Statalist


[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: st: random effects panel model - interpretation of rho=0


From   John Antonakis <john.antonakis@unil.ch>
To   statalist@hsphsun2.harvard.edu
Subject   Re: st: random effects panel model - interpretation of rho=0
Date   Wed, 20 May 2009 20:15:37 +0200

One definition of Rho (I am not sure if Stata uses this definition) is :

(MSb - MSw)/((MSb + (k-1)*MSw))

Where, k is the average group size; MS=mean square; b=between; w=within.

So, a rho of zero is telling you that that the variablilty within clusters is large relative to between; thus, your xtreg estimates are convering to OLS estimates.

What does -xttest0- tell you? Having random effects means that the intercept varies between subjects.

Also, when you run in OLS (with cluster robust SEs) are your estimates different from the xtreg estimator (they should not be)?

Best,
J.

____________________________________________________

Prof. John Antonakis
Associate Dean Faculty of Business and Economics
University of Lausanne
Internef #618
CH-1015 Lausanne-Dorigny
Switzerland

Tel ++41 (0)21 692-3438
Fax ++41 (0)21 692-3305

Faculty page:
http://www.hec.unil.ch/people/jantonakis&cl=en

Personal page:
http://www.hec.unil.ch/jantonakis
____________________________________________________



On 20.05.2009 19:52, P K wrote:
Hi,

I run a random effects panel model of 64 subjects for 10 years each and have a question concerning the results: My output tells me that
0% of the variance of the dependent variable is between subjects and 100% is within subjects (rho).

I am not sure whether I understand the interpretation correctly, but I find this result a bit uncommon given that I test for a random effects model and
my theory predicts a lot of variation between my subjects.
Could there by any mistake in the analysis?
Thanks, Pat

Here's the STATA output:
Random-effects GLS regression                   Number of obs      =       545
Group variable: number                          Number of groups   =        64

R-sq:  within  = 0.0138                         Obs per group: min =         2
       between = 0.1741                                        avg =       8.5
       overall = 0.0168                                        max =        10

Random effects u_i ~ Gaussian                   Wald chi2(7)       =      9.16
corr(u_i, X)       = 0 (assumed)                Prob > chi2        =    0.2416

------------------------------------------------------------------------------
AverCh~2adj2 |      Coef.   Std. Err.      z    P>|z|     [95% Conf. Interval]
-------------+----------------------------------------------------------------
duration_c~e |  -.9013214   .3550697    -2.54   0.011    -1.597245   -.2053976
   sicchange |   2.527742   4.073774     0.62   0.535    -5.456709    10.51219
       ukdum |   .5491039   1.126746     0.49   0.626    -1.659277    2.757485
     frendum |  -1.780095   1.522111    -1.17   0.242    -4.763378    1.203189
 regnlistdum |  -.6389011    1.09439    -0.58   0.559    -2.783865    1.506063
      mutdum |   .9720512   1.790017     0.54   0.587    -2.536318     4.48042
lgcount_st~y |  -.2016355   .6181168    -0.33   0.744    -1.413122    1.009851
       _cons |   1.249498   1.361879     0.92   0.359    -1.419735    3.918731
-------------+----------------------------------------------------------------
     sigma_u |          0
     sigma_e |  11.774903
         rho |          0   (fraction of variance due to u_i)
------------------------------------------------------------------------------


*
*   For searches and help try:
*   http://www.stata.com/help.cgi?search
*   http://www.stata.com/support/statalist/faq
*   http://www.ats.ucla.edu/stat/stata/
*
*   For searches and help try:
*   http://www.stata.com/help.cgi?search
*   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