In recent years, there has been increasing focus on the issue of multiple hypotheses testing in education evaluation studies. In these studies, researchers are typically interested in testing the effectiveness of an intervention on multiple outcomes, for multiple subgroups, at multiple points in time or across multiple treatment groups. When multiple hypotheses are tested, the probability of committing at least some Type I errors increases and more dramatically with a greater number of tests. Multiple testing procedures (MTPs) adjust p-values for statistical estimates upwards to counteract this problem. MTPs are being increasingly applied in impact evaluations in education. For example, the IES technical methods report, "Guidelines for Multiple Testing in Impact Evaluations," (Schochet, 2008) recommends multiple testing procedures as one of several strategies for dealing with the multiple hypotheses issue. In addition, the What Works Clearinghouse applies a particular procedure, the Benjamini-Hochberg procedure (Benjamini & Hochberg, 1995) to statistically significant findings in studies under review that have estimated effects for multiple measures and/or groups (U.S. Department of Education, 2013). However, an important consequence of making adjustments for multiplicity is a change in the statistical power for detecting true effects. This paper provides critical alternatives to current practice in two ways. First, it presents alternatives to how power is typically defined in studies with multiple tests. Second, for multiple definitions of statistical power under multiplicity, this paper presents methods for estimating power while accounting for p-value adjustments using one of five common multiple testing procedures (MTPs)--Bonferroni (Dunn, 1959, 1961), Holm (Holm, 1979), single-step and step-down versions of Westfall-Young (Westfall and Young, 1995), and Benjamini-Hochberg (Benjamini and Hochberg, 1995) procedures.