has an innocent cause too, so answer the question yourself, without the key, before changing a mark.

You have marked the quiz and one question looks wrong. Two thirds of the class missed it, and the students who missed it usually miss nothing. The likeliest cause is not a forgotten lesson: the question was marked against the wrong answer key.

A wrong key takes a mark from every student who knew the answer and hands one to some who did not.

EdProcto, which publishes this guide, prints "Check the key" in the full analysis of its results when the pattern appears.

What does a wrong answer key look like in the results?

A wrong answer key shows as four signs on one question, and the more of them appear together, the surer you can be.

  • Most of the class is marked wrong.
  • The strongest students are wrong more often than the weakest.
  • One "wrong" option drew more students than the key.
  • The discrimination figure is negative: the top scorers did worse on it than the bottom ones. See the discrimination index.

Each sign has an innocent cause

Sign With a wrong key The innocent cause How to tell
Most of the class marked wrong The class was right A topic not yet taught, or a hard question Untaught: answers spread over every option. Hard: the few right are your top scorers
Strong students wrong more than weak ones They agree on one other option An ambiguous stem The strong students split between two options
One wrong option ahead of the key Its takers outscored the key's on the whole quiz A shared misconception Its takers scored lower than the key's
Negative discrimination The right answer is marked wrong Two defensible answers, or a small class Both options draw strong students, or one student turns the figure

Texas State University (n.d.) gives its lecturers three readings of a negative figure: a question unlike the rest, an ambiguous one, or a mis-keyed one. Finding misconceptions and what the numbers mean cover the innocent causes.

A wrong key can leave no sign at all

The signs appear only when students disagree with the key. Wainer (1981) studied a mathematics question keyed incorrectly on the Preliminary Scholastic Aptitude Test and answered by almost 830,000 examinees. His abstract reports that detailed statistical analysis of even that sample would not have pointed to the error. Our reading: when the keyed answer is the one most students believe, only somebody working the question finds it.

How often do flawed and miskeyed questions turn up in real tests?

Flawed questions are common. A wrong key is rare, but reaches even papers that several people have checked.

  • Classroom papers. Tarrant et al. (2006) examined 2,770 multiple-choice questions set in one nursing department from 2001 to 2005: 46.2% broke an item-writing guideline.
  • Wrong keys, counted. Srisomsak et al. (2025) reviewed 14,238 questions from six years of medical school exams. Eighty-one were corrected after the sitting, 40.7% of them for a wrong key: 33 questions, about one in 430. On those, the truly right option drew more students than the keyed one and correlated better with the total. Where two answers were right, the correlations did not differ.
  • National examinations. Ofqual (2025), England's regulator, reports 98% of GCSE, AS and A level question papers free of error in summer 2025, and 93 errors in assessment materials: 72 in question papers, three in a mark scheme or answer key.

What one wrong key does to marks and grades

One question can move a grade. Ofqual (2026) fined one board £270,000 over 12 errors in physics papers and mark schemes sat by over 14,000 students. Correcting one mark scheme error after results day raised four of 13,278 A level entries by a grade; full marks for all on one faulty AS question raised 32 of 1,068.

A classroom quiz is shorter, so each question weighs more: on ten questions, a wrong key moves each student it touches by ten points. Studies that re-scored papers without their flawed questions are in when to throw a question out. Of Tarrant et al., Wainer, and Abozaid et al. below, we read the abstracts only.

How do you check a suspect key in ten minutes?

Work the question yourself before you touch a mark. The five steps take about ten minutes.

  1. Hold the marks (1 minute). If no score has gone out, keep it so. An EdProcto quiz with "Show the score" at "When released" sends students no score or mark until you press "Release the scores".
  2. Find the question (1 minute). In EdProcto's results, "Each question" gives "Got it", the share right, on every plan. Open any question under about 30%, and any your top scorers missed.
  3. Read who chose what (2 minutes). Which option leads, and how did its takers do on the whole quiz? On the Educator plan, press the question's row. On Free or Tutor, open your three highest scorers' attempts: each shows the option chosen.
  4. Answer it cold (3 minutes). Cover the key, work the question from the stem and write your answer down. Find the line in the textbook that settles it, and give a colleague the bare question.
  5. Name the fault (3 minutes). A slip of the key, two defensible answers, an unclear stem, an untaught topic, or none. Only the first is a wrong key.

