Durbin-Watson test

The Durbin-Watson test is a statistical tool used to detect whether there’s a pattern in the errors or differences between observed and predicted values in a regression analysis. Specifically, it checks for autocorrelation, which is when errors are related over time or sequence—meaning the error at one point influences the next. If the test indicates autocorrelation, it suggests the model might be missing some important trend or pattern. A value close to 2 suggests no autocorrelation; values approaching 0 or 4 indicate positive or negative autocorrelation, respectively.