Re: st: mfx with ivprobit and ordinal variable

[JOHN ANTONAKIS <john.antonakis@unil.ch>]

Actually, -mfx- and -adjust- give the same result when using the -at- 
option in mfx and if all the predictors are put into the -adjust- 
command. For example:

. sysuse auto
. quietly reg price weight length foreign
. adjust weight length, by(foreign)

-----------------------------------------------------------------------------------------
Dependent variable: price Command: regress
Covariates set to mean: weight = 3019.4595, length = 187.93243
-----------------------------------------------------------------------------------------

----------------------
Car type | xb
----------+-----------
Domestic | 5102.99
Foreign | 8676.08
----------------------
Key: xb = Linear Prediction

. mfx, at (foreign=0) nose

warning: no value assigned in at() for variables weight length;
means used for weight length

Marginal effects after regress
y = Fitted values (predict)
= 5102.9862
-------------------------------------------------------------------------------
variable | dy/dx X
---------------------------------+---------------------------------------------
weight | 5.774712 3019.46
length | -91.37083 187.932
foreign*| 3573.092 0
-------------------------------------------------------------------------------
(*) dy/dx is for discrete change of dummy variable from 0 to 1

. mfx, at (foreign=1) nose

warning: no value assigned in at() for variables weight length;
means used for weight length

Marginal effects after regress
y = Fitted values (predict)
= 8676.0781
-------------------------------------------------------------------------------
variable | dy/dx X
---------------------------------+---------------------------------------------
weight | 5.774712 3019.46
length | -91.37083 187.932
foreign*| 3573.092 1
-------------------------------------------------------------------------------
(*) dy/dx is for discrete change of dummy variable from 0 to 1

Best,
J.
____________________________________________________

Prof. John Antonakis
Associate Dean
Faculty of Business and Economics
University of Lausanne
Internef #618
CH-1015 Lausanne-Dorigny
Switzerland

Tel ++41 (0)21 692-3438
Fax ++41 (0)21 692-3305

http://www.hec.unil.ch/people/jantonakis&cl=en
____________________________________________________



On 14.01.2009 16:51, Maarten buis wrote:
> --- John L Worrall <worrall@utdallas.edu> wrote:
>> I am using ivprobit. My endogenous variable of interest is ordinal
>> (it’s called “custcode� and ranges from 1-5). I’m interested
>> in obtaining predicted probabilities for each of its values. From
>> what I can gather, the best approach is “mfx, predict(p) at
>> (custcode=1) nose,� “mfx, predict(p) at(custcode=2) nose,� etc.
>
> This will give you marginal effect instead of predicted probabilities,
> to get predicted probabilities you can use -adjust-, see -help adjust-.
>
> 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 N515
>
> +31 20 5986715
>
> http://home.fsw.vu.nl/m.buis/
> -----------------------------------------
>
>
>
> *
> * 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/

*
*   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/