Pooled cross sectional data: How to test for and handle autocorrelation?


1) My goal:
I am making a difference-in-difference analysis to estimate the development in political efficacy for a certain group of people relative to another group. I want to test for autocorrelation and handle it (if the test shows that autocorrelation is present).

2) My data:
I work with survey data selected in the years 1994, 1998, 2001, 2005 and 2007. The same questions are asked every year, but the sample is not the same. Therefore my data is not panel data but pooled cross sectional data over time.

3) My problem:
I have searched econometric literature and the internet to find out how to test for autocorrelation, which I have learned is a general problem when working with time series data. When I read about autocorrelation techniques the Durbin-Watson test is mentioned many times, but it is only related to studies working with panel data.

Therefore my question is: Can I use the Durbin-Watson test on pooled cross sectional data, or should I use another test? Are lagged effects a possible solution if autocorrelation is present?

Hoping for your help!

Best wishes Matilde