You reduce cheating on an online practice test by making lookup unprofitable rather than impossible - application questions instead of recall, numbers you changed yourself, real negative marking, section clocks set to the real exam's timing, and live rounds where the whole room answers at once. None of it needs a camera, which is fortunate: MockSetu has no proctoring. No webcam, no AI invigilator, no lockdown browser, no tab-switch detection, no screen recording. A published mock is public, so anyone with the link can open it. If a seat or a certificate rides on your test, buy a proctoring vendor. If it is a practice paper - a weekly test, a chapter test, a full-length mock - here is why a camera would not have fixed your problem anyway, and what does.
The Honest Inventory, Before Any Advice
A comparison table is a poor place to learn what a platform cannot do. Here is what is missing, up front.
- No webcam or AI proctoring. Nobody watches the student and there is no recording to review.
- No lockdown or secure browser. A second tab, window or phone is available the whole time, and the paper will never know.
- No tab-switch detection. There is no focus-loss counter, no warning banner, no integrity flag on the report.
- No question shuffling. Two students sitting next to each other get the same questions in the same order.
- No attempt limit. The same student can attempt the same paper again, and again.
- Published papers are public. There is no private delivery to one batch, no access code, no paywall.
Once you accept that the software will not police the student, you make the paper do the work instead - the only part of this problem you were ever in control of.
Who the Cheating Student Actually Defrauds
A mock test produces one thing of value: an honest reading of where a student stands weeks before the real paper. A student who looks up answers has not stolen marks from you, from a classmate, or from a rank list that does not exist. They have corrupted their own diagnostic - they go into the next week believing their coordinate geometry is fine when it is not, on the strength of a score the exam hall will not reproduce.
This is different from a board exam or a recruitment test, where a score converts into something scarce. On a practice paper the incentive is weak and the punishment self-inflicted. Say that to the batch once, plainly, and the problem usually shrinks to the handful who were going to copy regardless - students a webcam on a shared family laptop would not have stopped either.
A student who copies on a mock has not beaten you. They have deleted their own diagnostic and will find out in the exam hall.
Design the Paper So Looking It Up Does Not Pay
The single most effective anti-cheating measure is a question that cannot be answered by pasting it into a search box. Recall questions - who, when, which formula, what is the SI unit - are perfectly copyable and always will be. Application questions are not: a student has to understand the setup well enough to type a useful query, and by then could have solved it.
- Put the data in the stem. A numerical with values that appear nowhere else cannot be matched against a bank of past papers.
- Change the numbers, not just the wording, when you reuse a previous year question - a reworded stem is found instantly by its original phrasing.
- Use numeric-answer questions for the parts you care most about. There is no option list to guess from and no four-choice pattern to recognise.
- Use multi-correct questions with part marks where the subject allows it - a half-understood lookup produces a half-right selection, which the marking scheme already handles.
- Set a real negative mark for a wrong answer. A student guessing from a half-copied answer is now taking a risk, not a free shot.
MockSetu lets you set marks for a correct, wrong and unattempted answer on every question, and multi-correct questions can award part marks or be all-or-nothing, with the wrong-answer penalty charged once or per wrong option. That is not a security feature, but it is the lever that makes guessing expensive. The question-writing mechanics are in how to write good multiple choice questions, and they matter more here than anything on a vendor's security page.
Make the Clock Do the Invigilating
Lookup takes time. A student who pauses to search has to find the question, read a result, decide whether it matches, and come back. On a loosely timed paper that costs nothing; on a paper timed close to the real thing, it costs them questions they could have solved themselves.
On MockSetu, locking section switching gives each section its own clock, and two or more adjacent sections can share one pooled clock through a timing group; leaving switching open instead puts the whole paper on a single clock. Either way the paper auto-submits when time expires. Lock the switching and a student cannot park in section three while hunting for an answer in section one. Take your timings from the real paper rather than from what feels comfortable - JEE Main Paper 1 allows 180 minutes for all 75 questions, and SSC MTS Session II is a 45-minute session carrying a one-mark penalty. For any other exam, read the duration off the current official bulletin rather than off memory.
One caveat: a tight clock punishes a slow honest student exactly as hard as it punishes a copier. Use the real exam's timing, not an artificially cruel one.
Keep the High-Stakes Test in Your Own Room
If a test genuinely decides something - batch allocation, a scholarship slot - run it in your centre, on devices you can see, with you walking the aisle. That is not a workaround, it is the correct answer. A teacher at the back of the room beats any browser lock, whose failure mode is a second device it cannot see.
- Collect phones at the door for any test that decides something, or have them face-up on the desk where you can see the screen.
- Seat students so that no two people attempting the same paper share a sightline to each other's screen.
- Announce a fixed start time and start the room together - a published mock has no scheduled start of its own, so this one is yours to call.
