Obtaining bootstrap CI's around goodness of fit measures when performing SEM using STATA (fun extra)
HTML-код
- Опубликовано: 11 сен 2024
- This video is mainly for Stata nerds like myself (who for some reason) really like syntax :)
I provide a walkthrough of a strategy for obtaining bootstrap confidence intervals around goodness of fit measures when you are performing SEM using Stata Version 17. The example involves a path analysis model. In the video I talk through a do-file where I go over code I wrote for the model I was running; however, you can also generate the code necessary for bootstrapping by using graphical user interface (GUI). Simply take the code from the output (with any remaining modifications to make sure the syntax is laid out correctly) and plug it into the do-file.
You can download a copy of the data here:
drive.google.c...
You can obtain a copy of the referenced do-file here:
drive.google.c...
A little side note: It is possible for TLI to be negative or greater than 1 (see e.g. Shi et al. (2018); journals.sagep.... As such, it is possible that you could end up with a lower bound for your confidence interval that is negative or upper bound greater than 1.
*Shi, D., L.A., Lee, & Maydeu-Olivares, A. (2018). Understanding the model effect size on SEM fit indices. Educational and Psychological Measurement, 79, 310-334.
Thanks for the video. I see a Stata 17 upgrade from past videos. The ML models finally fly : )
I really like your videos and find them very helpful. So I was wondering if you also have a video on how to use fixed or random effects while using SEM?