[Limdep Nlogit List] standard errors in random effects regression
William Greene
wgreene at stern.nyu.edu
Fri Oct 16 02:12:00 EST 2009
Your LIMDEP command is missing ;PANEL. You are not computing
GLS estimates, you are computing OLS with a 10 period Newey-West
correction. A striking thing is that Stata seems to be reporting
the OLS estimates with some other kind of standard errors.
It looks like neither set of results is the feasible GLS estimates
you appear to be looking for.
/Bill Greene
----- Original Message -----
From: "Andreas Drichoutis" <adrihout at aua.gr>
To: "Limdep and Nlogit Mailing List" <limdep at limdep.itls.usyd.edu.au>
Sent: Thursday, October 15, 2009 11:07:39 AM GMT -05:00 US/Canada Eastern
Subject: [Limdep Nlogit List] standard errors in random effects regression
Can please someone explain the difference I'm getting from a random effects
regression in Limdep and Stata? Coefficient estimates are the same but
standard errors are completely different. I tend to get more statistical
significant variables in Limdep and tried this with several regressions.
Which ones should I report?
According to Limdep my treatment variable (TREAT_TR) is significant at the
10% level but is not significant according to Stata. In other regressions
I'm getting larger differences.
Regards,
Andreas Drichoutis
--> namelist ; x1=gender,age, period,treat_pr,treat_tr,one $
--> skip $
--> regress ; lhs=bet2lot5 ; rhs=x1 ; pds=10 $
+----------------------------------------------------+
| Ordinary least squares regression |
| Model was estimated Oct 15, 2009 at 05:47:51PM |
| LHS=BET2LOT5 Mean = 1.532028 |
| Standard deviation = 1.210199 |
| WTS=none Number of observs. = 710 |
| Model size Parameters = 6 |
| Degrees of freedom = 704 |
| Residuals Sum of squares = 966.4366 |
| Standard error of e = 1.171657 |
| Fit R-squared = .0692925 |
| Adjusted R-squared = .0626823 |
| Model test F[ 5, 704] (prob) = 10.48 (.0000) |
| Autocorrel Durbin-Watson Stat. = .4687073 |
| Rho = cor[e,e(-1)] = .7656464 |
| Robust VC Newey-West, Periods = 10 |
+----------------------------------------------------+
+--------+--------------+----------------+--------+--------+----------+
|Variable| Coefficient | Standard Error |b/St.Er.|P[|Z|>z]| Mean of X|
+--------+--------------+----------------+--------+--------+----------+
|GENDER | .35515 .22113828 1.606 .1083 .3943662|
|AGE | -.01294 .05868952 -.221 .8255 20.732394|
|PERIOD | .06909*** .02287902 3.020 .0025 5.5000000|
|TREAT_PR| .01037 .20414538 .051 .9595 .4929577|
|TREAT_TR| .34429* .20105428 1.712 .0868 .5070423|
|Constant| 1.10060 1.23528145 .891 .3729 |
+--------+------------------------------------------------------------+
| Note: ***, **, * = Significance at 1%, 5%, 10% level. |
+---------------------------------------------------------------------+
xtreg bet2lot5 gender age period treat_pr treat_tr, re
Random-effects GLS regression Number of obs =
710
Group variable: id Number of groups =
71
R-sq: within = 0.0876 Obs per group: min =
10
between = 0.0612 avg =
10.0
overall = 0.0693 max =
10
Random effects u_i ~ Gaussian Wald chi2(5) =
65.57
corr(u_i, X) = 0 (assumed) Prob > chi2 =
0.0000
----------------------------------------------------------------------------
--
bet2lot5 | Coef. Std. Err. z P>|z| [95% Conf.
Interval]
-------------+--------------------------------------------------------------
--
gender | .3551551 .2502641 1.42 0.156 -.1353535
.8456638
age | -.0129431 .0767783 -0.17 0.866 -.1634259
.1375397
period | .0690943 .0088273 7.83 0.000 .051793
.0863956
treat_pr | .0103745 .2413856 0.04 0.966 -.4627327
.4834816
treat_tr | .3442994 .2429732 1.42 0.156 -.1319193
.820518
_cons | 1.100601 1.635651 0.67 0.501 -2.105216
4.306418
-------------+--------------------------------------------------------------
--
sigma_u | .98865592
sigma_e | .67559506
rho | .68168039 (fraction of variance due to u_i)
----------------------------------------------------------------------------
--
_______________________________________________
Limdep site list
Limdep at limdep.itls.usyd.edu.au
http://limdep.itls.usyd.edu.au
More information about the Limdep
mailing list