If the Durbin-Watson statistic has a value close to 0, which assumption is violated? In other words, which assumption is the Durbin-Watson statistic checking to see is violated?

The assumption of Independence of errors is violated.

Explanation:

Durbin-Watson statistic is used to measure the serial correlation or autocorrelation in residuals by using regression analysis. An assumption is that the errors are normally distributed within a mean of zero and are stationary. Autocorrelation determines the connection or relationship between the current and the past values of a variable.

The static should essentially be in between 0 to 4. When the static is below 2, it indicates a positive autocorrelation and when it is close to 0, the independence of errors no more exists.

With the help of autocorrelation, we can do technical analysis related to the prices of securities and their trends.