Page 1 of 1

Testing for individual effects in panel data

Posted: Thu May 12, 2011 3:24 pm
by christina.sakali
Hi all,

After running the redundant fixed effects testing on EViews, both tests (F, LM (X2)) strongly reject the Ho that there are no individual effects in my data. However when I run the Hausman test in order to test which model is more appropriate for my data (fixed VS random effects) I get the following message: * Cross-section test variance is invalid. Hausman statistic set to zero.


I just wanted to know what it means, when EViews cannot calculate the Hausman statistic, is this due to the nature of my data or does it mean that the random model is not appropriate for my data? And should I try to calculate the Hausman statistic manually?

I have also read the answers to previous similar posts and cannot understand whether this means that the random effects model is not appropriate or whether it is not possible to make any inferences.

I would GREATLY appreciate any responses as I am currently preparing a paper for a conference and I have to justify the choice of my method

Thank you in advance!

Re: Testing for individual effects in panel data

Posted: Thu May 12, 2011 3:42 pm
by EViews Gareth

Re: Testing for individual effects in panel data

Posted: Sun May 15, 2011 5:21 pm
by christina.sakali
Thank you Gareth.

Re: Testing for individual effects in panel data

Posted: Mon May 30, 2011 2:10 am
by carmomi
Dear Gareth,

Would it be possible to post the answer/google link again (the link is not working).

Many thanks
Miguel

Re: Testing for individual effects in panel data

Posted: Tue May 31, 2011 11:12 am
by EViews Glenn