A large school district wants to estimate the average GPA of all 15,000 students using a budget of $5,000. Two sampling strategies are available: (1) Simple random sampling at$$5$per student, (2) Cluster sampling (schools as clusters) at$$1$ per student. If within-school GPA variation is much larger than between-school variation, which strategy is preferable and why?