I only get the Digest, so do not see the Stata list until the next day.
I submitted a rather extensive article to the Stata Journal a while
back that shows how to
create a number of different types of synthetic data sets using Stata.
I also provide
code that can be amended by the user to modify the models as they
desire.. Included
are, in part, logit and probit both binary and proportional response,
Poisson with and
without offsets and with a separate cluster variable, negative binomial
with and without
offsets -- I show for each NB2, NB1, quasipoisson, and NB-C. I also
show multinomial
and ordered logit and probit models, and synthetic hurdle models. The
user can specify
how many predictors they what and their coefficient values, and for NB
models can specify
the desired value of alpha, the ancillary parameter. For categorical
response models, the user
can select how many levels and the coefficients in each level, and for
ordered models,
the values of coefficients and cut values.
I used and explained many of these models in my text, "Logistic
Regression Models", but
the article goes into greater length for non-binomial models. If you
are having a need to
create a synthetic model of this type, or others which can use the same
logic for
construction, let me know and I'll send it to you. Email me a
[email protected]
I am a strong advocate of using synthetic models to better understand
real models.
Fortunately, Stata has all of the tools necessary to create a wide
variety of synthetic
models.
Joseph Hilbe
========================================
Date: Sat, 17 Oct 2009 14:37:59 +0200
From: "Martin Weiss" <[email protected]>
Subject: st: RE: R: RE: R: Simulation using Stata (flag: 9.2/SE version)
<>
Well, the MUS book is pretty comprehensive in its coverage of MCs, and
the
presentation by Ian White at the London UGM this year may fill some
blank
spots. Stata.com seems to be down at the moment, so cannot post the
link...
A book totally devoted to MC would be a bit of a stretch, IMHO...
HTH
Martin
- -----Original Message-----
From: [email protected]
[mailto:[email protected]] On Behalf Of Carlo Lazzaro
Sent: Samstag, 17. Oktober 2009 13:34
To: [email protected]
Subject: st: R: RE: R: Simulation using Stata (flag: 9.2/SE version)
Dear Martin,
In my previous posting I meant something like "Performing Monte Carlo
simulations with Stata".
IMHO, this textbook should cover the theoretical building blocks of
Monte
Carlo simulations, including double loop exercises, and present a lot of
examples drawn from different research fields (eg: risk analysis;
epidemiology; statistics; econometrics and, of course,
microeconometrics;
health economics; ecology)with related Stata codes discussed to the gory
detail.
I do not know if someone at Stata or among the most prominent Statalist
contributors are interested in publishing such a textbook.
Kind Regards and a nice W_E to you and to all Statalisters,
Carlo
- -----Messaggio originale-----
Da: [email protected]
[mailto:[email protected]] Per conto di Martin Weiss
Inviato: sabato 17 ottobre 2009 12.09
A: [email protected]
Oggetto: st: RE: R: Simulation using Stata (flag: 9.2/SE version)
<>
" Unfortunately, Stata textbooks entirely devoted to this topic haven't
be
published, so far."
Microeconometrics Using Stata, by Cameron and Trivedi (2008), is full of
examples for -simulate-...
HTH
Martin
- -----Original Message-----
From: [email protected]
[mailto:[email protected]] On Behalf Of Carlo Lazzaro
Sent: Samstag, 17. Oktober 2009 11:24
To: [email protected]
Cc: 'Ali Rowhani-Rahbar'
Subject: st: R: Simulation using Stata (flag: 9.2/SE version)
Dear Ali,
as far as my knowledge is concerned, the only examples of simulations
with
Stata (9.2/SE version) are reported in - help simulate -, that you might
want to invoke from within Stata.
Unfortunately, Stata textbooks entirely devoted to this topic haven't be
published, so far.
However, sticking to your query, performing a probabilistic sensitivity
analysis on the observed OR is straightforward with Stata.
- ----------------------------begin example
--------------------------------
sysuse nlsw88.dta
logistic union married
set obs 10000
g Sens_OR= (.7800495 +.0863818 *invnorm(uniform()))
- ----------------------------end example
--------------------------------
In the example sketched above you can draw 10,000 random values from
the OR
sampling distribution by simply plugging in the observed OR and its
standard
error.
HTH and Kind Regards,
Carlo
*
* 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/
