Effect Sizes and Optimum Significance Levels

_ In statistical analysis, standardized difference of means among conditions is represented by effect size. Corresponding to effect sizes, probability distributions of alternative hypotheses are given. So, effect sizes and significance levels are important components of statistical analysis. For a given effect size, a significance level determines probabilities of statistical results, e.g., power of the test. The significance level should be determined by an objective criterion. One method to set objective significance level is by utility. Expected utilities of significance levels can be estimated for effect sizes. Expected utilities of significance levels are dependent on many factors. In this report, dependency of the optimum significance level on effect sizes and on utilities of false alarm were studied. For details, check the file e_size_s_level.pdf in the box under the Files tab. _