統計散記-Vivian 經驗分享

今天審稿時，看到一篇文章: 3組之間連續值比較直接做兩兩比較，所以有3個P-value: p of A vs B, P of  A vs C, P of B vs C，這時問題來了~

1. p of A vs B, P of  A vs C, P of B vs C 是否有經過Bonferroni or Dunn-Sidak adjustment?

   (看統計方法及Table note 均無說明，應該是沒做修正了~)

2.應該要再提供一個完整的比較: Overall p-value，通常 Overall p-value 顯著了，才會再進一步做事後檢定

3.此篇探討的是Post hoc comparisons for the Kruskal-Wallis test，是針對3組以上連續資料無母數分析之案例，若是針對3組以上連續資料母數分析，應採用ANOVA and 相應的事後檢定分析

4.那類別資料又該如何做事後檢定呢?

可參考下列SPSS官方文章及找到的一篇paper: 

1.Post hoc comparisons for the Kruskal-Wallis test

https://www.ibm.com/support/pages/post-hoc-comparisons-kruskal-wallis-test)

2.Suo X, Lei D, Li N, Peng J, Chen C, Li W, Qin K, Kemp GJ, Peng R, Gong Q.

Brain functional network abnormalities in parkinson's disease with mild cognitive impairment. Cereb Cortex. 2022;32(21):4857-4868.

-P-values by post hoc Sidak correction for three comparisons unless otherwise indicated. (Table 1)

-For the post hoc tests, P < 0.05/3 was considered significant (Table 2)

延伸閱讀:

 連續變項vs.類別變項的統計流程(The flow chart of comparison between two or more than two groups)