1
$\begingroup$

I was running a couple of Tobit regressions on SAS. Surprisingly, I found that estimates from one model show 0 standard errors, missing (.) t-values and p-values. Why did this happen? What does this scenario imply? If this is not normal, how can I get rid of it?

Please allow me to share the output.

And much thanks for your guidance.

enter image description here

enter image description here

Improve this question
$\endgroup$
0

1 Answer 1

Reset to default
1
$\begingroup$

This typically happens when your Hessian matrix is nearly singular. The following SAS article describes your exact situation I believe.

http://support.sas.com/kb/31/714.html

Try re-scaling the variables that have zero standard errors as this is most likely the cause of the problem you are experiencing. I'm guessing that these variables are measured on a scale that is significantly larger/smaller than your other variables, which is leading to the invertibility problems with the Hessian matrix.

You're likely to resolve this problem by standardizing the predictors before running proc qlim. See proc standard for this. Verify that you have correctly specified your model and didn't accidentally include some variables twice. It seems this has lead to some other problems noted here (the answer there by @placidia is a good one and may help you resolve your issue with proc qlim).

If that does't work, try reading through this article and try some of the suggestions there.

Improve this answer
$\endgroup$
1
  • $\begingroup$ Previously I have noticed that when SAS encounters such problems, it gives warnings in the log. However, while running this model I have not seen any warnings. About standardizing, the problems are that the first 4 variables are dummies, and sine and cosine variables are actually sine and cosine values. Anyways, do you think the other parameters estimates would still be reliable? $\endgroup$
    – Deepan Das
    Commented May 30, 2019 at 3:03

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.