Open-Book and Applied Assessment

Opening the book without changing the questions produces an easier exam and a worse measurement. The questions have to change with it. A team using the scope example can make redesign work and reviewer effort visible while leaving the assessment itself open-book and focused on application.

The failure mode

Recall questions with the book open become lookup exercises.

Marks compress at the top, discrimination collapses, and the assessment stops distinguishing anybody.

Staff conclude that open-book does not work, when what did not work was a closed exam with one rule removed.

What to change

From "what is X" to "given this situation, what would you do and why".

From "list the factors" to "these three factors conflict here; which governs and on what basis".

From a single correct answer to a defensible position with reasoning.

And from general cases to specific ones drawn from the module's material.

Personalised inputs

Each student gets a different dataset, case or parameter set.

Generated from a template, which is an hour of work and removes answer-sharing almost entirely.

The marking scheme has to work against the generator rather than a fixed answer, which is the part that takes design effort.

For quantitative subjects this is the single most effective measure available.

Time

Open-book assessment needs more time per mark, not less.

Under heavy time pressure it rewards having material pre-organised, which advantages those who anticipated the questions and reintroduces the problem.

Longer windows with fewer questions discriminate better.

Marking

Reasoning-based answers need rubrics rather than answer keys.

Which is slower and is where the real cost sits.

Moderation matters more, because judgement varies between markers in a way that tick-box marking does not.

Budget for it honestly rather than discovering it in the marking period.

What students need told

That open-book does not mean easier.

That reproducing source material scores nothing.

And what good looks like: a worked example released in advance, which is the single most effective piece of guidance and reduces the advantage of students who have seen this format before.

Collaboration

Open-book does not authorise collaboration, and students frequently assume it does.

State the rule explicitly and specifically: may they use the internet, their notes, each other, an AI tool.

Ambiguity here produces honest students breaching rules they did not know existed, which is an avoidable injustice.

What this does to integrity cases

Institutions that move to well-designed open-book assessment generally report fewer cases, partly through fewer opportunities and partly because the boundary is clearer.

Measure it for your own modules rather than taking the claim on trust.

What to check

Were your open-book questions rewritten, or is it a closed paper with the restriction lifted?

Are inputs personalised anywhere they could be?

Have students been told explicitly what is and is not permitted, including AI tools?

And has the marking time been budgeted?

The point

Opening the book without changing the questions produces an easier exam and a worse measurement.

What failed in most attempts was a closed paper with one rule removed.

Worth stating

Release a worked example in advance.

It is the single most effective piece of guidance for open-book assessment and it reduces the advantage held by students who have seen the format before.

Also worth knowing

Open-book assessment needs more time per mark, not less.

Under heavy time pressure it rewards having material pre-organised, which advantages those who anticipated the questions and reintroduces the problem it was meant to solve.

And finally

Reasoning-based answers need rubrics rather than answer keys, which is slower and is where the real cost sits.

Moderation matters more, because judgement varies between markers in a way tick-box marking does not.

Summary

State explicitly what is permitted: internet, notes, other students, AI tools. Ambiguity produces honest students breaching rules they did not know existed.

In summary

Opening the book without rewriting the questions produces an easier exam and a worse measurement.

What most institutions concluded does not work was a closed paper with one rule removed. For wider institutional context, consult IEEE.