Overview
Power analysis is frequently treated as a checkbox — entering α=0.05, power=0.80, and Cohen's d=0.50 into a calculator and reporting the resulting N. The critical missing step is justifying the effect size: d=0.50 is "medium" by Cohen's conventions, but in many applied domains, a d=0.50 effect is either unrealistically large (making the study undemanding) or irrelevantly large (a d=0.10 effect has clinical significance in pharmacology). The effect size must be anchored to the domain's threshold of practical significance — not to Cohen's generic benchmarks.
The Statistical Power Analysis Framework justifies effect sizes from domain-specific thresholds, calculates required sample sizes across a range of plausible effect sizes, and designs the sample acquisition strategy that achieves the target power within the study's resource constraints.