A Comparison of Methods for Estimating Confidence Intervals for Omega-Squared Effect Size
Statistics testing is an approach to let us know whether models or specific parameter estimates are significant or not. However, outcome of significance involves sample size, meaning sample size once large, statistics testing tends to report significance. Therefore, we need an effect size which not being affected by sample size and moreover can provide us information about variables’ effect. That is why effect size so important to report. Back to the paper, authors conducted a series of simulations for comparing three confidence interval calculation methods applied in an effect size index of omega-square. Results illustrated that sample size and magnitude of the effect influence coverage rates and interval width. Among three methods, the percentile bootstrap method had the widest intervals and higher coverage rates in the smaller effect size conditions.
Questions:
1) It is not easy to understand the simulation conditions due to the fact that no setting details were depicted in the paper.
2) From table 3, the larger the sample size the smaller the confident interval width would be. Does it mean that sample size would influence stand error, so as CI width would reduce with larger sample size?