Stata The Stata listserver
[Date Prev][Date Next][Thread Prev][Thread Next][Date index][Thread index]

Re: st: Marginal effects and -parmest-


From   Friedrich Huebler <[email protected]>
To   [email protected]
Subject   Re: st: Marginal effects and -parmest-
Date   Mon, 17 Oct 2005 12:24:24 -0700 (PDT)

Roger,

Thank you for the reference to -parmcip-. I tried to generate
variables with the confidence limits and p-values but could not
reproduce the output from -mfx-. In the end I found another solution.
It is possible to post the results from -mfx- with -ereturn- so that
the results can be saved with -parmest-. In the example below I don't
include the Stata output but the values listed with -parmest- are
identical to those listed with -mfx-. I did this in Stata 8.2.

Friedrich

. sysuse auto
. logit foreign mpg weight
. mfx
. matrix mfx = e(Xmfx_dydx)
. matrix se = diag(e(Xmfx_se_dydx))
. matrix var = se*se
. ereturn post mfx var
. parmest, format(estimate %9.7f stderr %7.5f z %4.2f p %5.3f min95
max95 %8.6f) list(,)

--- Roger Newson <[email protected]> wrote:
> At 16:45 17/10/2005, Friedrich Huebler wrote:
> >Can -parmest- create a dataset with marginal effects and
> >associated standard errors, etc.? If this is not possible what
> >other package could I use for this purpose?
> 
> I presume that Friedrich is planning to use -mfx- to estimate
> marginal effects and elasticities. If you use -parmest- after
> using -mfx- after using an estimation command, then -parmest-
> will create a resultsset containing the original model parameters,
> not the marginal effects and/or elasticities. To create a
> resultsset with 1 observation for each marginal effect or
> elasticity, I think Friedrich's best strategy is to use -svmat-
> to convert the e() matrices returned by -mfx- into variables
> containing the estimates and standard errors, and then to use
> -parmcip- (which is part of the SSC package -metaparm-) to add
> variables containing the confidence limits and P-values.



	
		
__________________________________ 
Yahoo! Mail - PC Magazine Editors' Choice 2005 
http://mail.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/



© Copyright 1996–2024 StataCorp LLC   |   Terms of use   |   Privacy   |   Contact us   |   What's new   |   Site index