Unfortunately, my question is not very specific, but anyway ... I got
somewhat stuck trying to do cross-sectional time series analysis. As I am
pretty new to applying this kind of design, I am looking for some advice.
My problem is that, though I am familiar with the basic elements of each of
the various techniques, I do not manage to figure out when to use which
kind of estimator. To be sure: My current data set suffers from severe
inter-section autocorrelation. So I decided to skip basic fixed or random
effects models using xtreg, fe(re). Later on I discovered xtregar and
xtpcse. (The latter now seems to be a first choice in analysing
cross-sectional time series in political science.) How do I tell when to
use which command?
Probably nobody will be able to give a short answer to this kind of
question, but do you know some accessible kind of reference or instruction?
Any advice would be greatly appreciated!
