[Date Prev][Date Next][Thread Prev][Thread Next][Date index][Thread index]

From |
"Kevin Amess" <[email protected]> |

To |
<[email protected]> |

Subject |
Re: st: Question about random effects mode |

Date |
Tue, 20 Dec 2005 21:33:06 +0000 |

Tao, You wrote: "If the hausman test reject the hypothesis, does that mean the random effects is correlated with regressors?" Yes, so GLS is not consistent but OLS, the fixed effects estimator is. You wrote: "How can I test this assumption except hausman test?" I am guessing (but I do not think it is a wild one), but my guess is that you have conducted a Hausman test but did not get the result you 'wanted' or 'anticipated' and are in search of an alternative test to suit your purpose. My understaning is that the hausman test is standard for comapring fixed and random effects estimation. You wrote: "...fixed effect model will drop some time invariant variables in first stage." Stata will do you a favour and drop variables that are perfectly correlated. If you have time invariant regressors, these are fixed effects, so Stata is dropping them because your model has more than one set of fixed effects. I do not think that statistics, econometrics, or software can help you with this problem. You need to look at the model you seek to estimate and ask yourself what the theoretical motivations are for the estimating equation. More specifically, you need to consider what the binary (or time invariant) variables are "capturing" that the fixed effects are not capturing. Clearly, your time invariant variables are capturing permanent features of what you are modelling, but this is what the fixed effects are doing. Kevin >>> [email protected] 12/20/05 8:02 pm >>> Dear Sir or Madam, I saw the answers about panel data on stata website. I have one question about the random effects model. I am using two step panel data model ( xtivreg ), however, fixed effect model will drop some time invariant variables in first stage. therefore I only can choose random effect model. The RE model has an assumption that: the random effects are uncorrelated with regressors. How can I test this assumption except hausman test? If the hausman test reject the hypothesis, does that mean the random effects is correlated with regressors? Thanks your reply. Tao University of Bath * * For searches and help try: * http://www.stata.com/support/faqs/res/findit.html * http://www.stata.com/support/statalist/faq * http://www.ats.ucla.edu/stat/stata/ This message has been checked for viruses but the contents of an attachment may still contain software viruses, which could damage your computer system: you are advised to perform your own checks. Email communications with the University of Nottingham may be monitored as permitted by UK legislation. * * For searches and help try: * http://www.stata.com/support/faqs/res/findit.html * http://www.stata.com/support/statalist/faq * http://www.ats.ucla.edu/stat/stata/

- Prev by Date:
**st: returning value from a program** - Next by Date:
**st: RE: returning value from a program** - Previous by thread:
**st: Question about random effects mode** - Next by thread:
**st: RE: graph legend** - Index(es):

© Copyright 1996–2024 StataCorp LLC | Terms of use | Privacy | Contact us | What's new | Site index |