what does it means too low Durbin-Watson 0.146384 in OLS multiple regression?
A rule of thumb is that test statistic values in the range of 1.5 to 2.5 are relatively normal. Values outside of this range could be cause for concern. Field(2009) suggests that values under 1 or more than 3 are a definite cause for concern.
The concern here is first order autocorrelation. See a book like Kutner, et al, Applied Linear Statistical Models to see how you can fix the problem. Best wishes, David Booth
Positive autocorrelation. DW can be used for detecting AR(1) only. It has areas that you cannot decide and cannot be used with lag of the dependent as explanatory. Better use Breusch-Godfrey.
In such a case and under the restrictions mentioned you have to solve the problem with Durbin's 2 step, or FD or Cochranne-Orcutt
What test can I use instead of Durbin Watson if it is a lagged variable?
Positive autocorrelation. You better use B-G to avoid the usual problems of D-W
- Badges
- Science topic
- Similar topics
- Medicine
- Counseling Psychology