A researcher wants to study the effect of a school-based mindfulness program on student anxiety. She randomizes schools, not students. What is the primary reason for this choice?
ATo increase statistical power by using a larger unit of randomization
BBecause the intervention is delivered at the school level and individual randomization within a school would cause contamination
CBecause students within a school are too heterogeneous for individual randomization to work
DTo avoid needing informed consent from individual participants
When an intervention is delivered by school staff and affects the entire school environment, randomizing individuals within the same school is not feasible — the treatment affects everyone in the school, not just those 'assigned' to it. Contamination (treated and control students interacting, or teachers behaving differently toward different students) would also undermine the comparison. School-level randomization is the natural choice.
Question 2 Multiple Choice
In a cluster randomized trial, a high intraclass correlation coefficient (ICC) implies:
AIndividuals in different clusters are more similar to each other than individuals in the same cluster
BCluster size does not affect statistical power
CA larger design effect, meaning you need more participants to achieve the same power as an individually randomized trial
DThe standard errors from individual-level analysis will be too large
A high ICC means individuals within clusters are similar (much of the variance is between clusters, not within), which reduces the amount of independent information each additional individual contributes. The design effect 1 + (m−1)×ICC increases with ICC, inflating required sample size. Standard errors from naive individual-level analysis will be too *small* (not too large), because the analysis wrongly assumes independence.