An Empirical Investigation of Variance Design Parameters for Planning Cluster-Randomized Trials of Science Achievement