How to plan competency-based assessment that holds up

Competency-based assessment is easy to fake and hard to do. The fake version is a booklet of short-answer questions, a tick-and-flick observation that says “satisfactory” twelve times, and a cover sheet with a unit code copied from last year’s intake. The real version is a planned judgement: this person can do the work this unit describes, in conditions that look like the job, and I can show another assessor how I know.

If you trained through Certificate IV in Training and Assessment, this is supposed to be the centre of the job. What the Cert IV actually prepares you for is the wider brief. This guide is the assessment plan on the desk: how to read a unit, choose methods, map them, and leave a trail that holds up when someone asks why the result was competent.

It belongs with the other practice notes in the training and assessment hub. It is not a sales page for an assessment software vendor, and it will not tell you to enrol.

Start with the unit, not the quiz

Open the unit on training.gov.au. Not a Word rewrite your RTO saved in 2019. The live release. Read it in this order, even if it feels backwards.

Assessment conditions first. They tell you what must be in place: workplace or simulated environment, required equipment, timeframes, whether a real customer or a real load is expected. If your plan cannot meet the conditions, you do not have an assessment. You have a classroom approximation, and you need to say so.

Performance evidence next. This is what the person must actually do, how many times, and in what combinations. Knowledge evidence is what they must understand to do it safely and correctly. Elements and performance criteria are the structure of the work. Foundation skills are the reading, writing, oral communication, numeracy and learning demands sitting under the task. Ignore those and you will pass people who cannot read the permit or calculate the mix.

Then check the AQF level of the parent qualification. A Certificate III unit and a Diploma unit can share a topic and still demand different depth. Assessing a level 3 unit as a toolbox quiz is one error. Assessing a level 5 unit as a memory test is another.

The principles and the rules are not decoration

The principles of assessment — fair, flexible, valid and reliable — describe the process. The rules of evidence — valid, authentic, current and sufficient — describe the proof. Both have to be true at once.

Valid, twice, is not a typo. A valid process measures the unit, not a nearby skill you find easier to test. Valid evidence is evidence of that unit, not of general enthusiasm. Authentic means it is the candidate’s work. Current means they can still do it. Sufficient means you have enough coverage, across enough of the criteria, that a reasonable assessor would reach the same judgement.

Fair is the principle new workplace assessors skip. Fair does not mean easy. It means the candidate knew what would be assessed, had a reasonable chance to practise, and was not ambushed by a method that has nothing to do with the job. Flexible means you can gather the evidence in a way that fits the workplace and the candidate — including through recognition of prior learning — without rewriting the standard.

Reliable means two assessors, given the same evidence and the same tool, should not be tossing a coin. That is why validating assessment tools is part of planning, not a once-a-year ritual after the damage is done.

Choose methods that can actually produce the evidence

Methods are not a buffet. Each one produces a different kind of proof.

Observation, questions and products

Observation of the real task is the default for performance evidence. It fails when the checklist lists personality traits instead of performance criteria. Questioning is strong for knowledge evidence and weak as the only method when the unit is about doing. Work samples help when the unit leaves something behind. Third-party reports can support authenticity and currency; “please sign if they’re good” is not a brief.

Simulation is honest when the real task is rare, unsafe or blocked. It is dishonest when it is just cheaper. RPL uses the same rules with a different start: the candidate claims the skill, and you still need evidence that is valid, authentic, current and sufficient. How to gather RPL evidence assessors will accept is the applicant’s half of that mapping.

Map before you write the instrument

Mapping is the unglamorous table that saves you in an audit and, more usefully, in an argument with yourself. Every assessment task should point at specific elements, performance criteria, performance evidence items and knowledge evidence items. If a criterion has no task, you have a gap. If a task has no criterion, you have decoration.

Do not map by keyword. “Communication” in a unit about handing over a patient is not satisfied by a general presentation on listening skills. Read the performance criterion in full. Then write the task so a candidate who can do the job will look competent, and a candidate who can only talk about the job will not.

Cluster units only after you have mapped them separately. Clustering is a delivery and assessment convenience. It is legal and sensible when the job is already clustered — a maintenance shutdown, a service, an admission. It becomes a mess when you bury a weak unit inside a strong one and hope nobody notices the missing knowledge evidence.

Write the instrument after the map. Instructions to the candidate, instructions to the assessor, the observation criteria, the questions, the benchmark or exemplar, and the place where you record the judgement. If the instrument can be used by a colleague who was not in your head, you are closer to reliability.

Plan for dimensions, not just the happy path

The usual dimensions — task skills, task management, contingency, and job or role environment — are a planning prompt, not extra criteria. If you only assess a quiet bench with a perfect kit, you have missed sequencing, the angry client, and the booked forklift.

Mixed groups go sideways here. Experienced workers hide contingency behind speed; new workers look slow and can still be competent. Delivering training to mixed-experience workplace groups covers delivery. On assessment, do not let the fastest person turn the observation into a race.

Build in LLN, reasonable adjustment and a resit path

Foundation skills belong in the plan. If the unit requires a written report, say so in the candidate instructions. If the workplace equivalent is a verbal handover plus a short form, and the unit allows that, assess that. Do not invent a 1,200-word essay because it is easier to mark at the kitchen table.

Reasonable adjustment is planned, not improvised under panic. Extra time, a reader, a larger print job sheet, assessment on a different shift, or oral questions instead of a written paper can all be fair. Deleting a performance criterion cannot. If you are unsure where the line is, language, literacy and numeracy support in vocational training is the companion piece.

Plan the not-yet-competent path before you need it. Who tells the candidate, in plain Australian English, which criteria were not met? What practice is available? How soon can they be reassessed without pretending the first attempt never happened? A plan that only imagines competent results is not a plan. It is optimism.

Leave a record another assessor can follow

Name the units, methods, mapping, conditions, authorised assessors and how results are recorded. Completed tools should show what was seen, asked and produced, and why you judged as you did. “Competent — good bloke” is not a record.

Where ASQA is the regulator, it looks at whether the RTO’s system produces consistent outcomes — not at your individual learner. The same logic holds elsewhere: another professional should reconstruct the judgement. If you cannot explain the plan in ten minutes to a leading hand, it is too clever. Assessment that holds up is usually slightly boring. The excitement belongs in the work, not the cover sheet.