For this hypothetical paper exercise, imagine independent groups watching either a praised model or a neutral model before completing the same ten-minute task. This is not an instruction to experiment on pupils.
The independent variable is which model condition is shown. The dependent variable is the number of answer-checking actions recorded during the task. Define what counts as one check so the measure is repeatable. Keeping the task and observation period the same is a control; it does not remove every possible difference between groups. A design answer should also consider consent, withdrawal and protection from harm.
Invented check counts: praised-model group: 2, 3, 3, 4; neutral-model group: 1, 1, 2, 4. The first mean is (2 + 3 + 3 + 4) ÷ 4 = 3 checks; its range is 4 − 2 = 2 checks. The second mean is 8 ÷ 4 = 2 checks; its range is 4 − 1 = 3 checks.
A useful interpretation is: “The praised-model group checked more on average in these data, while the neutral-model group had a wider range.” These descriptive values alone do not establish statistical significance or prove a causal explanation. To choose an inferential test, establish the question, design and measurement level, then use your board’s requirements.
Retest: Replace the last neutral-group score with 8. Its mean becomes 3 and range 7. The means now match, but the spread does not: reporting only the mean would hide that difference.