Exam Practice — the full guide

1. What Exam Practice is

Exam Practice is the part of CampusTutor that simulates the test, not just teaches the material. Two distinct experiences live under this name:

• Structured assessments (Topic Quiz, Course Test, Exam Simulation) — timed, scored, and locked on submit, drawn from a question bank seeded with institution-specific past questions plus AI-generated questions calibrated to your level.
• Practice chat at /exam — a free-form, one-question-at-a-time conversation with the AI in examiner mode. Good for when you want to warm up or check a single concept without committing to a 30-minute test.

Questions are drawn from your institution, your course, and your level, then graded against the same mark scheme structure your lecturers use. The goal isn't to get a high score on the practice — it's to find out, before the exam, where you'd lose marks.

2. Plans & session limits

Free-plan students can see a blurred preview of a Topic Quiz — full question text, blurred answer feedback — so you can feel how the assessment flow works before upgrading.

3. The three assessment modes

Topic Quiz

8 questions on a single topic. Untimed. Practise at your own pace. Best for spot-checking one specific topic the night before a continuous assessment or after a difficult lecture. Available on Free (with the blurred preview cap) and unlimited on Plus and above.

Course Test

20 questions across all topics. 30-minute timer. Simulates a mid-semester CA — broad but quick. Best for checking course-wide preparedness about a week before a CA. Plus and above.

Exam Simulation

35 questions under strict exam conditions. 90-minute timer. Locked on submit. The full dress rehearsal — biased toward exam-weight topics, includes long-answer and short-answer questions in addition to MCQs, and runs the adaptive engine so the difficulty shifts as you answer. Plus and above.

4. Starting an assessment

1. From the dashboard sidebar, click Tutor Mode and pick the course you want to test on.
2. In the course workspace, click the Start Assessment button (or hover a specific topic in the panel and click the Brain icon for a quick Topic Quiz).
3. Pick a mode from the launcher dialog (Topic Quiz / Course Test / Exam Simulation). Locked modes show a Plus badge — tap to see the upgrade prompt.
4. If you have an Area of Concentration set for this course, a "Bias toward my AOC" checkbox appears for Course Test and Exam Simulation. Tick it to weight questions toward your high-priority topics.
5. Click Start Assessment. Questions generate (this takes a few seconds) and the exam screen takes over the full view.

5. Timer & navigation

The exam screen has three sections:

• Top — Timer & submit. For Course Test and Exam Simulation, a countdown ticks down in the top bar. When it crosses the 5-minute mark, the timer turns amber. At 1 minute it turns red. When it hits zero, your answers are auto-submitted as-is.
• Centre — Question Card. The current question. MCQs show four options; SHORT_ANSWER and LONG_ANSWER questions show a textarea. Use Next and Previous or press the arrow keys to move between questions.
• Right — Navigation Panel. A grid of every question number, colour-coded: grey = not visited, blue = current, green = answered, amber = flagged for review. Click any number to jump.

6. Question types

CampusTutor questions come in three flavours:

Multiple Choice (MCQ)

Four options, one correct. Graded instantly — no AI delay. The bulk of Topic Quiz and Course Test questions are MCQs. Each MCQ has a difficulty tag (Easy / Medium / Hard) that affects how much mastery it contributes.

Short Answer

A textarea for a one-to-three sentence answer (definitions, brief explanations, formula recall). Graded by AI against a model answer and a keyword rubric. Worth roughly twice the mastery contribution of an MCQ.

Long Answer (Essay)

A textarea for an essay-length response. Graded by AI against a rubric with multiple criteria (e.g. understanding, accuracy, structure, depth, examples). Worth roughly three times the mastery contribution of an MCQ. Available in Exam Simulation and in some Course Tests.

7. AI-graded structured answers

Short-answer and long-answer questions are graded by a dedicated AI grading engine, not the same model that wrote them. It runs a double-pass for accuracy:

1. First pass grades your answer against the model answer and rubric.
2. If the first pass returns low confidence (below 75%), a second pass runs.
3. If the two passes disagree by more than 10%, the scores are averaged and a Low confidence badge is shown — meaning "this score is approximate, not gospel."

