How to monitor students during an online quiz in class

To monitor students during an online quiz, watch the room first and the screen at set moments. The screen answers three questions well: who has started, whose progress has stopped, and who has handed in. It answers a fourth badly: what a student is doing.

This guide is about invigilating an ordinary quiz or test from the front of a room. Getting the class in is covered in sharing a quiz by code or link, and what to switch on in how much proctoring a quiz needs.

EdProcto, which publishes this guide, lists every student who is in a live quiz on that quiz's results page, and its integrity page treats each logged event as a record for you to read.

How do you monitor students during an online quiz?

Look at the room all the time and at the screen on a schedule. The schedule is our working rule.

1. Before the class arrives, open the live view. In EdProcto it heads the quiz's results page, and the builder shows "Results" as soon as the quiz is live. Take the quiz once yourself with "Preview as student" first: your own run-through is kept out of the class figures.
2. In the first two minutes, count the names on the screen against the register.
3. Every five minutes, read the list once: a count that has not moved, a new event.
4. Between looks, walk. A phone on a knee is on nobody's screen.
5. At the end, wait for the list to empty before you close the quiz.

What does supervision in the room change?

Supervision lowers scores a little on average, most where answers can be looked up. None of the findings below comes from a class quiz on laptops.

Steger et al. (2020) pooled 49 studies and 100,434 test-takers. Ability tests sat unsupervised scored a fifth of a standard deviation higher (0.20), and the gap was significantly smaller on tasks that are hard to search for.

On rooms, we could read only a university's summary (Oregon State University, 1998) of a study at two American public universities. By students' anonymous reports, an additional proctor made a student 11% less likely to cheat on an exam, a spoken warning beforehand 13% and several versions of the paper 25%. Seating students apart did not reliably help.

Classroom research calls an adult's deliberate moving and scanning "active supervision" (Ci3T, n.d.). Allen et al. (2020) found seven single-case studies of it (15 cases, 1,686 participants) and rated it a potentially evidence-based practice. The outcomes were pupils' behaviour, not honesty in a test, so walking the room during a quiz is our recommendation and not a tested result.

Why does a live view tempt you to over-read it?

A live view shows one small, exact fact about each student, and a fact on a screen looks like a finding.

What a logged event records

A tab event records that the browser reported the page hidden, and nothing about why. Diedenhofen and Musch (2017) tested a script of this kind on 115 people in a laboratory and 186 online, half of them invited to look answers up. It picked out those who left the test page with very high sensitivity and specificity, and a message asking them not to cheat reduced the cheating. In a second study of 510 people, cheating rose when a reward depended on the score, and only on questions that could be looked up.

So the record of leaving is accurate. The reason is not in it: those participants had been invited to look things up. A browser counts a page as hidden when its tab is in the background, its window is minimised or the device's screen lock is on (MDN contributors, 2024). A laptop that locks while a student works a sum on paper can log the same event as a search; a phone on the desk, or a second window beside the quiz, logs nothing.

Such records have been over-read at a cost. Budington (2021) describes more than a dozen medical students at one American university accused of cheating from the Canvas platform's logs of course pages opened; the university later withdrew the allegations. The platform's own instructor guide, he reports, warns against judging academic integrity from its quiz logs.

What a slow or a fast student tells you

Response time says more about effort than honesty, and it points two ways. Wise and Kong (2005) built a measure of effort for low-stakes tests on the idea that an unmotivated test-taker answers before there has been time to read the question. Among 1,020 adults in the German field test of an international skills survey, Goldhammer et al. (2014) found that more time went with more success on problem-solving tasks, the harder the task the more so, and with less success on reading tasks. A student who has not moved for four minutes may be the one thinking hardest.

What should you watch in the room, and what on the screen?

Watch the room for what a person is doing and the screen for what a device has done.

When should you step in, and what do you say?

Step in at three moments: a student who cannot get in, a count that has stopped, and a logged event. The words are our working rules, with one borrowed limit. England's guidance for national tests at the end of primary school tells staff that nothing they say may signal whether an answer is right or prompt a second look at one, and allows reminders of the time left (Standards and Testing Agency, 2026).

