Stress-test your assessment task before you hand it out

Stress-test your assessment task before you hand it out

The loophole in an assessment task is the one the strongest student finds in the first ten minutes and the rest of the class copies by recess. You usually only see it after you have a pile of 24 papers, three of which got top marks without doing the thinking you set out to assess. This prompt finds it on Friday afternoon instead of Tuesday morning.

The prompt

You are an experienced [SUBJECT] teacher and assessment moderator in an Australian [PRIMARY / SECONDARY / TERTIARY] classroom. You stress-test assessment tasks by role-playing the students who will sit them, not by polishing the wording.

The task I am about to hand out: [PASTE THE FULL TASK INSTRUCTIONS AS STUDENTS WILL SEE THEM, including any rubric or success criteria. If it is long, paste it all anyway.]

Year level: [YEAR LEVEL]
Subject and unit: [SUBJECT, plus where this sits in the unit, e.g. "Year 9 history, end-of-unit task on the Industrial Revolution"]
Time students have: [LESSON / TAKE-HOME / EXAM CONDITIONS, plus minutes]
What this task is actually meant to assess: [ONE PLAIN SENTENCE in your own words, e.g. "Whether students can use two sources to support a claim about cause"]
Curriculum link: [ACARA v9.0 content descriptor, VCE study design dot point, or your own success criteria, copied verbatim]
Class context: [ONE SENTENCE, e.g. "Mixed ability, six EAL/D, three students with ILPs, one student typically finishes in half the time"]

Run three passes over the task. Do not skip any.

PASS 1: The strategic student.
Role-play a clever, time-poor Year [X] student whose goal is the highest mark for the least real thinking. Read the task as they would. Tell me:
a. The shortest path to a passing or strong mark that does NOT require the thinking I said this task is meant to assess. Be specific. If the loophole is "answer the easier of the two parts and ignore the other", say so.
b. The exact wording in the task that lets them do it (quote the line).
c. What I would have to change in the task to close the loophole without making it harder for the kids who are genuinely trying.

PASS 2: The confused student.
Role-play a genuinely capable but anxious Year [X] student who is trying to do the task properly but cannot work out what I am asking for. Tell me:
a. The first place they stall when reading the task. Quote the exact wording that loses them.
b. The most likely misreading: what they will do if I am not in the room to clarify (and I will not be, if it is take-home).
c. The smallest edit to the wording that would prevent that misreading without dumbing the task down.

PASS 3: The rubric reality check.
Look at the rubric or success criteria I pasted. Tell me:
a. The criterion most likely to be applied inconsistently by two teachers marking blind. Quote it. Explain in one sentence why.
b. The criterion that sounds rigorous but is actually unmarkable in the time I will have per script. Name the time per script you are assuming.
c. The bit of the task students will be assessed on that is NOT in the rubric. (This is the silent killer in moderation meetings.)

Then, "The one fix": if I only have time to change ONE thing before I print this tomorrow, what is it? Pick from passes 1, 2, or 3 and justify in one sentence against the year level and class context.

Then, "Equity check": one sentence on which group of students in this class context is most likely to be disadvantaged by the task as written. Do not give me a generic "support EAL/D students" line. Name the specific feature of THIS task that creates the disadvantage.

Finally, "Sanity check": one sentence on what would make you, the model, wrong about the strategic-student loophole. If the loophole you named is more common in textbook tasks than in real Australian classroom tasks at this year level, say so.

Australian English throughout. Do not use the phrases "deep dive", "unpack", or "scaffold their thinking".

What it does well

Forces the model to read the task adversarially, which is the move that catches the loopholes a teacher who wrote the task cannot see. The strategic-student pass is the one that earns its keep. It surfaces the "answer two of the four prompts and average the marks" style shortcut that every Year 9 cohort eventually finds. The rubric reality check is the bit that quietly saves a moderation meeting in Week 9, by naming the criterion that two teachers will mark differently in good faith.

Where it falls over

The model will sometimes invent a loophole that no real student would actually find, because it pattern-matches to "loophole-sounding wording" without modelling effort. Read the strategic-student pass against the cohort you actually have. If the move requires more cunning than your class typically deploys, ignore it. The confused-student pass also tends to over-index on vocabulary, when the real source of confusion is often the implicit step between two instructions, which the model is bad at noticing because it has no body in the room. The prompt also will not save a task built on the wrong question. If the task is testing recall when the unit was about analysis, no stress test fixes that. That is a redesign job.

Try it like this

For a Year 10 English analytical response on a set text: the loophole the strategic student almost always finds is "quote-pile, no analysis", supported by a rubric line that says "uses textual evidence". The fix is usually a single added line in the criterion that says "each quote must be followed by an inference that goes beyond the quote".

For a VCE SAC draft (any study design): the rubric reality check is where this prompt earns its keep. The criterion most likely to be marked inconsistently is the highest-band descriptor, the one with "sophisticated", "nuanced", or "insightful" in it. Get the model to rewrite that descriptor in the same plain language the lower bands use.

If you only have five minutes today, skip passes 1 and 2 and run pass 3 against the rubric only. The rubric reality check is the highest-yield pass for the time, and the one most likely to save you a marking dispute in the meeting after results go home.

The task you stress-test on Friday is the one you do not have to defend on Monday.