Hi,
I wish to know the reason why i get different mediated effect while using the Stata `mediate` command with the following specifications:
mediate (outcome cov1 cov2, logit) (mediator cov1 cov2, logit) (exposure), vce(cluster var) nointeraction
VERSUS
The traditional way:
logit outcome exposure cov1 cov2
estimate store M1
logit outcome exposure mediator cov1 cov2
estimate store M2
suest M1 M2, vce (cluster var)
lincom_b[M1_outcome: exposure]-_b[M2_outcome: exposure], eform
I wish to know the reason why i get different mediated effect while using the Stata `mediate` command with the following specifications:
mediate (outcome cov1 cov2, logit) (mediator cov1 cov2, logit) (exposure), vce(cluster var) nointeraction
VERSUS
The traditional way:
logit outcome exposure cov1 cov2
estimate store M1
logit outcome exposure mediator cov1 cov2
estimate store M2
suest M1 M2, vce (cluster var)
lincom_b[M1_outcome: exposure]-_b[M2_outcome: exposure], eform