Online exams: how to prevent cheating and keep them fair

Online exams have moved from an emergency measure to a normal part of assessment in schools and universities across the Gulf and the wider Arab world: weekly quizzes, midterms and, sometimes, full final exams. With them comes the question every principal and teacher asks: how do we know a grade reflects what the student actually knows?

The short answer is that no single tool prevents cheating. It takes a set of small decisions: how you write questions, how you configure each attempt, what you monitor, who reviews results before they count, and how you explain the rules to students and parents. This guide walks through each decision and is honest about what technology can and cannot do.

Match the protection to the stakes

A practice quiz does not need the protection of a final exam that feeds into a student’s GPA. Treat every test with the same strictness and you wear out students and teachers for nothing, and the rules lose their meaning.

Agree these levels with heads of department before term and write them into your assessment policy, so every teacher knows which settings to use and sections do not each invent their own rules.

Question banks: your first line of defence

The most common cheating in online exams is simple: a student finishes and posts the questions in the class WhatsApp group. The remedy is for each student to see a different version, and that starts with a question bank larger than the exam.

• Write more questions than you need. For a 20-question exam, aim for three or more equivalent questions per learning objective, so the platform can draw a different one for each student if it supports random selection.
• Tag every question by unit, learning objective and difficulty. Tagging keeps different versions equally hard, so no student gets an easier paper than the person next to them.
• Change the numbers, not the idea. In maths and physics, one question with different values gives you several equivalent versions.
• Review the bank every year. Remove questions whose answers have circulated and questions that everyone answers correctly, because they no longer tell students apart.

Building the bank as a subject team saves time and keeps exams consistent across sections, which parents notice when they compare results.

Write questions that are hard to look up

A question that can be answered by pasting it into a search engine or an AI tool is weak in an unsupervised exam, whatever protection surrounds it. Keep objective questions, but balance them with questions that ask for application or explanation.

• Ask students to apply a rule to a new case rather than define it.
• Use data, texts or diagrams you presented in class that do not exist elsewhere.
• Add a short question asking the student to explain why they chose an earlier answer.
• In open-book exams, design questions on the assumption that students will use the book, and assess understanding rather than recall.

These may sound like teaching tips rather than technology, but they are the strongest integrity tools you have. Nothing reduces cheating like a question where cheating does not help much.

Random order and time limits

Next come the attempt settings. Most platforms make them easy to apply; the value is in choosing them well.

• Random order for questions and options, so two students sitting side by side cannot swap answers by question number.
• A time limit that is calculated, not guessed. Give enough time for a prepared student to answer and check, but not enough to look up every answer. Take the exam yourself, start with roughly double your time and adjust after the first sitting.
• One window for everyone. Open the exam for all sections at the same time where you can, or use a different version for each section if the times differ.
• A clear number of attempts: one for formal exams, several for practice.
• The right moment for results. Show correct answers after the exam has closed for everyone, not as soon as the first student submits.

Showing one question per page with no going back reduces answer sharing, but it raises anxiety for some students and takes away the chance to review. Use it for short quizzes and avoid it for long exams.

Integrity signals: what they measure and what they do not

Many platforms record browser events during an exam, such as a student leaving the exam page for another tab, copying and pasting, or the window losing focus. Some combine these events into an integrity score for each attempt and highlight the attempts worth reviewing.

These signals are useful because they point a teacher to a handful of attempts instead of all of them. But they are a signal, not a verdict. A student whose connection dropped, who switched to a message on their phone or who opened the calculator can look, in the data, just like a student searching for answers.

Be realistic, too. A browser tool cannot see a second phone on the desk or a friend in the room. Webcam proctoring adds protection but raises privacy questions, needs clear consent under data protection laws such as those in Saudi Arabia and the UAE, and depends on a strong connection that not every family has. For most unit tests, browser signals and well-written questions are enough; hold the exams that really matter in a supervised room, even if students sit them on devices.

