The six mistakes teachers make most often when setting question papers, and a practical fix for each one, from mark distribution to MCQ clues to final review.
Faria Haque
6 min read
Setting a question paper is the least discussed part of a teacher's job and one of the most consequential. A good paper tells you what your students actually learned. A weak paper tells you who memorised the most, or who guessed the pattern best. The difference comes down to a handful of small decisions, and most of them get made in a hurry the night before.
The six mistakes below show up repeatedly in school, college and coaching papers alike. Each one comes with a fix you can apply to the next test you set.
The most common mistake happens before the first question is written. The teacher opens the syllabus and starts writing from whichever chapters come to mind. The result is predictable. The chapter taught most recently, or taught best, takes far more marks than its share, while a chapter that got fewer classes may not appear at all.
Students read that signal correctly. They learn which chapters pay and which ones can be skipped.
The fix is old and well tested. Before writing a single question, build a table of specifications, sometimes called a test blueprint. Chapters down one side, question types across the top, marks in the cells. Assessment research treats this table as the simplest way to secure content validity, because it forces the paper to represent the syllabus rather than the teacher's memory of it.
It does not need to be elaborate. A one page table is enough, and once built it carries over to next year.
The creative question format separates knowledge, comprehension, application and higher order skills for a reason. In practice many papers keep the shape of those four levels without the substance. When the stimulus is copied almost word for word from a textbook example, the application question is a recall question wearing a costume.
There is a quick way to check. Go through your paper and mark every question where the answer could be lifted directly from the book or a set of notes. If most questions get a mark, the paper is measuring memory, not understanding.
The fix is to change the context. Same concept, new situation. The physics principle stays the same but the scenario is one the student has not seen. The accounting rule stays the same but the numbers and the business are new. Assessment guidelines call this using novel material, and it is the main condition for testing anything above recall.
Students frequently get MCQs right without knowing the content, because the structure of the item gives the answer away. This damages the paper directly, since it narrows the gap between the student who studied and the student who did not.
The clues that appear most often:
The correct option is longer and better phrased than the rest. Teachers write the correct answer carefully and the distractors in a rush, and students notice.
The wrong options are so implausible that eliminating them requires no knowledge at all.
The correct answer sits in the same position repeatedly, usually B or C.
Options like "all of the above" and "none of the above" are used. Assessment research advises against both, because they change item difficulty without measuring knowledge.
The fixes are straightforward. Keep the options similar in length and phrasing, make every distractor plausible, and build distractors out of the errors students actually make. When the paper is done, count how often each position holds the correct answer.
One finding runs against common practice. Research suggests three well written options are sufficient in most cases, because writing three genuinely plausible distractors is difficult in the first place. Public examination formats will fix the number at four, but class tests and batch exams have room to move.
Questions like "which of the following is not correct" are not bad in themselves. The problem starts when the negative sits quietly inside the sentence. A student reading at speed misses the "not" and answers wrongly despite knowing the material. At that point the question is measuring reading care, not subject knowledge.
Standard item writing guidance is specific here. If a negative must be used, put it in capitals or bold so the eye cannot skip it.
Length creates the same problem. Unnecessary preambles, information repeated twice, a three line stimulus where one line would do, all of it raises the reading load. The student who knows the subject but reads slowly is the one penalised. Keep questions short and put the central idea in the question itself rather than in the options.
Trick questions, meaning questions designed mainly to catch students out, should be removed. They measure nothing about learning.
Almost every institution has at least one exam where half the students never reach the last question. That is not a difficulty problem. It is an arithmetic problem.
The simple rule is to sit down and answer your own paper in full before finalising it, with a clock running. A teacher writing at full speed takes roughly a third of the time an average student needs. If the paper takes you 25 minutes, it fits a 90 minute exam. If it takes you 40, cut something.
This habit does a second job. Solving your own paper exposes ambiguous wording, wrong figures and missing information immediately, while there is still time to fix them.
The last mistake is the most expensive. The paper gets written the night before, printed in the morning, and the error surfaces in the exam hall. Correcting it then means an announcement, lost time, and some students never hearing it.
Research into examination errors points at process rather than individual competence. Errors cluster around time pressure and around papers that only one person ever read. That means process is also where the fix lives.
Three habits work:
Leave at least a day between writing the paper and reading it again. Read your own writing while it is fresh and your brain supplies what you meant, not what you wrote.
Have a colleague read the paper once. They do not need to teach the subject, because spotting unclear wording does not require subject knowledge.
Write the marking scheme at the same time as the paper, not afterwards. If you have not decided what earns each mark before the exam, your marking will drift across scripts.
Does every chapter get the marks the blueprint assigned it.
How many answers can be lifted straight from the textbook.
Are MCQ options similar in length and phrasing, and is the correct position varied.
Are negatives bolded.
Have you solved the paper yourself against a clock.
Was the marking scheme written alongside the paper.
Has at least one colleague read the whole thing.
Clearing five of these seven puts your paper above the average standard at most institutions.
Haladyna, T. M., Downing, S. M., and Rodriguez, M. C. (2002). A Review of Multiple-Choice Item-Writing Guidelines for Classroom Assessment. Applied Measurement in Education, 15(3), 309-334. https://eric.ed.gov/?id=EJ660246
Fives, H., and DiDonato-Barnes, N. (2013). Classroom Test Construction: The Power of a Table of Specifications. Practical Assessment, Research, and Evaluation, 18(3). https://doi.org/10.7275/cztt-7109
Johnson, M., and Mehta, S. (2024). What gives rise to errors in examination papers. Oxford Review of Education. https://www.tandfonline.com/doi/full/10.1080/03054985.2024.2308548
Schreyer Institute for Teaching Excellence, Penn State University. Multiple Choice Item Writing Rules. https://www.schreyerinstitute.psu.edu/pdf/Multiple_Choice_Item_Writing_Rules.pdf
Google Meet or Zoom for online classes in Bangladesh, compared on free tier time limits, recording, breakout rooms, data cost and joining friction, with official sources.
Faria Haque
4 min read
A four step structure for talking to parents about student progress, what belongs in a progress report, and the measured effect of one sentence messages home.
Faria Haque
5 min read
Take a 15-minute demo with your own batches, attendance, and fee figures.