Writing knowledge checks and scenarios that test decisions

The recognition trap

A typical training quiz: "Which of these is a step in the closing procedure?" The worker glances back at the bullet points and picks the matching one. They passed. Within a week they have forgotten it, because nothing asked them to use it.

Take a food safety module that ends with "At what temperature should fresh chicken be stored?", options 4°C, 7°C, 10°C and 15°C. The learner read "at or below 4°C" three paragraphs earlier and taps 4°C. That is recognition, the easiest operation there is, and passing it says very little about whether they will act on it.

Replace it with: "You are putting away a delivery and the cooler reads 8°C. The chicken pallet is inside. What do you do first?" Now the learner has to retrieve the limit from memory, notice that 8°C is above it and decide what to do. Same content, completely different mental work.

Research on the testing effect points the same way: recalling information from memory produces better long-term retention than reading it again (Roediger and Karpicke, 2006; Karpicke and Roediger, 2008).

What makes a check stick

The feedback is the second teaching moment, often more memorable than the module itself. A wrong answer followed by one clear sentence of explanation is worth more than a right answer followed by a score.

Scenarios: questions that test decisions

A frontliner does not need to recite the food safety procedure. They need to know what to do when a colleague skips the temperature check on a busy Saturday. A scenario puts them in that moment: a realistic situation, a small set of options and feedback that explains the principle behind the right one.

Scenarios fit judgment: customer situations, what to do when you see a rule broken, when to stop the line, when to escalate. For a pure procedure, such as operating the dishwasher or counting down the till, show the procedure and have them do it. A scenario is overkill there.

What to avoid

• Trick questions. Negative phrasing ("Which of these is NOT a step?"), double negatives, options that are right on a technicality. They test test-taking, not the job.
• Long stems. A question that takes ninety seconds to read is a reading test, and on a mixed shift it tests language first.
• True or false on decisions that matter. Half the options, a coin toss's worth of evidence.
• The score as a grade. A pass mark with consequences teaches people to click until they pass.

Ask again, a few shifts later

Where a check sits matters as much as how it is written. At the end of the module it confirms what was just covered, but on its own it mostly produces short-term memory. The version that lasts asks again later, from memory: spacing retrieval over time beats cramming the same amount into one sitting (Cepeda et al., 2006). Spaced repetition is the name for it.

On a frontline floor, count in shifts, not calendar days, because the roster is what people actually live by. Ask the key question again on the person's next shift, a few shifts later and a few weeks later: in the handover, in the toolbox talk or as a one-question check on the phone. Two further placements are worth knowing. A question before the module surfaces what people think they already know, which is useful where misconceptions are common. And the same question asked across all sites shows whether the content landed everywhere.

Spacing is wasted in three places: a procedure people use every shift (daily use is the spacing), reference content that should be looked up rather than remembered, such as an allergen matrix, and content that changes every week.

Reading the results

Check results are a signal about the training, not only about the learner.

A result is not a competence assessment. It shows the content landed; where a standard asks for demonstrated competence, the observed task stays on the floor, as covered in HACCP training in food production.

How Aristotl handles it

Every Aristotl course carries a knowledge check, built from your own procedures and instructions and reviewed by your team before it is published. Courses take 3-5 minutes on the worker's phone, so the check comes while the content is fresh, and results are recorded per person. Writing the decision scenarios is still your team's work: the rule, the tempting mistake and the moment on the floor come from people who know the job.

Take your most recent quiz and mark every question whose answer was visible on the screen before it. Rewrite the worst one as a moment from a real shift.