topic Re: Durbin–Watson test in proc reg in Statistical Procedures

Durbin–Watson test in proc reg

LengYi — Sat, 03 Mar 2018 14:20:07 GMT

Hi everyone,

I have a question to read the DW test output from proc reg using option dwProb.

This is the output from SAS/STAT(R) 9.2 User's Guide, so this is autocorrelation or not?

How can I know d, dU and dL of test

Thanks

Re: Durbin–Watson test in proc reg

Miracle — Sat, 03 Mar 2018 16:17:28 GMT

I assume your d is Durbin-Watson test statistic. If so, it is 1.191
Yes, there is a significant first-order autocorrelation in the disturbances because Pr<DW=0.005.
You can refer to Durbin-Watson Significance Tables to find out dL and dU. They depend on the number of observations in your data and number of parameter estimates(excluding intercept) in your regression model.

Re: Durbin–Watson test in proc reg

AlexS332 — Fri, 03 May 2019 09:04:36 GMT

Hi everyone,

I'm also working on the durbin watson test for linear regression, with SAS.
I use the dwprob argument to get a pvalue for the test.. Then I get 2 pvalues : one for testing for positive autocorrelation, another to test for negative autocorrelation... But I want a "general" test with alternative "autocorrelation" (positive or negative)..
Do you know how I can get a "general" pvalue for this test from the pvalues givent by sas?
Can I multiply Pr<DW by 2? is it correct?

Thank you very much for help

Alex