What do examination bodies do when a key is found wrong after the sitting?

They do one of three things: correct the key and mark again, credit more than one answer or every answer, or take the question out. Boards call the check key verification.

Re-key and re-mark

The North American Board of Naturopathic Examiners (n.d.) describes the routine: a psychometrician flags questions whose statistics look faulty, subject experts read each, and the correct answer may be changed, more than one answer credited, or the scoring left alone. Texas State University (n.d.) lists possible mis-keyed questions on a lecturer's report and, on request, changes the key and processes the exam again.

Where the key was a slip and the question is sound, the mark belongs to the right answer; students who chose the old key bear the cost.

Credit more than one answer, or every answer

Ofqual (2025) says a board whose error may have confused students can widen the mark scheme to accept a range of plausible answers and, if fairness cannot be kept otherwise, award the marks to all or exclude the question. In summer 2025 boards amended a mark scheme for 28 question-paper errors and discounted the question for six.

Nobody loses a mark, which suits a fault in the question; the cost is a mark for students who knew nothing.

Drop the question

In Germany's state medical examinations, as Möltner (2018) describes them, a flawed question leaves the count for everyone except a candidate who answered it defensibly, for whom it still counts. The rule rests on a 1995 ruling of the Federal Administrative Court.

It suits a question with no right option, at the cost of a shorter test.

A working rule: re-mark a slip of the key, credit a fault in the question, drop a question with no right answer. Decide by reading the question, never by who gains.

A worked example: one wrong key in a class of twenty-four

This class is invented, and the figures are what EdProcto's analysis code returned for its answers. Twenty-four students took ten four-option questions on speed, distance and time, one point each, with a pass mark of 60%.

Question 6: a car travels 150 km in 2.5 hours, so what is its average speed? Building the quiz, the teacher selected 375 km/h as the key. Question 9 asked what an object is doing when its line on a distance-time graph is horizontal, keyed "It is not moving".

Q6 Q9
Got it 21% 29%
Separates −0.19 0.83
The keyed option 375 km/h: 5 students, mean 52% It is not moving: 7 students, mean 84.3%
The leading other option 60 km/h: 15 students, mean 70.7% It is moving at a steady speed: 14 students, mean 54.3%
The page's top line Check the key A shared wrong answer

Most of the class missed both, and there the likeness ends. On Q6 the fifteen who chose 60 km/h outscored the five on the key, and the index is negative. Worked cold, 150 ÷ 2.5 is 60: the key was 150 × 2.5, a slip. On Q9 the seven on the key are among the highest scorers, and the fourteen read the graph as a speed-time graph: that key is right.

Marked against 60 km/h, fifteen students gain a mark and five lose one. The mean rises from 60.8% to 65%. Fourteen passed before and fifteen after: two students move from 50% to 60%, and one who chose 375 km/h falls from 60% to 50%. Q6 now reads 63% right with an index of 0.92, and the reliability figure moves from 0.66 to 0.80.

How do you tell the class?

Tell the whole class at once, in three sentences, and call the mistake yours: which question, what was wrong, what was done.

"Question 6: the answer I had marked correct, 375 km/h, was my error. The correct answer is 60 km/h, and every paper has been marked again against it. Fifteen marks went up and five went down."

Speak first, in private, to anyone whose corrected score falls below the pass mark.

How do you stop a wrong key before the quiz?

Three habits catch a wrong key before a student sees the quiz.

