Dear Statalisters,

I hope you have all been healthy and well. I have a question that may require the expertise of someone who is familiar with the development of Stata, perhaps someone from StataCorp?

I have been analyzing a dataset at an on-site data analysis center from which the data cannot be exported. My option of software has therefore been limited to what they have installed on the computers at the said center. Recently they updated their Stata 13 to 16, and when I re-run my SEM model with bootstrapping, I noticed that the bootstrapped standard error and the 95% CI produced by Stata were different from what I obtained during my previous visits when I was still using Stata13.

I am therefore wondering if there has perhaps been a change in Stata 16 regarding the algorithm or estimator used to run bootstrapping, which may be responsible for the different estimates I got from using the two versions. I have checked my commands and they are the same as the ones I used. Sadly I am not able to further validate my assumption by re-running my command in Stata 13 as it is now updated to 16.
If anyone can offer any information or suggestion on this I will be extremely grateful!
Thank you in advance!

Best regards,
Kai-Yuan