Statalist


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

Re: Re: st: est2tex, significance levels


From   "Michael Blasnik" <michael.blasnik@verizon.net>
To   <statalist@hsphsun2.harvard.edu>
Subject   Re: Re: st: est2tex, significance levels
Date   Wed, 18 Jul 2007 11:56:21 -0400

...

Just a guess, but it may be worth looking at the d.f. used by est2tex vs the d.f. that should be used given the cluster option in the command -- perhaps est2tex is looking at the wrong d.f., which will change the stars for some of your coefficients.

Michael Blasnik


----- Original Message ----- From: "Yun Liu" <liuyun3@gmail.com>
To: <statalist@hsphsun2.harvard.edu>
Sent: Wednesday, July 18, 2007 11:04 AM
Subject: Re: Re: st: est2tex, significance levels



Thank you,  Valérie. I tried the estimates command and confirmed the
significance levels. So does it mean est2tex has a bug? The latex file
it generates does not assign correct number of stars in my
regressions.

On 7/18/07, Valérie Orozco <Valerie.Orozco@toulouse.inra.fr> wrote:
The p-value is directly obtained from the t distribution.
You can also try the following command :
estimates table,star(0.10 0.05 0.01)   /*respectively a significance of
90%, 5% and 1% */
it will give you the stars associated at each coefficient.

*From*    "Yun Liu" <liuyun3@gmail.com <mailto:liuyun3@gmail.com>>
*To*      statalist@hsphsun2.harvard.edu
<mailto:statalist@hsphsun2.harvard.edu>
*Subject*         Re: st: est2tex, significance levels
*Date*    Wed, 18 Jul 2007 20:05:33 +0800



Sure. On of my regression command is: xi:xtreg aQ rds fyage lassets sp
de i.year, fe i(id) r cl(year) nonest The stata output is at the end of
this message. est2tex gives rds three stars, not two stars. The author
Marc-Andreas Muendler kindly explained that "The stars option is based
on the assumption of normally distributed standard errors. The code uses
t-distributed errors to assess significance, not the p-values." But I am
still puzzled. My regressions include fixed effects, and use robust
standard deviation with cluster options. But the estimated output still
shows t and P>|t|. Doesn't stata calculate p value based on t
distributions? Anyway, the bottom line is, I should add the stars
according to p-values reported in stata output, right? Thank you!

_Iyeara_1996-2004 (naturally coded; _Iyeara_1996 omitted)

Fixed-effects (within) regression Number of obs = 13040 Group variable
(i): id Number of groups = 48 R-sq: within = 0.0533 Obs per group: min =
1 between = 0.0784 avg = 271.7 overall = 0.0645 max = 1378 F(13,8) =
214300.19 corr(u_i, Xb) = 0.0478 Prob > F = 0.0000 (Std. Err. adjusted
for 9 clusters in yeara)
------------------------------------------------------------------------------
Robust
aQ | Coef. Std. Err. t P>|t| [95% Conf. Interval]
-------------+----------------------------------------------------------------

rds | .3686158 .1319893 2.79 0.023 .064248 .6729837
fyage | -.0057774 .0022823 -2.53 0.035 -.0110403 -.0005145
lassets | -.3862244 .0592108 -6.52 0.000 -.5227648 -.249684
sp | 1.345573 .1895998 7.10 0.000 .9083551 1.782791
de | -.1117767 .1149431 -0.97 0.359 -.376836 .1532826
_Iyeara_1997 | .2387015 .0062056 38.47 0.000 .2243913 .2530117
_Iyeara_1998 | .5235133 .0096897 54.03 0.000 .5011689 .5458577
_Iyeara_1999 | .9848062 .0212358 46.37 0.000 .9358364 1.033776
_Iyeara_2000 | .671056 .0209304 32.06 0.000 .6227905 .7193216
_Iyeara_2001 | .2876369 .020132 14.29 0.000 .2412124 .3340613
_Iyeara_2002 | .0992386 .0191882 5.17 0.001 .0549906 .1434866
_Iyeara_2003 | .0816156 .0185712 4.39 0.002 .0387904 .1244408
_Iyeara_2004 | .0807065 .018253 4.42 0.002 .038615 .1227979
_cons | 2.874104 .433527 6.63 0.000 1.874389 3.873819
-------------+----------------------------------------------------------------
sigma_u | .61513864
sigma_e | 2.6455573
rho | .0512914 (fraction of variance due to u_i)
------------------------------------------------------------------------------


*
*   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