Review and approve before results go out

On paper, grades usually pass the teacher and then a head of department or an exams committee before they are announced. Keep that step online. Automatic marking is fast, but it will not notice a wrong answer in the key.

1. The system marks objective questions; the teacher marks written answers against a clear rubric.
2. The teacher reviews the attempts highlighted by integrity signals and notes a decision on each one.
3. A head of department or exams officer reviews unusual results: a question most students got wrong, or a section whose results differ sharply from the others.
4. Results are approved, then released to students and moved into the gradebook.

When most strong students miss a question, the question is usually ambiguous or the key is wrong; the fault lies with the question, not the students. Approval lets you fix it before results reach students and parents, not after complaints arrive. It also protects what comes later: final exam results feed final grades and the GPA, as our guide to calculating GPA explains.

In universities, also check eligibility before the exam opens. A student over the absence limit may be barred from the final, and that should be settled before exam day, not argued about after it. Our guide to university attendance policy covers how to set and apply the limit.

Accommodations and equal opportunity

Fairness does not mean treating everyone the same. Some students need accommodations so that the exam measures their knowledge rather than their circumstances.

• Extra time for students with learning difficulties or disabilities, according to a plan the school has approved.
• An alternative to sitting at home for students without a suitable device or a stable connection, such as taking the exam in the school’s computer lab.
• A clear procedure for dropped connections. What should a student do if the internet cuts out mid-exam, and who decides whether to reopen the attempt?
• Language load. If the exam is in English for students whose first language is Arabic, make sure the wording does not become the obstacle, especially in science subjects.

Record every accommodation in the student’s file and apply it in every exam, not just some. When someone later asks about a decision, you have a clear written reason.

Explain the rules before the exam, not after

Rules that were never explained clearly can turn into a dispute with a student or a parent. Is a calculator allowed? Is the exam open book? May students use their class notes? Does using an AI tool count as cheating?

Write the rules for each type of exam on one page, publish them on the platform and send them to parents at the start of the term. They should cover:

• What is allowed and what is not: books, notes, calculators, websites and AI tools.
• What the platform records during the exam, such as leaving the exam page, and how that information is used.
• What to do if a technical problem happens, and whom to contact.
• How suspected cases are reviewed, and the student’s right to explain.
• The consequences of a breach, as set out in your school’s behaviour policy.

Ask students to acknowledge the rules before their first exam. It also helps to open the first exam with an ungraded practice question, so students meet the interface before they are assessed on it.

The limits of technology in online exams

No platform prevents cheating completely, and anyone who promises that is overselling. Technology can make cheating harder, detect some cases and give you an organised record on which to base a fair decision. The rest depends on how you design assessment and on the culture of your school.

So vary how you assess: projects, short oral presentations, portfolios and assignments written in stages. A student assessed in several ways rarely earns a grade through cheating alone, and a teacher who knows how a student writes soon notices an answer that does not sound like them.

A checklist before every online exam

• You have set the level of the exam: practice, unit test or final.
• The bank holds more questions than the exam, tagged by objective and difficulty.
• Random order is switched on for questions and options.
• You have taken the exam yourself and set the time limit from that.
• You have decided when results and correct answers appear.
• Students and parents know the rules, and students have acknowledged them.
• There is a clear plan for dropped connections and for accommodations.
• You know who reviews flagged attempts and who approves the results.

If you are comparing platforms, the assessment section of our LMS RFP checklist turns these points into questions you can send to vendors.

Where Ilmoria fits

In Ilmoria’s quizzes and exams, teachers set the time and number of attempts, randomise the question order and choose when results appear. In final exams, each attempt receives an integrity score based on browser events such as leaving the exam page, and suspicious attempts are flagged for the teacher to review, with no webcam recording. Final exam grades reach students only after an approval step, then flow into the gradebook. To try it with one of your own exams, book a demo.