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

Re: st: dirifit

From   Viktor Slavtchev <>
Subject   Re: st: dirifit
Date   Wed, 05 Mar 2008 09:45:05 +0100

- Maarten, thank you very much!
I am also about to finish my PhD. I need three more months. I know how it feels.
Good luck

Maarten buis wrote:

--- Viktor Slavtchev <> wrote:

My dependent variable are proportions in five categories which add
to one (y1-y5). I want to test the impact of a set of independent
variables (a, b, c), which are common for each y.
Additionally, I wan to test the impact of x1, x2, x3, x4, x5 whereat
x1 enters only the equation for y1, x2 enters only y2, and so on.
I think -dirifit- is the right technique. So I run fist:
dirifit y1 - y5, alpha(a b c) alpha1(x1) alpha2(x2) alpha3(x3) alpha4(x4) alpha5(x5) However, as I am interested in the marginal effects I have to use the

-mu- option (if I've got it correctly).
My problem is, however, that with -mu- option I was not able to
specify that x1 hat to enter only y1, x2 only y2, ...
Any help is greatly appreciated.
There are -mu1-, -mu2-, etc options, but they don't work well with -ddirifit- which I assume is the problem you are running in to. You
could add x1 till x5 to all equation and add constraints that their
parameters are 0 in all equations where you don't want them. Generally
speaking the -ddirifit- command (the command that computes the marginal
effects) is too fragile for my taste (I am one of the authors, so I
don't consider it bad taste to beat up my own program). It is on my
todo list to repair that. But, as I have said on some other occations
on this list, I am in the process of finishing my dissertation, which
has priority right now.

-- Maarten

Maarten L. Buis
Department of Social Research Methodology
Vrije Universiteit Amsterdam
Boelelaan 1081
1081 HV Amsterdam
The Netherlands

visiting address:
Buitenveldertselaan 3 (Metropolitan), room Z434

+31 20 5986715

Sent from Yahoo! Mail.
A Smarter Inbox.
* For searches and help try:
*   For searches and help try:

© Copyright 1996–2015 StataCorp LP   |   Terms of use   |   Privacy   |   Contact us   |   What's new   |   Site index