# st: xtreg with and without the mle command

 From "Fouskakis Dimitris" To Subject st: xtreg with and without the mle command Date Thu, 3 Nov 2005 14:28:38 +0200

```Dear Statalist,

I am facing the following problem concerning the command xtreg with and
without mle.
I have a dataset of 103 kids, that they belong to two different groups
(pre-term, full-term) and we measure a kind of suppression in both ears. I
want to see weather or not the group variable is a risk factor of
suppression.
After I reshape my data I run the xtreg command and I give you the two
different results, with and without the mle command at the end of the email.

As you can see from the output with the mle command the sigma_u is almost
zero, thus the estimates are almost like the OLS. Any idea why this is
happening? Shall I report the results without the mle? Does this mean that I
dont have random effects, and I should run the xtgee command instead and
report these results?

Thanks for your help.

Dimitris

============================================================================
==============================

. xtreg supp  ear group, i(kid)

Random-effects GLS regression                   Number of obs      =
184
Group variable (i): kid                         Number of groups   =
103

R-sq:  within  = 0.0430                         Obs per group: min =
1
between = 0.1318                                        avg =
1.8
overall = 0.0834                                        max =
2

Random effects u_i ~ Gaussian                   Wald chi2(2)       =
16.24
corr(u_i, X)       = 0 (assumed)                Prob > chi2        =
0.0003

----------------------------------------------------------------------------
--
supp |      Coef.   Std. Err.      z    P>|z|     [95% Conf.
Interval]
-------------+--------------------------------------------------------------
--
ear |   .1515081   .1097378     1.38   0.167    -.0635741
.3665903
group |   .4414064   .1157561     3.81   0.000     .2145287
.6682841
_cons |   .3400142   .1839055     1.85   0.064    -.0204339
.7004624
-------------+--------------------------------------------------------------
--
sigma_u |  .18900564
sigma_e |  .74538925
rho |  .06041168   (fraction of variance due to u_i)
----------------------------------------------------------------------------
--

. xtreg supp  ear group, i(kid) mle

Fitting constant-only model:
Iteration 0:   log likelihood = -218.05492
Iteration 1:   log likelihood = -217.81992
Iteration 2:   log likelihood = -217.81635
Iteration 3:   log likelihood = -217.81626
Iteration 4:   log likelihood = -217.81626
Iteration 5:   log likelihood = -217.81626

Fitting full model:
Iteration 0:   log likelihood = -209.84604
Iteration 1:   log likelihood = -209.81413
Iteration 2:   log likelihood = -209.80715
Iteration 3:   log likelihood = -209.80605
Iteration 4:   log likelihood = -209.80601
Iteration 5:   log likelihood = -209.80601
Iteration 6:   log likelihood = -209.80601

Random-effects ML regression                    Number of obs      =
184
Group variable (i): kid                         Number of groups   =
103

Random effects u_i ~ Gaussian                   Obs per group: min =
1
avg =
1.8
max =
2

LR chi2(2)         =
16.02
Log likelihood  = -209.80601                    Prob > chi2        =
0.0003

----------------------------------------------------------------------------
--
supp |      Coef.   Std. Err.      z    P>|z|     [95% Conf.
Interval]
-------------+--------------------------------------------------------------
--
ear |   .1472708   .1116638     1.32   0.187    -.0715861
.3661278
group |   .4354625   .1116638     3.90   0.000     .2166055
.6543195
_cons |   .3475279   .1851485     1.88   0.061    -.0153566
.7104123
-------------+--------------------------------------------------------------
--
/sigma_u |   3.09e-24   .4801257     0.00   1.000     -.941029
.941029
/sigma_e |   .7567756   .0394497    19.18   0.000     .6794557
.8340955
-------------+--------------------------------------------------------------
--
rho |   1.67e-47   5.18e-24                             0
1
----------------------------------------------------------------------------
--
Likelihood-ratio test of sigma_u=0: chibar2(01)=    0.00 Prob>=chibar2 =
1.000
============================================================================
================================

----------------------------------------------------------
Dr. Dimitris Fouskakis
Department of Mathematics
National Technical Univ. of Athens
Zografou Campus, Athens 15780
Greece

Tel.: +30 210 772 1702
Fax:  +30 210 772 1775
-----------------------------------------------------------

*
*   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/
```