Boostrapping分析方法

使用Bootstrapping計算Effect size的方法
以下為使用R軟體

> data=read.table(file.choose(),sep=',', header=TRUE)
> data
        X1      X2 Y1 Y2
1  4.33964 7.42030 22 33
2  7.17865 8.70926 33 34
3  7.40154 7.40154 30 32
4  1.06128 4.75095  8 23
5  2.69063 7.16065 11 29
6  3.74045 7.02058 11 31
7  6.80245 9.39384 31 36
8  4.33964 8.70926 20 33
9  1.06128 4.33964  9 22
10 2.69063 4.33964  6 19
11 6.19895 7.74492 32 35
12 7.44349 7.72457 28 34
13 6.52419 7.72457 32 35
14 6.90474 7.72457 34 34
15 0.03023 3.66253  2 18
16 6.19895 8.05381 30 34
17 2.69063 7.12063  9 30
18 3.74045 9.39384 14 36
19 7.42030 7.40154 33 32
20 3.44653 6.52419 14 32
21 1.93182 7.17865 11 33
22 4.06875 7.13855 18 31
23 5.37969 7.72457 30 32
24 2.99228 5.37969  5 22
25 0.03023 4.33964  2 20
26 3.44653 4.33964 11 19
27 6.10720 6.86523 20 31
28 0.03023 4.33964  4 20
29 2.69063 4.33964 15 23
30 5.14221 7.44349 27 33
31 3.44653 5.65632 13 30
32 3.44653 6.13888 11 28
33 3.44653 5.37969 12 27
34 6.15484 7.13855 25 32
35 3.66253 5.53187 22 23
36 5.31453 6.86523 29 30
37 0.03023 4.01905  3  8
38 6.13888 7.50640 31 33
39 2.69063 5.05935 12 28
40 6.13888 7.44349 27 34
41 3.66253 7.53005 14 34
42 6.12336 6.86523 30 30
43 1.06128 2.69063  5  4
44 3.66253 6.13888 19 31
45 2.99228 6.90474 14 33
46 5.48398 7.23850 24 34
47 5.37969 7.83892 33 34
48 6.13888 7.83892 32 34
49 8.15931 8.70926 34 34
50 6.86523 7.09863 31 32
51 1.06128 2.69063  1  5
52 1.06128 4.33964  5 26
53 7.79237 7.79237 32 32
54 3.17450 7.23850  9 33
55 3.66253 5.53187 16 21
56 3.66253 4.33964 15 19
57 2.24595 4.33964 13 19
58 5.37969 7.72457 29 33
59 4.33964 5.53187 20 25
60 3.66253 5.59073 21 28
61 6.72564 8.05381 31 35
62 5.53187 6.13888 25 30
63 6.13888 8.74302 30 36
64 9.39384 9.39384 36 36
65 3.66253 6.10720 18 29
66 4.75095 6.13888 24 30

ES=matrix(0,1000,3)
for (i in 1:1000){
x=sample(66,replace=T)
data1=data[x,]
diff1=data1[,2]-data1[,1]
diff2=data1[,4]-data1[,3]
sd1=sd(data1[,1])
sd2=sd(data1[,3])
m1=mean(diff1)
m2=mean(diff2)
ef1=m1/sd1
ef2=m2/sd2
es=ef1-ef2
ES[i,1]=ef1
ES[i,2]=ef2
ES[i,3]=es
}
> quantile(ES[,3],c(0.025,0.975))
       2.5%       97.5%
-0.01287047  0.27372242

註：使語法為設定進行1000個boostrap samples，有66筆資料。X1和X2表示X工具第一次和第二次的測量分數，Y1和Y2表示Y工具第一次和第二次的測量分數。