What statistical value does the Durbin Watson statistic take on when testing for autocorrelation?

The Durbin-Watson statistic is specifically designed to detect the presence of autocorrelation in the residuals from a regression analysis, particularly first-order autocorrelation. The statistic ranges between 0 and 4, where:

• A value around 2 suggests no autocorrelation.

• Values towards 0 indicate positive autocorrelation.

• Values towards 4 indicate negative autocorrelation.

The range of 0 to 4 is key because it allows analysts to effectively interpret the results of their regression models regarding the independence of residuals. Therefore, stating that the Durbin-Watson statistic takes a value between 0 and 4 accurately reflects the expectations for this statistic in practice, making it the correct choice.