Statalist The Stata Listserver


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

st: RE: panel data and hausman test


From   "Schaffer, Mark E" <M.E.Schaffer@hw.ac.uk>
To   <statalist@hsphsun2.harvard.edu>
Subject   st: RE: panel data and hausman test
Date   Tue, 1 Aug 2006 19:42:06 +0100

Ekaterina,

> -----Original Message-----
> From: owner-statalist@hsphsun2.harvard.edu 
> [mailto:owner-statalist@hsphsun2.harvard.edu] On Behalf Of 
> Ekaterina Rashkova
> Sent: 01 August 2006 19:20
> To: statalist@hsphsun2.harvard.edu
> Subject: st: panel data and hausman test
> 
> Dear all, 
> 
> I am trying to decide whether to use fixed or random effects 
> model. In doing that I ran both models and then the hausman 
> test. The results fro mthe test suggest that fixed effects 
> might be superior (prob>chi sq is smaller than
> 0.05) but does that make sense substantively?

This means that the coefficients of interest are different in the two
estimations, in a statistical sense.  Since under the null (both RE and
FE are consistent) they shouldn't be different, and under the
alternative (only FE is consistent) they may differ, you have evidence
that only FE is consistent.

> I am studying 
> the determinants of voter turnout and my data consists of  
> 550+ observation and 77 countries. 
> Both models (FE and RE) do not provide different signs or 
> significance, 

You mean significance in the sense of significantly different from zero.

Another way to put it is that the Hausman test suggests you should use
your FE results, but with respect to interpreting the coefficients,
there is no practically signficant (as opposed to statistically
significant) difference between the RE and FE results.

> but the R-sq. stat differs quite a bit (3 vs 
> 29%).

You probably want to check the Stata manual on within/between/overall
R2s.

HTH.

Cheers,
Mark

> Any suggestions and advice is welcome. 
> This is the hausman test output
> :
> hausman fixed random
> 
>                  ---- Coefficients ----
>              |      (b)          (B)            (b-B)     
> sqrt(diag(V_b-V_B))
>              |     fixed        random       Difference          S.E.
> -------------+------------------------------------------------
> ----------
> -------------+------
>    stability |    .8612319     .5429065        .3183254       
>   .130613
>     PRHetero |    17.92476     16.23667        1.688093       
>   4.88696
>          age |   -.1873375    -.1401071       -.0472304       
>  .0171591
>          gdp |    .0858235     .0801026        .0057209       
>  .0105805
>    mandatory |    8.434539     6.504805        1.929734       
>  3.078364
>           pr |   -7.481201    -6.438582       -1.042618       
>  1.307992
>          elf |   -18.81114    -21.91486        3.103719       
>  10.77812
> --------------------------------------------------------------
> ----------------
>                            b = consistent under Ho and Ha; 
> obtained from xtreg
>             B = inconsistent under Ha, efficient under Ho; 
> obtained from xtreg
> 
>     Test:  Ho:  difference in coefficients not systematic
> 
>                   chi2(7) = (b-B)'[(V_b-V_B)^(-1)](b-B)
>                           =       25.94
>                 Prob>chi2 =      0.0005
> 
> 
> I appreciate your help.
> 
> Regards,
> Ekaterina
> 
> --
> Ekaterina R. Rashkova
> Graduate Student
> Department of Political Science
> Washington University
> Campus Box 1063
> One Brookings Drive
> St. Louis, MO 63130
> (314) 935-5856 (Fax)
> 
> *
> *   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/
> 
> 

*
*   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/



© Copyright 1996–2014 StataCorp LP   |   Terms of use   |   Privacy   |   Contact us   |   What's new   |   Site index