In Chapter 1, we noted that there are two general ways that power analysis might be used. First, power analysis is an extremely useful tool for planning research. Critical decisions, such as how many subjects are needed, whether multiple observations should be obtained from each subject, and even what criterion should be used to define “statistical significance” can be better made by taking into account the results of a power analysis. Decisions about whether or not to pursue a specific research question might even depend on considerations of statistical power. For example, if your research idea involves a small (but theoretically meaningful) interaction effect in a complex experiment, power analysis might show that thousands of subjects would be needed to have any reasonable chance of detecting the effect. If the resources are not available to test for such an effect, it is certainly better to know this before the fact than to learn it after collecting your data.