1 Star 0 Fork 76

wujun0329/Causal_Inference_book

加入 Gitee
与超过 1200万 开发者一起发现、参与优秀开源项目,私有仓库也完全免费 :)
免费加入
文件
该仓库未声明开源许可证文件(LICENSE),使用请关注具体项目描述及其代码上游依赖。
克隆/下载
chapter11.do 3.43 KB
一键复制 编辑 原始数据 按行查看 历史
/***************************************************************
Stata code for Causal Inference: What If by Miguel Hernan & Jamie Robins
Date: 10/10/2019
Author: Eleanor Murray
For errors contact: ejmurray@bu.edu
***************************************************************/
/***************************************************************
PROGRAM 11.1
Figures 11.1, 11.2, and 11.3
Sample averages by treatment level
***************************************************************/
**Figure 11.1**
*create the dataset*
clear
input A Y
1 200
1 150
1 220
1 110
1 50
1 180
1 90
1 170
0 170
0 30
0 70
0 110
0 80
0 50
0 10
0 20
end
*Build the scatterplot*
scatter Y A, ylab(0(50)250) xlab(0 1) xscale(range(-0.5 1.5))
*Output the mean values for Y in each level of A*
bysort A: sum Y
*Save the data*
save fig1, replace
**Figure 11.2**
clear
input A Y
1 110
1 80
1 50
1 40
2 170
2 30
2 70
2 50
3 110
3 50
3 180
3 130
4 200
4 150
4 220
4 210
end
scatter Y A, ylab(0(50)250) xlab(0(1)4) xscale(range(0 4.5))
bysort A: sum Y
save fig2, replace
**Figure 11.3**
clear
input A Y
3 21
11 54
17 33
23 101
29 85
37 65
41 157
53 120
67 111
79 200
83 140
97 220
60 230
71 217
15 11
45 190
end
scatter Y A, ylab(0(50)250) xlab(0(10)100) xscale(range(0 100))
save fig3, replace
/***************************************************************
PROGRAM 11.2
2-parameter linear model
Creates Figure 11.4,
parameter estimates with 95% confidence intervals from Section 11.2,
and parameter estimates with 95% confidence intervals from Section 11.3
***************************************************************/
**Section 11.2: parametric estimators**
*Create the data, or reload with the command "use fig3, clear"*
clear
input A Y
3 21
11 54
17 33
23 101
29 85
37 65
41 157
53 120
67 111
79 200
83 140
97 220
60 230
71 217
15 11
45 190
end
*Plot the data*
scatter Y A, ylab(0(50)250) xlab(0(10)100) xscale(range(0 100))
*Fit the regression model*
regress Y A, noheader cformat(%5.2f)
*Output the estimated mean Y value when A = 90*
lincom _b[_cons] + 90*_b[A]
*Plot the data with the regression line: Fig 11.4*
scatter Y A, ylab(0(50)250) xlab(0(10)100) xscale(range(0 100)) || lfit Y A
save fig3, replace
**Section 11.3: non-parametric estimation*
*Create the data, or reload with the command "use fig1, clear"*
clear
input A Y
1 200
1 150
1 220
1 110
1 50
1 180
1 90
1 170
0 170
0 30
0 70
0 110
0 80
0 50
0 10
0 20
end
*Fit the regression model*
regress Y A, noheader cformat(%5.2f)
*E[Y|A=1]*
di 67.50 + 78.75
/***************************************************************
PROGRAM 11.3
3-parameter linear model
Creates Figure 11.5 and
Parameter estimates for Section 11.4
***************************************************************/
*Create the data, or reload with the command "use fig3, clear"*
clear
input A Y
3 21
11 54
17 33
23 101
29 85
37 65
41 157
53 120
67 111
79 200
83 140
97 220
60 230
71 217
15 11
45 190
end
*Create the product term*
gen Asq = A*A
*Fit the regression model*
regress Y A Asq, noheader cformat(%5.2f)
*Output the estimated mean Y value when A = 90*
lincom _b[_cons] + 90*_b[A] + 90*90*_b[Asq]
*Plot the data with the regression line: Fig 11.5*
scatter Y A, ylab(0(50)250) xlab(0(10)100) xscale(range(0 100)) || qfit Y A
clear
马建仓 AI 助手
尝试更多
代码解读
代码找茬
代码优化
其他
1
https://gitee.com/wujun0329/Robins.git
git@gitee.com:wujun0329/Robins.git
wujun0329
Robins
Causal_Inference_book
master

搜索帮助