Dear esteemed Stata Users
Re: How to perform mediation analysis on a unbalanced panel dataset (using *medsem* command - in STATA 17 or *mediation* command- in STATA 18?)
Can one perform mediation analysis on an unbalanced panel dataset using the *medsem* or *mediation* command and add several controls?
I understand that one can do so on multilevel data based on the following sites
https://stats.oarc.ucla.edu/stata/fa...data-method-1/
https://stats.oarc.ucla.edu/stata/fa...data-method-2/
That said, I have an unbalanced panel dataset, where I’ve collected three variables by firm.
Let’s say there are 1,000 firms in my unbalanced dataset
From my understanding, one can perform mediation analysis in STATA using the medsem package.
In STATA 18, there is a new *mediation* command for .
I have the following variables that I have collated, where I have collected it by firm
M (moderator) variable = Capital structure
X (independent) variable – Corporate Governance
Y (dependent) variable – Firm performance
Control variables: FirmAge, Firm size, growth
I’m having trouble in ascertaining whether the direct and indirect effects can be determined in an unbalanced setting by either using the mediation or medsem package?
Can someone please advice as to the appropriateness of the *mediation* or *medsem* command on an unbalanced panel dataset where you can add controls to the regression models?
Thanks,
R
Re: How to perform mediation analysis on a unbalanced panel dataset (using *medsem* command - in STATA 17 or *mediation* command- in STATA 18?)
Can one perform mediation analysis on an unbalanced panel dataset using the *medsem* or *mediation* command and add several controls?
I understand that one can do so on multilevel data based on the following sites
https://stats.oarc.ucla.edu/stata/fa...data-method-1/
https://stats.oarc.ucla.edu/stata/fa...data-method-2/
That said, I have an unbalanced panel dataset, where I’ve collected three variables by firm.
Let’s say there are 1,000 firms in my unbalanced dataset
From my understanding, one can perform mediation analysis in STATA using the medsem package.
In STATA 18, there is a new *mediation* command for .
I have the following variables that I have collated, where I have collected it by firm
M (moderator) variable = Capital structure
X (independent) variable – Corporate Governance
Y (dependent) variable – Firm performance
Control variables: FirmAge, Firm size, growth
I’m having trouble in ascertaining whether the direct and indirect effects can be determined in an unbalanced setting by either using the mediation or medsem package?
Can someone please advice as to the appropriateness of the *mediation* or *medsem* command on an unbalanced panel dataset where you can add controls to the regression models?
Thanks,
R
