Current practice for conducting power analyses in hierarchical trials using survey based ICC and effect size estimates may be misestimating power because ICCs are not being adjusted to account for treatment effect heterogeneity. Results presented in Table 1 show that the necessary adjustments can be quite large or quite small. Furthermore, power estimates without adjusting the ICC could be either too large or too small, depending on the covariance of school specific treatment effects with control group school means. The paper illustrates the need for the field to obtain better empirical evidence about likely values of this parameter in order to conduct accurate power analyses. One table is appended.