teffects for SEM with clustered errors

[Jennifer Nelson]

I am having trouble estimating teffects, in order to get indirect and total effects, after I estimate a SEM with standardized coefficients and clustered robust standard errors (in the case of my data, clustering by school). When I run estat teffects after a standardized SEM estimation without clustering, the standardized direct effects given by the SEM estimation and teffects match. However, when I do the same operation following a clustered SEM, the standardized coefficients for direct effects do not match those in the SEM. Please advise.

[Richard Williams]

It might help if you showed us some of your code and output. Use code tags (see pt 12 of the FAQ).

[Jennifer Nelson]

Thanks, Richard.

I was able to resolve my problem, but will follow the format next time I post.

[Richard Williams]

Good. In case anyone else ever has the same problem, you might post your solution.

[Cristina Santos]

Dear Richard,

I am actually having a similar problem to Jennifer, above.

When I run "estat teffects, standardized" after clustering my SE on SEM, the reported direct effects are a match to the unstandardized solution of clustering my data and not to the standardized coefficients of the strucutural model. Is this because I am clustering data?


I am only able to get the values for the standardized effects if I use the commands return list and matrix list r(direct_std)