How to conduct an online test for your students

Six decisions, in the order they bite. Most of this is true whatever tool you use; the last section is how ViJEEta does it.

1. The paper

Everything else on this page is logistics. Decide the syllabus, the count per subject, and the question types first — and match the real exam's shape rather than a round number of questions. If your students are sitting JEE Main, a paper of 20 single-choice plus 5 numeric per subject teaches the pace they will actually face. A 30-question mixed paper does not, however good the questions are. The exam shapes are on the paper-setting page.

2. The window

Give one start time and one end time. "Any time before Sunday" is not a test — it is an open-book assignment with a timer, and it makes the scores incomparable, because the student who sat it on Sunday night had four extra days of preparation on the same syllabus.

A narrow window is also the only anti-copying measure that costs an honest student nothing. And decide the grace: a student who started five minutes before the close should be allowed to finish. On ViJEEta an attempt already in flight may land up to 30 minutes after the window closes.

3. The marking scheme — say it before, not after

Negative marking changes what a score means and what a student should do in the last ten minutes. Put it in the message that carries the link. "+4 correct, −1 wrong" and "+4 correct, 0 wrong" produce completely different papers from the same students, and a class that was not told will produce a mixture of both — which you then cannot compare.

One exam makes this vivid: ISI's B.Stat/B.Math UGA paper awards +1 for a question left blank and 0 for a wrong answer. A student who does not know that will answer everything and score below a student who left the same questions alone.

4. The device: assume the worst phone in the class

The realistic floor in most Indian batches is a shared Android phone on mobile data. Every extra requirement — install this, use Chrome, enable the camera — removes students, and removes the weaker ones first. A test that opens from a link in a browser reaches everybody you sent it to.

Ask for one thing only: a screen they can read a diagram on. A phone is fine for objective papers; long numerical work is easier on anything bigger.

5. What happens when the internet drops

It will. Decide in advance and tell them: can they reopen the link and continue, or is the attempt gone? On ViJEEta the answers are held on the server as the student goes, one scored attempt is allowed per student, and a second attempt is refused rather than silently overwriting the first — so a student who reconnects is not accidentally graded twice, and a student who wants a re-run has to ask you.

6. Copying — the honest version

You cannot proctor a phone test taken at home. Browser lockdown, tab-switch counters and webcam checks are all defeated by a second phone lying on the table, and they punish the student with weak internet far more than the student who is cheating. Anyone selling you certainty here is selling you something.

What actually works is changing what the test is for:

• Same paper for everyone. Randomising the questions per student destroys comparability and buys very little — it is the same syllabus either way.
• Short window. Two hours, not two days.
• Read it as diagnosis, not selection. A chapter-level signal — say, most of the class losing marks in Rotational Motion — survives a few students copying. A rank list does not.
• Keep selection for the tests you invigilate. Online for frequency, offline for stakes.

Then: read the result properly

Three numbers, in this order.

• How many actually sat it. If you shared to forty and twelve sat it, you do not have a hard paper — you have a broken message, a bad time slot, or a link that did not open. Fix that before you look at any score.
• Accuracy by chapter, not the average. The average tells you the paper's difficulty. The chapter breakdown tells you what to teach on Monday.
• The questions answered wrong fast. A quick wrong answer is a confident misconception; a slow wrong answer is a gap in method. They need different lessons.

And be careful comparing this test's average with the last one's unless both papers sat at the same difficulty. ViJEEta prints the change against your previous test with a warning attached when they did not — a class can look like it is improving purely because the second paper was easier.

Doing all six on ViJEEta

All six, free, in one place. The paper comes out at the real exam's pattern and difficulty; the link opens in any phone browser; the server grades it with that paper's own marking scheme. You get back who actually sat it, the accuracy per chapter, and each student's paper.

See what Test Papers does The message to send

Questions teachers ask

How do I conduct an online test for my students?

Decide six things before you send anything: the paper, the window it is open, the marking scheme, what device a student needs, what happens if their internet drops, and what you will do with the result. Then send one link with all six stated in the message. Most online tests fail on the sixth — a test whose result you will not read is homework with a timer on it.

Can I stop students from copying in an online test?

Not on a phone in their own home — no browser lock, tab-switch counter or webcam check survives a second device sitting next to the first one, and the ones that try mostly punish honest students with weak internet. Design around it instead: keep the window short, give everyone the identical paper so the ranking is at least internally consistent, and read the result as a diagnosis of which chapters the class is weak in rather than as selection. A chapter-level signal survives a few students copying; a rank list does not.

What device do students need for an online test?

Assume the worst one in the class. A phone browser is the realistic floor in most Indian batches — not a laptop, not an app install, not a specific browser. Anything that requires more than opening a link loses the students who most need the practice.

How long should an online test be?

Match the real exam's per-question time rather than picking a round number: JEE Main gives 180 minutes for 75 questions, NEET 180 minutes for 180, ISI UGA 120 minutes for 30. A 20-question paper in 30 minutes teaches a pace that does not exist in the exam.