Question about the significance level of Durbin-Watson Test in Stata Manual

[Alex Mai]

On P5 of the Stata Manual of "regress postestimation time series", the interpretation of the result of -estat dwa- is based on 1% significance level in the Durbin-Watson Statistics Table (K=1 and n=22: dL=0.997). However, in some other materials people choose 5% significance level in the Durbin-Watson Statistics Table.

For my case, I estimate trend rate by fitting the equation using OLS, y=a+bx, where x is the time variable and b is the trend rate. The existence of autocorrelation depends on which significance level (1% or 5%) I use.

So I would like to ask how to choose the significance level for Durbin-Watson test.

Thank you very much!

[Phil Bromiley]

You didn't get a quick answer. You'll increase your chances of a useful answer by following the FAQ on asking questions - provide Stata code in code delimiters, readable Stata output, and sample data using dataex. I'm not sure how much sense it makes to worry about serial correlation if you are just regressing on a time trend.

As to the significance level to use, this is a general issue - it applies to all testing, not just DW. Some disciplines routinely use .05 and some .01. There is no good answer. Some suggest that if you have a very large sample, a higher significance level is appropriate.