用R进行中介效应分析

中介效应分析用mediation包来完成。分析之前需要先安装mediation包，安装方法，请在R控制台输入：install.packages("mediation")。

脚本与注释

library(mediation)   # 调用mediation包
b <- lm(job_seek ~ sex + age+ econ_hard+ treat, data=jobs)             # 自变量到中介变量的回归方程
c <- lm(depress2 ~ sex + age+ econ_hard+ treat + job_seek, data=jobs)  # 自变量+中介变量到因变量的回归方程
contcont <- mediate(b, c, sims=50, treat="treat", mediator="job_seek") # 采用准贝叶斯近似估计法，也可以改用其他方法
summary(b)                                                             # 显示b的结果
summary(c)                                                             # 显示c的结果
summary(contcont)                                                      # 显示中介效应分析的结果

结果

> summary(b)

Call:
lm(formula = job_seek ~ sex + age + econ_hard + treat, data = jobs)

Residuals:
    Min      1Q  Median      3Q     Max 
-3.1658 -0.4092  0.0577  0.5819  1.1121 

Coefficients:
             Estimate Std. Error t value Pr(>|t|)    
(Intercept)  3.670585   0.125169  29.325   <2e-16 ***
sex         -0.007637   0.048725  -0.157   0.8755    
age          0.004586   0.002319   1.978   0.0482 *  
econ_hard    0.053162   0.024598   2.161   0.0309 *  
treat        0.065615   0.051472   1.275   0.2027    
---
Signif. codes:  0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ‘ ’ 1

Residual standard error: 0.7259 on 894 degrees of freedom
Multiple R-squared:  0.01147,   Adjusted R-squared:  0.007048 
F-statistic: 2.593 on 4 and 894 DF,  p-value: 0.03529

> summary(c)

Call:
lm(formula = depress2 ~ sex + age + econ_hard + treat + job_seek, 
    data = jobs)

Residuals:
    Min      1Q  Median      3Q     Max 
-1.5192 -0.4461 -0.1503  0.3597  2.7568 

Coefficients:
              Estimate Std. Error t value Pr(>|t|)    
(Intercept)  2.2076414  0.1480548  14.911  < 2e-16 ***
sex          0.1068049  0.0411471   2.596   0.0096 ** 
age          0.0006489  0.0019625   0.331   0.7410    
econ_hard    0.1485434  0.0208269   7.132 2.04e-12 ***
treat       -0.0402647  0.0435059  -0.926   0.3550    
job_seek    -0.2399550  0.0282433  -8.496  < 2e-16 ***
---
Signif. codes:  0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ‘ ’ 1

Residual standard error: 0.613 on 893 degrees of freedom
Multiple R-squared:  0.1203,    Adjusted R-squared:  0.1154 
F-statistic: 24.43 on 5 and 893 DF,  p-value: < 2.2e-16

> summary(contcont)

Causal Mediation Analysis 

Quasi-Bayesian Confidence Intervals

                Estimate 95% CI Lower 95% CI Upper p-value
ACME           -0.017484    -0.041055     0.000761    0.12
ADE            -0.044809    -0.132919     0.025923    0.28
Total Effect   -0.062292    -0.136771     0.012892    0.12
Prop. Mediated  0.278977    -0.374172     2.677228    0.16

Sample Size Used: 899 


Simulations: 50