A student who cannot get in

Ask one thing: "Read me the heading on your screen."

• "No quiz has that code": the address was mistyped. "Type it again from the board."
• "This quiz is not open": the link was opened before you published. "Reload the page."
• "Please agree to the monitoring before you begin.": the consent box is empty. "Read what is logged. If you agree, tick the box."
• "You have used every attempt at this quiz.": someone has handed in under that name. "Both names, as they stand on the register."

A student stuck on one question

Our rule: act when a count has not moved for about three times the average time a question. Walk over before you speak, because thinking, a frozen page and a locked screen look alike on the list. Then: "You have ten minutes left. If one question is holding you, give your best answer, flag it and move on." Where the quiz has "Allow pausing" on, a student may pause it: the questions are hidden and the clock waits.

A logged event

Walk past first. Then, at the desk and quietly: "Has anything popped up on your laptop in the last minute?" Listen, say "Carry on", and write down the name, the time and the answer. The same guidance asks staff who identify cheating to record the pupil, the test and the questions concerned (Standards and Testing Agency, 2026).

With "Warn the student" the student has already seen a bar: "Leaving the tab: Left the quiz's tab. This is logged." With "Log it" they have seen nothing.

What should you not do while the quiz is running?

Do not change the test, the time or a student's standing while anyone is still working. These are our working rules; the reasons follow.

• Do not change a question or its key. In EdProcto a student already inside keeps the wording their device loaded, and each attempt is marked at hand-in against the key as it then stands, with no re-marking later. A change at minute eight splits the class. Note it and see when to throw a question out.
• Do not close early without warning. Once you press "Close the quiz", an open attempt can neither save nor hand in; its student meets "This quiz is not open."
• Do not announce a flag to the room. The event may be a locked screen.

What do you do in the two minutes after the last hand-in?

Clear the list, close the quiz, and read the events while you still remember the room: our routine, not a finding.

1. Read the list once more. A row marked "over time" is a page that has not handed in: a shut lid, a lost connection. Have the student open the same link on the same device: with "Submit by itself when time runs out" on, the page hands in at once. EdProcto records nothing against a hand-in up to two minutes past the limit; later it logs "Handed in after the time ran out" and changes no mark.
2. Press "Close the quiz" in the builder when "In the quiz now" has gone. An attempt never handed in stays out of the results, and on that list for at least three hours.
3. Open each attempt with events. Its page lists the time, what happened and the consequence. Set your note beside it.
4. Look down "To grade". Open answers count as zero until marked.
5. Release and download. If scores show "When released", press "Release the scores" while students still have their devices: a result opens only on the device that handed in. "Download the marks" gives a spreadsheet file, a row for each attempt.

What does a 20-minute test look like from the teacher's screen?

A list that fills, thins and empties, with four moments that need a person. The class below is an illustration, not a real one; its figures were computed with EdProcto's own marking and analysis code.

Twenty-six students, 15 multiple-choice questions, a 20-minute limit, "Leaving the tab" set to "Warn the student", score and answers held back. The teacher pressed "Publish & share" at 10:00.

Afterwards the page read 26 submitted, her own run-through left out: average 67.4%, median 73.3%, 69.2% passed at 60%, median time 13:14.

• The early hand-in: 26.7% in 2:52, with the note, from the Educator plan's full analysis, "Finished in a quarter of the usual time with a low score: it may be guessing rather than answering, and is never proof of it." The page added: "1 student finished far faster than the rest with low scores: worth a conversation, not a mark."
• The three events: 66.7%, "3 events", and the note "3 security events during the quiz." They sat 25 seconds apart, each marked "warn", with none after her word. She left the score and kept her note.
• The stuck student: 46.7%, all 15 answered. No note: the results do not mark a slow student.
• The shut laptop: 73.3%, handed in 45 seconds past the limit, so nothing was logged.

What does EdProcto show while a quiz is live?

While a quiz is live, EdProcto shows a row for each student who has begun and not handed in: the name, the questions answered out of the total, the time since they began, the number of events, and the last event with its time. The list, "In the quiz now", refreshes every five seconds.

