--- "Giulio Michele Bosio (giulio.bosio)" <[email protected]>
wrote:
> I need to estimate with stata a sequential logit model. It's a simple
> logit plus multinomial logit with 4 outcome.In this model there is no
> correlation between errors at first and second step.So I need 5
> vectors of betas (parameters).
Giulio:
A sequential logit with 4 outcomes would result in 3 vectors of betas.
With -logit- you model 3 transitions: The transition from outcome 1 to
outcome 2, the transition from outcome 2 to 3, and the transition from
outcome 3 to 4. Say we have the following 4 outcomes: less then
highschool, highschool, some college, complete college or more, and we
are willing to assume that someone must have passed all `lesser' levels
in order to have achieved his level. The example below shows how to
estimate a sequential logit in that case. The if statements in the
-gen- commands ensure that people who didn't get highschool get a
missing value for all subsequent transitions; they are no longer at
risk and drop out of the analsyses of those transitions.
HTH,
Maarten
*---------------- begin example --------------
sysuse nlsw88
gen tr0t1 = grade >= 12
gen tr1t2 = grade > 12 if grade >=12
gen tr2t3 = grade >= 16 if grade > 12
xi: logit tr0t1 south i.race
xi: logit tr1t2 south i.race
xi: logit tr2t3 south i.race
*---------- end example ---------------
-----------------------------------------
Maarten L. Buis
Department of Social Research Methodology
Vrije Universiteit Amsterdam
Boelelaan 1081
1081 HV Amsterdam
The Netherlands
visiting adress:
Buitenveldertselaan 3 (Metropolitan), room Z434
+31 20 5986715
http://home.fsw.vu.nl/m.buis/
-----------------------------------------
Send instant messages to your online friends http://uk.messenger.yahoo.com
*
* 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/
