The School Canteen 'Sugar Crash' Survey
A Year 11 class wants to test the claim that eating a sugary snack at lunch makes students sleepy in period 5. They survey 60 students who volunteered after an announcement. Students self-report what they ate at lunch and rate their period-5 sleepiness from 1 (alert) to 10 (very sleepy). The class finds that students who ate a sugary snack reported a mean sleepiness of 6.8, while those who did not reported 5.1. The teacher notes that the survey was run on a Friday after a school athletics carnival, and that students who stayed home from the carnival were more likely to complete the survey. The class concludes that sugar causes sleepiness and wants to submit it to a science fair.
1. Identify the independent and dependent variables in this investigation and explain why the design cannot support the conclusion that sugar causes sleepiness.
Read the worked answer
Variables: The independent variable is whether the student ate a sugary snack at lunch (self-reported, two categories). The dependent variable is the self-reported sleepiness rating in period 5 on a 1–10 scale.
Why causation fails:
- There is no manipulation or control. Students chose their own lunch, so the groups differ in many ways besides sugar (e.g. sleep the night before, activity level, whether they attended the carnival).
- The outcome is self-reported, so expectation and recall bias can inflate the difference.
- The sample is a volunteer sample, not random, so it may not represent the school population.
- The timing (after an athletics carnival) introduces a confounding variable: fatigue from the carnival could raise sleepiness in both groups unevenly.
- Selection bias is likely because students who skipped the carnival were more likely to respond, and these students may differ systematically (e.g. less physically tired).
A correlation of 6.8 versus 5.1 is consistent with many explanations, including reverse causation (tired students choose sugary snacks for energy). The design can suggest a hypothesis, but it cannot isolate sugar as the cause.
2. Propose two specific improvements to the design and explain what each improvement controls for.
Read the worked answer
Improvement 1 - Randomised allocation with a controlled snack. Randomly assign consenting students to either a standardised sugary snack or an identical-tasting low-sugar snack at the same time, with the same total energy content. This controls for self-selection and for confounding by other lunch components, because the only systematic difference between groups is sugar content.
Improvement 2 - Blinding and objective timing. Do not tell students which snack they received, and measure sleepiness with a simple reaction-time task plus a self-rating at fixed intervals (e.g. 30, 60 and 90 minutes after eating). This controls for expectation effects and standardises when measurements are taken, removing the variability caused by different period-5 schedules or carnival fatigue.
Supporting changes: recruit a random sample of the year group rather than volunteers, run the trial on a normal school day, and record sleep hours the night before as a covariate. Together these reduce selection bias, confounding and measurement bias, making a causal claim more defensible.