Designing Assessments

Assessment is the core of skills-based learning

In a content-first model, assessment is often an afterthought — a quiz added at the end of a module. In a skills-based model, assessment is the learning activity. The task is what produces the competence, not just what measures it.

Good skills-based assessment has three properties:

1. Realistic — the task resembles something the learner would actually do in a real context
2. Feedback-rich — the learner understands what worked, what did not, and why
3. Evidence-producing — the learner leaves with something they can show

Assessment types by maturity level

Level	Assessment approach	Example
1	Auto-marked interactive exercises	H5P quiz, Moodle quiz with instant feedback
2	Guided task completion	VPL coding task with test scripts, lab completion check
3	Scenario outcome	Investigate a dataset, configure a system, respond to an incident
4	Challenge / proof of solution	CTF flag, passing a full test suite, Kaggle-style submission
5	Performance-based exam	Live lab exam, scored practical assessment in a controlled environment
6	Team simulation	Group incident response, collaborative deployment exercise

How to design a practical task

A well-designed practical task specifies:

• The scenario — the realistic context the learner is placed in
• The inputs — what they are given (data, starter code, a broken system, a set of files)
• The expected output — what a correct or good solution looks like
• The feedback mechanism — how results are communicated (test output, rubric score, teacher review)
• The evidence artefact — what the learner keeps at the end (code repo, report, portfolio entry, badge)

Rubrics

For tasks that cannot be auto-graded, rubrics make assessment transparent and consistent. A skills rubric defines what competence looks like across a spectrum — not just pass/fail.

A simple rubric structure:

Criterion	Not yet	Developing	Proficient	Advanced
[Skill area]	Cannot perform the task	Completes with significant guidance	Completes independently	Extends or adapts the task

EduxPal helps design rubrics aligned with relevant frameworks — DigComp for digital skills, UNESCO AI Competency for AI literacy, NICE for cybersecurity roles, or custom frameworks for the institution's specific learning outcomes.

Automated feedback

Automated feedback is not just about convenience — it is a justice issue. When feedback is fast, specific, and always available, learners can practise more, iterate faster, and improve without depending on a teacher being available.

Tools for automated feedback:

• Moodle VPL — runs student code against test cases, returns pass/fail and output
• GitHub Classroom autograding — CI runs tests on every push, results visible in the repository
• H5P — inline feedback on interactive exercises
• CodeGrade — detailed inline code feedback, rubric scoring, plagiarism detection

Common mistakes to avoid

• Designing assessment before the learning activity — the task should drive the lesson, not follow it
• Grading only on output, not process — for higher levels, how the learner approached the problem matters
• No feedback loop — learners need to understand what to improve, not just whether they passed
• Over-automating — some tasks need human review; autograding is best for objective, verifiable outputs