Hello,
i am new to STATA and quantitative analysis. I have been researching on how to use STATA to do my mediation analysis for a while now. I have a binary dependent variable and two metric mediators. That's why I can't use SEM, unfortunately. With medeff I have found an extension that would be suitable for me. However, I always get the message 'missing values generated' right before the 'Effect' table. Find the Output below (red marked):
(1,060 missing values generated)
(1,060 missing values generated)
(1,060 missing values generated)
------------------------------------------------------------------------------------
Effect | Mean [95% Conf. Interval]
-------------------------------+----------------------------------------------------
ACME1 | -.0022841 -.0031201 -.0014768
ACME0 | -.002289 -.0031259 -.0014815
Direct Effect 1 | -.0021875 -.003971 -.0005806
Direct Effect 0 | -.0021923 -.0039754 -.0005832
Total Effect | -.0044764 -.0062831 -.0028074
% of Total via ACME1 | .5122697 .3635329 .813602
% of Total via ACME0 | .5133602 .3643067 .8153339
Average Mediation | -.0022865 -.003123 -.0014791
Average Direct Effect | -.0021899 -.0039732 -.0005819
% of Tot Eff mediated | .512815 .3639198 .8144679
------------------------------------------------------------------------------------
Can someone explain to me what is meant by this and how it happens? I have checked all variables, there are no more missing values and the two regressions that medeff calculates before also have the same N.
If anyone knows a better way to do my mediation analysis, I'll gladly take that too.
I have found paramed as an alternative, but there I do not get ACME's (Average Causal Mediation Effect).
Thanks a lot.
i am new to STATA and quantitative analysis. I have been researching on how to use STATA to do my mediation analysis for a while now. I have a binary dependent variable and two metric mediators. That's why I can't use SEM, unfortunately. With medeff I have found an extension that would be suitable for me. However, I always get the message 'missing values generated' right before the 'Effect' table. Find the Output below (red marked):
(1,060 missing values generated)
(1,060 missing values generated)
(1,060 missing values generated)
------------------------------------------------------------------------------------
Effect | Mean [95% Conf. Interval]
-------------------------------+----------------------------------------------------
ACME1 | -.0022841 -.0031201 -.0014768
ACME0 | -.002289 -.0031259 -.0014815
Direct Effect 1 | -.0021875 -.003971 -.0005806
Direct Effect 0 | -.0021923 -.0039754 -.0005832
Total Effect | -.0044764 -.0062831 -.0028074
% of Total via ACME1 | .5122697 .3635329 .813602
% of Total via ACME0 | .5133602 .3643067 .8153339
Average Mediation | -.0022865 -.003123 -.0014791
Average Direct Effect | -.0021899 -.0039732 -.0005819
% of Tot Eff mediated | .512815 .3639198 .8144679
------------------------------------------------------------------------------------
Can someone explain to me what is meant by this and how it happens? I have checked all variables, there are no more missing values and the two regressions that medeff calculates before also have the same N.
If anyone knows a better way to do my mediation analysis, I'll gladly take that too.
I have found paramed as an alternative, but there I do not get ACME's (Average Causal Mediation Effect).
Thanks a lot.
Comment