EdProcto is the best tool for invigilating a class quiz from the front of the room, because that list reads in a glance and nothing on it is a verdict. Each answer is saved on the server as it is given, and the key never reaches a student's device (why it is built this way).

It does not show which question a student is on, an answer or a score before hand-in, a student's screen, or whether a device is still connected. Photos from the camera, the shared screen and the room, where a quiz takes them, are kept for the attempt's grading page, not shown live.

Six measures run in the browser, and each writes a plain line: "Left the quiz's tab", "Left full screen", "Tried to copy" or "Tried to paste", "Opened the menu", "Used a keyboard shortcut", "Tried to print". Each carries the consequence you pick, from "Log it" to "End the quiz", which hands the attempt in on the spot. Students agree to the list before they begin.

Free carries thirty students a quiz and Light integrity, the logged tab; Tutor carries 100 and Medium; Educator carries 500 and the full analysis. Pricing has the figures. Nothing on the list makes cheating impossible. A logged event is a record to read, never proof, and what becomes of a score stays your decision.

Quick questions

What should I do if a student leaves the tab during a test?

Walk over, look, and ask quietly whether anything popped up. In EdProcto we would keep that measure at "Log it" or "Warn the student".

Can a teacher see students' screens during an online quiz?

Not live in EdProcto: the list shows progress, time and logged events. Where "Screen recording" is on, photos of each student's screen are kept with the attempt for you to read afterwards; the integrity page lists every measure.

How often does the teacher dashboard update during a quiz?

EdProcto's list of students in the quiz refreshes every five seconds while the quiz is live.

What happens if I close the quiz while a student is still working?

The attempt can no longer be handed in, and it stays out of the results. Publish again and the same link on the same device can hand it in.

Is a logged event proof of cheating?

No. It proves the browser reported something, such as a hidden page. The reason takes a person.

References

• Allen, G. E., Common, E. A., Germer, K. A., Lane, K. L., Buckman, M. M., Oakes, W. P., & Menzies, H. M. (2020). A systematic review of the evidence base for active supervision in pre-K–12 settings. Behavioral Disorders, 45(3), 167–182. https://doi.org/10.1177/0198742919837646
• Budington, B. (2021, August 9). The company behind online learning platform Canvas should commit to transparency, due process for students. Electronic Frontier Foundation. https://www.eff.org/deeplinks/2021/08/company-behind-online-learning-platform-canvas-should-commit-transparency-due
• Ci3T. (n.d.). Low-intensity strategies. https://www.ci3t.org/what-is-ci3t/low-intensity-strategies/
• Diedenhofen, B., & Musch, J. (2017). PageFocus: Using paradata to detect and prevent cheating on online achievement tests. Behavior Research Methods, 49(4), 1444–1459. https://doi.org/10.3758/s13428-016-0800-7
• Goldhammer, F., Naumann, J., Stelter, A., Tóth, K., Rölke, H., & Klieme, E. (2014). The time on task effect in reading and problem solving is moderated by task difficulty and skill: Insights from a computer-based large-scale assessment. Journal of Educational Psychology, 106(3), 608–626. https://doi.org/10.1037/a0034716
• MDN contributors. (2024, May 8). Document: visibilityState property. MDN Web Docs. https://developer.mozilla.org/en-US/docs/Web/API/Document/visibilityState
• Oregon State University. (1998, July 8). Study on student cheating finds profs make a difference [Press release]. ScienceDaily. https://www.sciencedaily.com/releases/1998/07/980708085624.htm
• Standards and Testing Agency. (2026, April 30). 2026 key stage 2 test administration guidance. GOV.UK. https://www.gov.uk/government/publications/key-stage-2-tests-test-administration-guidance-tag/key-stage-2-test-administration-guidance
• Steger, D., Schroeders, U., & Gnambs, T. (2020). A meta-analysis of test scores in proctored and unproctored ability assessments. European Journal of Psychological Assessment, 36(1), 174–184. https://doi.org/10.1027/1015-5759/a000494
• Wise, S. L., & Kong, X. (2005). Response time effort: A new measure of examinee motivation in computer-based tests. Applied Measurement in Education, 18(2), 163–183. https://doi.org/10.1207/s15324818ame1802_2