Question: a priori power analysis for LiNGAM

[lhl881210]

Dear all

Are there any recommended a priori power analysis methods for LiNGAM to estimate the required sample size before conducting a study? I found that some a priori power analysis methods for SEM do not seem applicable, because they don’t account for the number of latent variables. I would appreciate your help. Thank you!

[sshimizu2006]

I'm not aware of any power analysis methods specifically developed for LiNGAM.

[lhl881210]

Dear Shimizu-sensei

Thank you for the quick reply.

I have a preliminary idea of a priori power analysis for LiNGAM. Maybe this idea is incorrect. Can we assume the most conservative case with the maximum number of possible paths in a DAG, i.e., every element in the strictly lower triangular part of the adjacency matrix is non-zero? Then, based on this specific model, we can calculate the dof and use the semPower2 tool (https://sempower.shinyapps.io/sempower/) to compute the recommended sample size.

I look forward to your opinion. Thank you!