Second devices in remote exams: designing for the phone you cannot see
Admit the blind spot in the design brief
Proctoring comparisons are unusually honest about this gap. The Invigilator's lockdown comparison lists what a browser lock cannot determine, including whether a student is using a second smartphone or tablet and whether help is arriving from off camera. Starting a design meeting from that admission changes the shopping list. Money stops flowing toward stronger locks on the visible screen and starts flowing toward the exam itself.
It also changes what you promise. An institution that tells accreditors its remote exams are secured because a lockdown browser is installed has made a claim the vendor's own documentation contradicts. The defensible claim is narrower and stronger: device controls, identity checks, session recording, and assessment design operate together, and here is the evidence each layer contributes.
Make lookup a losing strategy
A phone helps when questions reward retrieval: definitions, dates, standard formulas, the first result for a well known problem. Tighten the clock so a lookup costs a question elsewhere, randomise values and scenarios per student so a photographed question is not the neighbour's question, and weight marks toward working shown inside the answer. The phone does not become useless, but its exchange rate collapses, and most students do that arithmetic quickly.
Rotating banks matter more than bank size alone. A small bank recycled every term leaks into homework sites and group chats within a year, after which the second device is not searching the internet, it is searching your own past papers with model answers attached. Retire questions on a schedule, track which items appear in leak reports, and treat bank maintenance as a standing cost of remote delivery, not a one off project.
The environment scan is a cost raiser, not a guarantee
A startup room scan, where the student shows the desk and the space around it to the webcam, will not catch a phone in a pocket. What it does is force a decision: the student must actively conceal the device, keep it silent, and use it without looking down in a pattern the recording will show. Each of those is a chance to be caught and a reason, for the marginal student, not to bother. Deterrence is a real effect even when detection is imperfect.
Pair the scan with clear rules stated before exam day, not discovered during it: one screen, desk clear, phone outside the room or face down beyond reach, depending on your policy. Students plan around rules they know. Ambiguity produces the worst outcomes, honest students anxious about innocent items on the desk and dishonest ones claiming they never heard a phone was banned.
A single glance downward proves nothing; people think with their eyes anywhere. Reviewers should look for rhythm: repeated glances to the same off camera point at question intervals, typing bursts after long pauses with no reading time, answers in a register the student's coursework never uses. Pattern evidence supports a fair conversation with the student. A lone flagged frame supports an argument you will probably lose on appeal, and deserve to.
Review recordings for patterns, not moments
Protect the reviewers too. Give them the student's prior work for comparison, a second reviewer for any case headed toward a misconduct finding, and a written standard for what counts as corroboration. The goal is a process a student can trust even when it finds against them, because that trust is what keeps the honest majority reporting the leaks and workarounds your software will never see.
Frequently asked questions
Keep reading: Question bank leaks: your exam is on a homework site, now what, Identity checks before the exam: the impostor is stopped at the door or not at all.
Frequently asked questions
Can any proctoring tool reliably detect a second phone?
Not reliably. Tools running on the exam device cannot see other devices, and webcam monitoring catches only careless use within the frame. Vendors say this themselves in their lockdown comparisons. The realistic approach is design: timed, randomised, application heavy questions that make a second device worth much less, plus environment scans that raise the effort of using one.
Do room scans actually stop phone use?
They do not guarantee it, and a phone can stay hidden. What a scan does is force concealment and create a recording where repeated use leaves a visible pattern. That raises both effort and risk enough to deter the marginal case, which is where most of the volume is. Treat scans as a deterrent layer, not a detection promise.
How often should question banks be refreshed for remote exams?
On a schedule, with items retired as they age or appear in leak reports and homework sites. A static bank leaks completely within a few terms of remote delivery, after which second devices stop being a search tool and become an answer key. Budget bank maintenance as a recurring cost, sized to how high the exam's stakes are.
What evidence is fair to use in a misconduct case from a recording?
Patterns, corroborated: repeated off camera focus at question intervals, answer register far from the student's coursework, timing that fits lookup rather than thinking. A single flagged moment is not enough, and a second reviewer should confirm any case heading to a formal finding. That standard protects honest students and makes real cases hold up on appeal.