What you get back per structured answer:

• Score arc — your score out of the question's max, visualised as an arc.
• Rubric table — every criterion, your score on that criterion, and a one-line note explaining why.
• What you did well — the strongest parts of your answer.
• What needs work — areas to focus on next time (we deliberately don't say "weaknesses").
• How to improve — concrete, specific advice for next time.
• Confidence badge — High / Medium / Low, so you know how much to trust the score.

Structured grading is asynchronous — when you submit a long-answer question, you'll see a spinner while the grading completes (usually 10–30 seconds per question). The page polls every 3 seconds; you don't need to refresh.

8. Adaptive testing

Exam Simulation runs an adaptive engine behind the scenes that adjusts question difficulty as you progress, mirroring how a real exam paper escalates.

For each topic, the engine tracks three numbers:

• Ability score (0–100) — a continuous estimate of your skill on that topic, updated after every answered question.
• Confidence score — rolling correct-rate over your last 10 answers.
• Stability index — how consistent your answers are. Stable means the system trusts the ability number; volatile means it's still calibrating.

Each correct answer raises ability by 3 (Easy), 6 (Medium), or 9 (Hard) points. Each wrong answer lowers it by the same. Then the engine picks your next question from a difficulty distribution that matches your current ability:

• Low ability (under 35): 70% Easy, 25% Medium, 5% Hard — build confidence first.
• Mid (35–65): 20% Easy, 60% Medium, 20% Hard — the calibration zone.
• High (over 65): 5% Easy, 30% Medium, 65% Hard — challenge mode.

9. Free-form practice chat

Open /examfor the practice-chat experience. It's a single scrollable conversation where:

1. You pick a course (and optionally one or more topics) at the start.
2. The AI asks you an exam-style question.
3. You answer in your own words.
4. The AI marks your answer using mark-scheme language (full marks / partial credit / common mistake to avoid) and immediately asks the next question.

Practice chat is lighter than a full assessment — no timer, no question grid — but feedback is just as detailed. It's the right tool for "quick five-minute revision before bed." Each user message counts as one session against your monthly cap.

10. The anxiety-friendly UX

Exam practice can be stressful — that's by design when the exam is real, but counter-productive when you're practising. CampusTutor's exam features are tuned to lower anxiety:

• Rotating reassurance messages. While the AI loads its next question, you'll see one of five calming messages ("Take your time", "You're building this muscle") instead of a generic spinner.
• Wrong-answer feedback is slate, not red. A getting-it-wrong moment is shown in a calm slate colour, not the alarming red used by most quiz apps. The goal is to teach you, not to punish you.
• Encouragement rotates. Eight different correct-answer messages, four different incorrect-answer messages. You never see the same response twice in a row.
• Panic words trigger calm responses. If you type something like "I'm going to fail" or "I don't understand anything", the AI catches that and shifts into a supportive, normalising tone before continuing.
• Score framing is supportive. Whether you score 30% or 90%, the framing focuses on what you can do next, not on the gap to perfect.

11. Reading your results

When you submit a structured assessment, the Result Screen shows:

• Headline score. Your overall percentage and a grade prediction band (A ≥ 70%, B ≥ 60%, C ≥ 50%, At Risk below 50%).
• Topic breakdown. Per-topic accuracy, so you can see exactly which topics dragged the score down vs which carried you.
• Question-by-question review. Tap any question to see your answer, the correct answer, and (for structured answers) the full rubric breakdown.
• Areas to focus on. A short list of 2–4 topics where you lost the most marks, with links straight back to Tutor Mode so you can study them.
• Time-to-answer. Average seconds per question — useful for spotting questions where you spent too long.

All results feed your topic mastery scores and your overall GPA forecast automatically.

12. Past attempts & history

Every assessment you complete is saved. From the course workspace, click Assessment History to see every attempt sorted by date, with the type (Topic Quiz / Course Test / Exam Simulation), the score, and a quick view button to re-open the full result screen.

