There is something seriously wrong with your code: The fact that you
have called -sum- doesn't mean that the variable will now contain the
mean, as you appear to think. The mean is stored in r(mean) and will
change every time you will call -sum-. So you will have to store it,
e.g. in a local or a scalar, and feed that to norm.
The formula for the marginal effect is probably something like:
norm(xb)*norm(-xb)*_b[whatever]. However I don't know if the
correlation between error terms is going to spoil that formula, so
you'd better derive it yourself (it's the first derivative)
Hope this helps,
Maarten
--- stata_user stata_user <[email protected]> wrote:
> Thanks a lot Maarten,
>
> I found your paper very interesting, I have followed your suggestion
> .
>
> Since I am interested on the change in the marginal probabilities
> from
> mvprobit , for example for one of my independent binary variables ,
> called
> DM, I have used the following syntax to obtain the the 3 marginal
> probailities for my 3 dependent binary variables (bn_1,bn_2 and bn_3)
>
> *------------------------------DM==0-----------------*
>
> mvppred xb_DM0 if DM ==0,xb
>
> sum xb_DM01 if e(sample)
> sum xb_DM02 if e(sample)
> sum xb_DM03 if e(sample)
>
> *------------------------------DM==1-----------------*
>
> mvppred xb_DM1 if DM ==1,xb
>
> sum xb_DM11 if e(sample)
> sum xb_DM12 if e(sample)
> sum xb_DM13 if e(sample)
>
> *--------------Marginal probabilities for DM =0 and
> DM=1-----------------*
> *bn_1,bn_2 and bn_3 are the 3 binary variables
> *For bn_1
> gen bn_01 = norm(xb_DM01)
> gen bn_11 = norm(xb_DM11)
>
> *For bn_2
> gen bn_02 = norm(xb_DM02)
> gen bn_12 = norm(xb_DM12)
>
> *For bn_3
> gen bn_03 = norm(xb_DM03)
> gen bn_13 = norm(xb_DM13)
>
> However I have noticed that even if the 3 beta coeffecients of DM
> were all
> positives, the change on the probabilites from DM=0 to DM=1 (having
> all
> other variables constant at their mean values) were either positive
> or
> negative, is this make sense ?
>
> The second point is how to obtain the change for a continuous
> variable ?
>
> Kind regads,
>
> Sami
>
>
> >From: Maarten buis <[email protected]>
> >Reply-To: [email protected]
> >To: [email protected]
> >Subject: RE: st: matrix product in stata after mvprobit
> >Date: Wed, 18 Apr 2007 09:04:02 +0100 (BST)
> >
> >--- stata_user stata_user <[email protected]> wrote:
> > > I have tried mvppred, but I have noticed that I can have
> predicted
> > > values for xb , p and other statistics only for the values of
> the
> > > x's (independent variables) that are already in my data.
> > >
> > > When I specify with the " if" statement , that I want the
> predicted
> > > for the means of the x�s, I have a message saying � no
> >observations�.
> > >
> > > I do not know if it is possible to get the predicted values on
> the
> > > mean values of the x's (independent variables)
> >
> >You can use the fact that the mean linear predictor is the linear
> >predictor at mean values of x. Note, however that the mean
> probability
> >is not the probability at mean values of x. See:
> >http://home.fsw.vu.nl/m.buis/wp/logit_adjust.html .
> >
> >So what you want to do is: 1) use predict to create the linear
> >predictors, 2) sumarize them (don't forget to ensure that you are
> doing
> >it on the same sample by adding: if e(sample)), and 3) plug the
> means
> >into the formula for creating probabilities.
> >
> >However, comparing mean predicted probabilities is in this case
> easier,
> >and in many cases theoretically nicer: you are comparing a typical
> >predicted probability for men with a typical predicted probability
> for
> >women, instead of comparing predicted probability for men who have
> >typical values on the other explanatory variables with predicted
> >probabilites for women who have typical values on the other
> explanatory
> >variables.
> >
> >Hope this helps,
> >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
> >
> >http://home.fsw.vu.nl/m.buis/
> >-----------------------------------------
> >
> >
> > ___________________________________________________________
> >Yahoo! Mail is the world's favourite email. Don't settle for less,
> sign up
> >for
> >your free account today
>
>http://uk.rd.yahoo.com/evt=44106/*http://uk.docs.yahoo.com/mail/winter07.html
> >*
> >* 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/
>
> _________________________________________________________________
> Get Hotmail, News, Sport and Entertainment from MSN on your mobile.
> http://www.msn.txt4content.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/
>
-----------------------------------------
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
http://home.fsw.vu.nl/m.buis/
-----------------------------------------
___________________________________________________________
Yahoo! Mail is the world's favourite email. Don't settle for less, sign up for
your free account today http://uk.rd.yahoo.com/evt=44106/*http://uk.docs.yahoo.com/mail/winter07.html
*
* 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/
