Doubt of the Cox model

[Ana Guerrero]

Hello everyone,

I am doing my thesis which seeks to find the variables that influence the survival of micro-enterprises in the 2015-2021 period, for this I have an unbalanced panel database that includes variables that vary over time, for example, I have the GDP by state that varies over time (that is, one value for 2015, another for 2016, etc), my question is, to estimate a Cox model, should I specify to stata that I’m using variables that change over time? I have read about it and I know that the extended Cox model exists, but I'm not sure that this extension of the model refers to these types of variables that change over time.
I also have another question: should I specify the ID variable for multiple records to the model or is it not necessary?


Thank you,
Ana

[Carlo Lazzaro]

Ana:
welcome to this forum.
Given your description, why not considering a panel data regression (-xtlogit-)?
As far as -stcox- is concerned, you may refer to the -tvc()- option, but I fear it does not apply to what you're probably after.

[Ana Guerrero]

Ana:
welcome to this forum.
Given your description, why not considering a panel data regression (-xtlogit-)?
As far as -stcox- is concerned, you may refer to the -tvc()- option, but I fear it does not apply to what you're probably after.

Hi Carlo, thanks for your answer.
I've read some documents that use an unbalanced panel and these types of variables with the Cox model, however I'm going to thoroughly review the two options that you propose to me.

Thank you very much again,
Kind regards,
Ana