- Keep one spare device and one hotspot ready - an integrity incident and a connectivity incident look identical from the front of the room.
The operational half of this - devices, network, start time, what to do when a student drops - is its own job, worth doing properly before you worry about integrity at all. The sequence is in how to conduct an online exam for students, and the wording for the paper's own instructions page is in this online exam instructions template.
A Live Exam Closes the Window Instead of Watching It
One format removes the lookup window rather than policing it. In a MockSetu live exam, students join from their phones with a single code - there is a one-time sign-in first, so leave room for it - and the whole room answers the same question at the same time, with the creator controlling when the paper moves on. You can set a scheduled start with a countdown, project the room on a screen, show or hide the answer options there, and reveal the correct answer once time on a question is up.
The reason this suppresses copying is structural, not technical: nobody is ahead, so there is no finished neighbour to copy from, and the per-question window is short enough that a search is a bad trade. Live answer bars show how the class split across the options; standings can be visible to everyone, only to you, or switched off; names can be hidden on the big screen. There is also a private button a student can tap to say they are lost, which only you see - the signal you actually wanted, and one no webcam would have produced.
A live room is not right for everything: put a very large cohort through a full-length paper as a published self-paced mock rather than one live session, and keep the live format for shorter diagnostic rounds.
Read the Pacing, Not the Faces
Cheating leaves a signature in timing, and timing is the one thing you can actually see. The MockSetu live report gives class accuracy, participation, drop-off, median score, median answer time per question, a split of fast and slow against right and wrong, a count of taps on the private lost button, a students tab and a shareable report link. The split is the interesting one: a cluster sitting in fast-and-right on a question the rest of the class laboured over is worth a second look.
Be careful about what this can and cannot tell you. Outside a live exam, MockSetu's creator analytics are aggregated on purpose - section-wise accuracy, per-question timing averaged across everyone who attempted that question, where the class as a whole struggled. The one place an individual surfaces is a top-three board, which names the three highest-scoring sittings by their handle and shows no email and no full name. You will see that the cohort answered a question implausibly fast; you will not get a per-student integrity report, because nothing else in a self-paced mock is reported per student and there is no CSV export to build one from. If naming a particular student afterwards is the point, this platform will not do it.
- A hard question answered faster than an easy one, across the class, usually means the paper is circulating, not that the class got clever.
- Near-perfect accuracy on the section you expected to be weakest is a paper-integrity signal, not a teaching win. Verify it with a short in-room test.
- Drop-off concentrated at one question is almost always a broken question or a rendering problem, not cheating. Check the question first.
Plan Around a Public Paper Instead of Fighting It
A published MockSetu paper sits in the public library where any student can attempt it as a guest, with no account. That is a distribution feature, and it is also the reason a paper cannot be kept to one batch. Treat the first publish as the moment the paper becomes public knowledge, because it is.
The consequences are small once you plan for them. Run the in-centre version first and publish afterwards, so the batch sits the paper before it is browsable in the public mock test library. Unpublish a paper whose key has leaked. Duplicate the exam for the next batch and change the numbers rather than rerunning the identical paper. If you find a wrong key after students have attempted, unpublish the paper to correct it, because a published exam is locked for editing, then publish again and tell the batch. Nothing re-scores an attempt already recorded - a paper is graded at the moment it is submitted and those marks stand. If you write replacement questions outside the platform, the bulk upload format is documented in the JSON upload guide - with one limitation to plan around: it does not carry numeric, TITA or match-the-column questions, so the very questions this article told you to build the paper around must still be typed into the editor by hand.
When You Genuinely Do Need a Proctoring Vendor
There is a real line here and it should be drawn honestly. If the result converts into something scarce and irreversible - an admission decision, a professional certification, a hiring shortlist, an assessment an external body will audit - you need identity verification, session recording and a defensible chain of evidence. That is a different product category at a different price, and a practice-test platform should not pretend to cover it. Buy the vendor and budget for the support load.
Everything short of that line - weekly tests, chapter tests, full-length mocks, a free paper you publish to build a following - is practice, and practice does not need surveillance. It needs a paper worth attempting honestly. MockSetu is free with no card, and what a creator can and cannot do on it is laid out on the page for creators.
A Working Setup, Start to Finish
One sequence for the next test. Build the paper on application questions and numericals with real negative marking; set each section's clock to the real exam's timing, lock switching, leave auto-submit on; preview it yourself end to end, which records nothing; run the graded sitting in your own centre with phones visible; use a live round for the follow-up diagnostic; then publish, duplicate with changed numbers, and unpublish anything whose key is out.
None of it involves a camera, and all of it is within reach this week. The institutes that worry least about cheating are the ones whose papers are hardest to cheat on - which makes this a question-writing problem, and question writing is the one thing on the list you were already good at.