Creating an online test isn't just digitizing the paper version. Done well, the quiz grades itself, ensures integrity, and even delivers data that improves teaching. Here is the step-by-step and the safeguards that make a difference.
Quick answer
- Build or import questions (bank, GIFT/CSV formats)
- Mark correct option, explanation, and concept on each question
- Set integrity: attempts, time, shuffling, access code
- Multiple-choice has auto-grading; grades go to the report
- Use item-level psychometrics to improve the test
Step by step
1. Define the goal and the concepts
Before the questions, list the concepts the test should assess. Tagging each question by concept feeds mastery reports and adaptive learning.
2. Build or import the questions
Create from scratch or import from a question bank (GIFT and CSV formats). Use several types: multiple choice, true/false, matching, ordering, fill-in-the-blanks, and open-ended with a rubric.
3. Configure integrity
Set an attempt limit, timed duration, shuffling of questions and options, an access code, and a review mode. These controls protect the assessment.
4. Publish and let the platform grade
Multiple-choice is graded automatically; open-ended uses rubrics. Grades go straight to the report.
Academic integrity (what matters)
| Feature | What it does |
|---|---|
| Attempt limit | Prevent infinite trial and error |
| Server-validated time | Prevent client-side bypass |
| Shuffling | Make copying harder |
| Access code | Control who enters |
| Answers only on server | Don't leak the key before submission |
Data that improves the test
A good platform shows item-level psychometrics: difficulty, discrimination (point-biserial), distractor analysis, and Cronbach's alpha. With these you discover which questions are poorly worded and where the class struggles — and improve the test and teaching.
Frequently asked questions
How do I create a test with auto-grading? Build/import questions, mark the correct one, set rules, and publish; multiple-choice grades itself.
What question types? Multiple choice, true/false, matching, ordering, fill-in-the-blanks, and open-ended with a rubric.
How do I ensure integrity? Attempts, server-side time, shuffling, access code, and answers only on the server.
What is a question bank? A reusable repository tagged by concept; import via GIFT/CSV.
Do quizzes generate useful data? Yes — item-level psychometrics reveal bad questions and class difficulties.
Studeia has a quiz engine with auto-grading, a question bank, and psychometric analytics. See the quiz engine and the question bank.