"A priori" power analyses allow researchers to estimate the number of participants needed to detect any effects of an intervention. However, power analyses are only as valid as the parameter estimates used. Effect sizes vary greatly depending on several study characteristics, including the nature of the intervention, the developer of the outcome measure, and the age of the participants. Researchers should have an awareness of how an effect size obtained in a pilot study might deviate from what is reasonable to expect in a cluster randomized trial. Our meta-analysis examines the relationship between science education intervention effect sizes and a host of study characteristics, allowing primary researchers to access better estimates of effect sizes for "a priori" power analyses.