Zhehui,
Try the Berndt-Hall-Hall-Hausman (BHHH) algorithm. Using this algorithm the
standard errors very nearly match (except for AGESQ, which probably a typo).
For example (data is at:
http://www.econ.ucdavis.edu/faculty/cameron/racd/racddata.html)
. do "C:\DOCUME~1\Scott_2\LOCALS~1\Temp\STD08000000.tmp"
. infile SEX AGE AGESQ INCOME LEVYPLUS FREEPOOR FREEREPA ILLNESS ACTDAYS ///
> HSCORE CHCOND1 CHCOND2 DVISITS NONDOCCO HOSPADMI HOSPDAYS MEDICINE ///
> PRESCRIB NONPRESC CONSTANT ///
> using "C:\Documents and Settings\Scott_2\Desktop\racd3.asc"
(5190 observations read)
. qui nbreg DVISITS SEX AGE AGESQ INCOME LEVYPLUS FREEPOOR FREEREPA ///
> ILLNESS ACTDAYS HSCORE CHCOND1 CHCOND2, dis(const) nolog tech(bhhh)
. est store NB1
. qui nbreg DVISITS SEX AGE AGESQ INCOME LEVYPLUS FREEPOOR FREEREPA ///
> ILLNESS ACTDAYS HSCORE CHCOND1 CHCOND2, dis(mean) nolog tech(bhhh)
. est store NB2
. est table NB1 NB2, se drop(_cons) b(%9.3f)
--------------------------------------
Variable | NB1 NB2
-------------+------------------------
SEX | 0.164 0.217
| 0.060 0.066
AGE | 0.280 -0.216
| 1.126 1.234
AGESQ | 0.020 0.609
| 1.199 1.381
INCOME | -0.135 -0.142
| 0.096 0.098
LEVYPLUS | 0.212 0.118
| 0.084 0.085
FREEPOOR | -0.537 -0.497
| 0.209 0.175
FREEREPA | 0.208 0.145
| 0.104 0.117
ILLNESS | 0.196 0.214
| 0.021 0.026
ACTDAYS | 0.112 0.144
| 0.006 0.007
HSCORE | 0.036 0.038
| 0.010 0.014
CHCOND1 | 0.133 0.099
| 0.075 0.077
CHCOND2 | 0.174 0.190
| 0.089 0.095
--------------------------------------
legend: b/se
end of do-file
Hope this helps,
Scott
> -----Original Message-----
> From: [email protected] [mailto:owner-
> [email protected]] On Behalf Of Zhehui Luo
> Sent: Thursday, January 27, 2005 10:49 AM
> To: [email protected]
> Subject: st: [email protected]
>
> Hi,
>
> To reproduce some part of Table 3.4 in Cameron and Trivedi (1998) book
> "Regression analysis of count data", I found this help file from Stata
> Corp
> FAQs (http://www.stata.com/support/faqs/stat/nbreg1.html).
>
> I ran nbreg, dispersion(constant) , but the resulted standard errors for
> beta are very different from what's reported in the MLE column and similar
> to the GLM column. Could someone tell me why?
>
> Thanks,
>
> Zhehui
*
* 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/
