EViews.com

HI, I am learning econometrics and am playing with some datasets to practise different modelling techniques. I have included a lagged dependent variable to remove much of the serial correlation but there is still some obvious serial correlation left but only for part of the series. I did think to just use a robust standard error but the coefficient on the lagged dependent variable could be inconsistent so I was wondering what the best thing to do in this sort of situation is?
Thanks