History is permanent — even on Free plan, your results from any assessments you completed before downgrading remain accessible.

13. Viva & proposal practice (postgrad)

If your academic profile is set to Postgraduate, opening /exam takes you straight into one of two specialised modes:

• Proposal Practice (when dissertation phase = PROPOSAL). The AI plays a panel member reviewing your work. You describe your research question and objectives; it probes your scope, justification, and methodological choices.
• Mock Viva (when dissertation phase = VIVA, or when you toggle viva-practice on from Tutor Mode). The AI plays your examiner. You name a chapter or your thesis title; it begins probing — about your contribution to knowledge, your methodology, the limits of your findings, and the choices you defended in your write-up.

Add your research area to your academic profile to get domain-relevant questioning. The closer your research area description, the sharper the viva.

14. Integrity checks

For structured answers, the AI grader runs three integrity checks alongside scoring:

• Short response, high score. Flags answers that scored very well despite being unusually short — sometimes a sign of memorised one-liners rather than understanding.
• Response time anomaly. Flags answers submitted unusually fast or unusually slow for their length.
• Pattern anomaly. Flags answers that look closely similar to other answers you've previously submitted, in case you're pasting from a previous attempt.

Integrity flags are signals to you, not penalties — they show up as a small banner on your result so you can reflect on whether you really understood the answer or just got lucky.

15. How to get maximum results

1. Do a Topic Quiz after every Tutor Mode session. Eight questions, untimed, no stakes. Catches the gap between "I read it" and "I actually know it" while the topic is still fresh.
2. Set your AOC before sitting Course Tests and Exam Simulations. The Bias-toward-AOC toggle makes practice match the real exam's emphasis.
3. Sit Exam Simulation under real conditions. Phone in another room, full 90 minutes, no breaks. The point is to feel the real fatigue before exam day, not after.
4. Always go back to the question review. The score is the least useful number — the per-question review is where the real learning happens. Spend 15 minutes after every assessment reading why you got things wrong.
5. Write long answers in your own words. Pasting a textbook paragraph triggers the pattern integrity check and won't actually help — the AI grader rewards understanding over memorisation.
6. Use the topic breakdown to plan the next week. Two topics where you scored under 60% becomes "next week, do a Tutor Mode topic session for each one, then re-do the Topic Quiz." This loop is the most reliable way to lift exam scores.
7. Don't binge on Exam Simulation. Once a week max — they're heavy and burnout is real. Course Tests are the right weekly cadence.

16. FAQ

Are the questions real past questions?

Topic Quiz, Course Test, and Exam Simulation pull from a mix of institution-specific past questions (where students or institutional partners have supplied them) and AI-generated questions calibrated to your level and topic. When a question is sourced from a real past paper, you'll see a small Past Q tag.

Can I see the correct answer before submitting?

No — that would defeat the purpose. Correct answers and full feedback reveal only on the Result Screen after submission. For practice chat (the /exam page), the AI marks each answer immediately before the next question, which is closer to a tutorial than an exam.

What happens if I lose internet mid-exam?

Your answers are saved to local state as you type. When connection returns, the page resumes. If the timer expired while you were offline, your answers are submitted as-is when you reconnect.

Can I retake a Topic Quiz I've already done?

Yes — Topic Quiz is unlimited on Plus and above. Each attempt pulls a fresh set of 8 questions; the engine deliberately avoids repeating questions you've recently answered correctly.

How is the grade prediction (A / B / C / Risk) calculated?

From your weighted percentage on the assessment: A ≥ 70%, B ≥ 60%, C ≥ 50%, At Risk below 50%. Long-answer and short-answer questions are weighted more heavily than MCQs (3× and 2× respectively) because they test deeper understanding.

Can the AI grader be wrong?

Occasionally, yes — especially on long answers where there's a legitimately different valid interpretation. That's why the confidence badgeis there: Low confidence means "the AI is not sure, take the score as approximate." You can also send disputes to support and we'll review them manually.