Despite availability of software, power analysis has not become standard practice among researchers in psychology. In this paper, the relationships among power, significance level, sample size, and effect size are discussed. Actual research data are used to illustrate that the power of a test is increased with an increase in effect size, a decrease in significance level, and an increase in sample size. Strategies for power analysis are described. Recommendations are discussed, including choosing a sample size that ensures adequate power for a planned study, pursuing research on effects with large magnitudes rather than with only small magnitudes, and reporting effect sizes to enable readers to better gauge the importance of research results.