Bookmark and Share

Notice: On March 31, it was announced that Statalist is moving from an email list to a forum. The old list will shut down at the end of May, and its replacement, statalist.org is already up and running.


[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: st: Trouble producing MNL table with outreg


From   John Luke Gallup <jlgallup@pdx.edu>
To   statalist@hsphsun2.harvard.edu
Subject   Re: st: Trouble producing MNL table with outreg
Date   Sun, 9 Sep 2012 13:55:04 -0700

Wameq,

I think you may have a problem using the option -addtable- for the -outreg- command, if you want to use it to combine tables.  This option creates a new table in the same document as a previous table, but the tables are completely separate.

Otherwise I don't see a problem.  Most of your options listed below have nothing to do with marginal effects per se.

I can produce a table of marginal effects for multinomial logit as follows:

. sysuse auto, clear
. mlogit rep78 mpg foreign, nolog baseoutcome(1)

If you just use the command

. margins, dydx(*)

you only get the marginal effects for the first outcome of the -mlogit- model.  To get a full table of marginal effects, you have to calculate them for each outcome, and use the -merge- option in -outreg- to combine them:

. margins, dydx(*) predict(outcome(2))
. outreg, marginal ctitle("",2)
. margins, dydx(*) predict(outcome(3))
. outreg, marginal merge ctitle("",3)
. margins, dydx(*) predict(outcome(4))
. outreg, marginal merge ctitle("",4)
. margins, dydx(*) predict(outcome(5))
. outreg, marginal merge ctitle("",5)

This produces the following table:

                 --------------------------------------------
                              2       3       4        5    
                 --------------------------------------------
                  mpg       -0.002  -0.010  -0.003   0.014  
                            (0.23)  (0.92)  (0.34)  (2.40)* 
                  foreign   -1.334  0.859   0.577    0.245  
                            (0.01)  (0.01)  (0.02)  (0.04)  
                  N           69      69      69      69    
                 --------------------------------------------
                              * p<0.05; ** p<0.01

You can put the same code into a loop:

. outreg, clear
. forvalues o = 2/5 {
	qui margins, dydx(*) predict(outcome(`o'))
	outreg, marginal merge ctitle("",`o') nodisplay
}
. outreg, replay

If you still find that -outreg- is not lining up columns as it should, please create an example using Stata's auto.dta dataset so that I can reproduce your code.

John


On Sep 8, 2012, at 4:05 PM, Wameq Raza <wameq.r@gmail.com> wrote:

> Hello STATA Users,
> I just started working with outreg recently and have had no trouble
> getting tables for my binomial models.
> But when I tried to get margins tables for a MNL model, the marginal
> effects and the standard errors come
> up all over the place. I've specified ctitles but the columns don't
> follow the titles either. I'm not using "nosub"
> option, but the standard errors are coming up beside 2 of the columns
> while the rest are grouped together
> and appear far below the main table. Here's my syntax:
> 
> mlogit hwseen_a tfhhh shg lntotalexp ......
> 
> margins, dydx (_all)
> 
> outreg using results, marginal ctitle("Variable", "Other", "Pharm",
> "Public", "Private") title ("MNL results for seeking   ///
> care for acute illnesses") starloc(1) starlevels(10 5 1) note("The
> figures show marginal effects and standard errors() for ///
> each provider") addtable
> 
> Can anyone suggest what I might be doing wrong?
> 
> Best
> Wameq
> 
> 
> 
> -- 
> W A M E Q   R A Z A
> *
> *   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/


© Copyright 1996–2014 StataCorp LP   |   Terms of use   |   Privacy   |   Contact us   |   Site index