Both are good tools to analyse such as data. There is an article that does comparison between these two strategies and others as attached:

https://pubmed.ncbi.nlm.nih.gov › ...

Structural equation modeling with longitudinal data: strategies for ...

SEM is a technique for performing regression with same assumptions of the original regression method. I believe both are good techniques for analysis. However, SEM has an advantage of presenting the mediators, direct effects and indirect effects. Your data requires GSEM (generalized structural equation modelling) since it is longitudinal.

SEM is a technique for performing regression with same assumptions of the original regression method. I believe both are good techniques for analysis. However, SEM has an advantage of presenting the mediators, direct effects and indirect effects. Your data requires GSEM (generalized structural equation modelling) since it is longitudinal.

Mixed random effects modelling is a more flexible strategy than SEM, as you can include continuous time as a predictor variable which gives you more power. The comparison is made in

http://eprints.lse.ac.uk/52203/1/Steele_multilevel_models_longitudinal_2008.pdf

For a short example of what can be done see

http://www.bristol.ac.uk/media-library/sites/cmm/migrated/documents/repeated-measures.pdf

For a very approachable book-length account see

https://www.pilesofvariance.com/

Standard regression is not recommended as the repeated measures induces dependence over time in the response.

Hello Priyadarshni,

I am agreed with Prof. Kelvyn. However, both mixed random effect modeling and SEM/GSEM can be done. But, would you please elaborate more on the nature of outcome variable. SEM is applied on continuous outcome variable while GSEM is applied on categorical. You can also create path diagram through SEM/GSEM. If you want to use STATA. You can read this document. You can contact me for further details or documents if required.

Regards,

Vijay

Thanks everyone for useful insights