Author:
Wang Huan,Sheng Li,Chen Dechang
Abstract
Statistical tests are often used to detect the difference in survival between two groups. Small p-values, say less than 0.05, are commonly used to declare significant differences. The problem is that p-values do not tell how much the differences are. An alternative is to use effect sizes to detect the difference in survival between two groups. Effect sizes provide numerical numbers to quantify the differences. In this study, we reviewed the effect size ESG that was developed recently by Wang, H., Chen, D., Pan, Q. et al. The effect size ESG is not only unaffected by the change in sample sizes but also applicable no matter if hazards are proportional. We presented some applications of the effect size in comparing different groups of patients with prostate cancer. The results showed that the effect size ESG performed well in detecting and quantifying the difference in survival between two groups.