  • A second reader. Abozaid et al. (2017) compared three final examinations at one medical college before and after formal peer review of questions began: discrimination rose by 0.14 and reliability by 0.12. Two year groups were compared, so other changes may share the credit. Give your colleague the questions without the key.
  • Your own quiz, answered cold. At Cambridge OCR, writes Malpass (2026), a "candidate proxy" attempts each paper as a candidate would and a marker marks that attempt with the draft mark scheme. Do the same: in EdProcto, press "Preview as student". With "Show the score" at "After submitting", the preview is marked against your key and stays out of the results. A question you miss is your slip or the key's.
  • A key fixed before anything is shuffled. A key written as letters breaks when options are reordered or a second version is printed. Texas State University's report flags students who may have been marked against the other version's key: its example scored 15 with one key and 44 with the other. In EdProcto you select the right option itself, under "Options · mark the right one", so "Shuffle the options" moves option and key together.

How does EdProcto show a wrong key and let you put it right?

EdProcto is the best tool for catching a wrong key when the scores go out tomorrow: its full analysis reads the four signs for you and says whether the pattern points at the key or at the class, as it did for Q6 and Q9 above.

Beside the question, in "What the numbers say":

  • "Most of the class missed Q6. Check the key first, then whether this was taught." Under 30% right, once three answers are marked.
  • "On Q6 the weaker students did better than the stronger ones. That usually means a wrong key or an ambiguous wording." The "Separates" figure is below zero; it needs six submissions.
  • "More students chose "60 km/h" than the right answer on Q6. Either the key is wrong or the class shares a misconception." A wrong option with at least three takers drew more than the key.

At the top, under "What it says about the teaching":

  • "Check the key on Q6: the stronger students got it wrong more often than the weaker ones." Every question with a negative figure.
  • The same opening ends "the students who chose the most popular wrong answer scored higher on the rest of the quiz than those who chose the keyed one" when the figure is not negative, but the takers of such an option have the higher mean score on the quiz, or lead on the rest of it by more than chance would give. Where nobody chose the key, the line ends "nobody chose the keyed answer."
  • Any other popular wrong option is filed under "A shared wrong answer on Q9", as a matter for the next lesson.

So the test is who chose the option, not how many. Pressing a multiple choice row opens "Option", "Chose it" and "Their mean", with the key tagged. Each sentence is a reason to reread the question, not a verdict.

These sentences, "Separates" and the option table are the full analysis, sent from the Educator plan up; Educator takes 500 students a quiz. Free takes thirty and Tutor 100: there the page gives "Got it" and "Median t" for each question, and every attempt shows the option chosen. See pricing.

Putting it right

A mark made by rule is made when the student hands in, against the key saved at that moment, and stands until you ask for the marks again.

  1. Correct the key in the builder. From that save on, every hand-in is marked against the right option, the answer shown to students is the corrected one, and "Duplicate this quiz" carries it to the next class.
  2. Press "Re-mark this quiz". On the results page, then "Yes, re-mark": every attempt handed in is marked again from the questions and keys as they stand, and the page says how many attempts have new marks. A mark you or Procto AI gave is kept. To credit every answer to a faulty question instead, open each attempt from "Students" and press "Add a note, or set the points by hand" under that answer: "Points", "Note for the student", "Save my grade".
  3. Tell the class. If the scores are still held back, release them once the marks are right. If students have seen them, each sees the new mark on opening the result again, so say what changed and why.

Quick questions

What is a miskeyed question?

A question marked against the wrong answer key: the option recorded as correct is not the right answer, so students who knew it are marked wrong.

Does negative discrimination always mean the key is wrong?

No. An ambiguous stem, two defensible answers or a question unlike the rest can give it too, and so can one student in a small class.

Should students who chose the old key lose the mark?

If that option is simply wrong, the mark belongs to the right answer. If the question gave them fair reason to choose it, credit both.

Does EdProcto re-mark a quiz when I correct the key?

When you ask it to. Every hand-in after the correction is marked against the new key, and "Re-mark this quiz" on the results page marks the earlier ones again.

References

The results

Difficulty, discrimination, the wrong option that beat the right one, reliability, and what to reteach, in plain sentences.

Basic on Free, full on Educator

What the results say

Read next