I want to check for stationarity among my variables in a N=10/T=20 panel dataset. I am using the -xtunitroot llc- command to conduct the Levin-Lin-Chu test for my dependent variable and my list of about 10 independent variables.

When I run the test on each of the variables, some are stationary and some are not. However, if I add the -trend- option to the ones that aren't stationary, they become stationary. When I add the -trend- option to some of the variables that were already stationary without the -trend- option, they become non-stationary. How do I interpret this? Is it okay to add different options (e.g. trends, lag-lengths, etc.) for different variables? Does this mean that I need to have a trend term in my regression?

Thanks!