ABSTRACT
INTRODUCTION Precision analyses (also known as accuracy in parameter estimation) focus on the width of confidence intervals (CIs). Precision analysis provides information that supplements power analyses and in some cases is more appropriate to research goals. Power analysis determines the likelihood of rejecting a null hypothesis given a particular population effect size, sample size, and Type I error rate. However, rejecting the null hypothesis is only half of the story. Another important issue is what range of values is reasonable to expect for the population given the sample result. A CI provides this information but can be very wide or very narrow. The wider the confidence limits are, the less precise the result is. We can design for more precise (i.e., narrower) CIs around effect sizes or raw values (e.g., mean differences). However, increasing precision requires larger samples or a better design to reduce error variability.
