2012年5月4日 星期五

研究設計與統計方法


Continuous variables 連續變項
Student's t test = paired t test = repeat test :two dependent sample, or two sample's variances are equal

unpaired t test = independent t test:two independent samples, or two sample's variances are unequal

One way ANOVA used for 3 or more  independent variables


non-parametric analysis(用於n<30)
Wilcoxon rank sum test 排序總和檢定(or Mann-Whitney U, 小樣本  兩組樣本皆<10, 或大樣本  任一組樣本數>10 ) used for  two independent samples
Wilcoxen signed-rank test  used for  two dependent non-parametric samples
Kruskal-Wallis test used for 3 or more  independent non-parametric samples
Freidman test  used for 3 or more  dependent non-parametric samples


Binominal or categorical variables 二項式或類別變項
Chi-square used in large one sample (compared with hypothesis)
Pearson Chi-square (goodness of fit test) for two independent larger groups, Yate's continuity correction p value)
Fisher exact used in small independent samples (non-parametric analysis)
McNemar's test for two small dependent groups (non-parametric analysis)


相關之統計方法







參考資料:

1. Wilcoxon rank sum test http://homepage.ntu.edu.tw/~clhsieh/biostatistic/10/10-2.htm

2. 研究設計與統計方法  
http://www1.cgmh.org.tw/intr/intr2/ebmlink/doc/%E7%B5%B1%E8%A8%88%E6%95%99%E5%AD%B89411.pdf

沒有留言:

張貼留言