The Tutoring Program Trial
A school introduces a peer-tutoring program for Year 10 students. Forty students volunteer; the coordinator assigns the first twenty on the alphabetical roll to tutoring and the remaining twenty to a waitlist. After eight weeks, tutored students' mean test score rises by 9 marks, while the waitlist rises by 3. The coordinator announces the program 'works' and plans to roll it out school-wide. A teacher notes that several tutored students had asked to join because they were already motivated to improve, and that tutors were the school's top Year 12 students, who sometimes completed parts of the worksheets. The coordinator wants a defensible interpretation before spending the budget.
1. Evaluate the strength of the coordinator's causal claim and identify the design features that limit it.
Read the worked answer
The claim is weak as stated. Assignment by alphabetical order is not random with respect to motivation, so the groups may differ before the program begins. The 6-mark difference is consistent with a program effect but also with pre-existing differences and with tutor assistance on worksheets, which contaminates the independent variable.
The design also lacks blinding: students and coordinator knew who was tutored, so expectancy could influence effort and scoring. The waitlist comparison is better than no control, but a waitlist group that knows it will receive tutoring may show less effort, inflating the difference.
A defensible conclusion is that the program is promising but unproven. To support rollout, the coordinator should randomly assign volunteers, standardise what tutors may do, use pre-test scores as a covariate, and ideally have markers unaware of condition.
2. Explain how the school could redesign the study to reduce selection bias, and state one ethical consideration.
Read the worked answer
Random allocation of the forty volunteers to tutoring or waitlist would distribute motivation and prior achievement roughly evenly, so post-test differences could be attributed more confidently to the program. Pre-testing all students and reporting change scores, or statistically controlling baseline scores, would further reduce bias. Tutors should be given a script and barred from completing work.
Ethically, a waitlist design is preferable to withholding tutoring indefinitely, because all volunteers eventually receive it. The school should obtain informed consent from students and caregivers, keep test results confidential, and ensure no student is identifiable in reporting.