Survival analysis with lagged variable

[lingbo zhao]

Hi all,

I am here to ask for your suggestions on how to run survival analysis with lagged variable.

I have a series of IV. After I generate a lagged variable of one of the IVs, should I just replace the variable with the new lagged one? Or should I make it consistent by making all the variables lagged?

Thank you very much!

Best,
Lingbo

[Clyde Schechter]

If you actually lag every variable, then you lag nothing. Every variable is then associated with the same time-period's values of every other variable--all you accomplish is removing the earliest observation from the analysis (because lagged values are necessarily missing).

More concretely, you need to sit and think about your research goal and the theory underlying your analysis. If the effect of a variable on your outcome is, according to your theory, delayed, then it makes sense to use one or more lags of that variable in your analysis. If not, you should leave it alone.

[lingbo zhao]

Dear Clyde,
thank you very much for your reply and your time on it. Yea now I understand it better.

Best,
Lingbo