Unable to use estadd with listcoef after logit and mlogit

[Avery Cohn]

I've searched the web, Statalist, and helpfiles, but I'm stumped with the following problem. I'm unable to use the estadd command to append listcoef postestimation results to e() following logit and following mlogit. In both cases, I get the error "option matrix not allowed r(198); ." A check of ado dir reveals that I have spost13_ado (updated 2/10/15) and estout installed. I'm running Stata 13 MP. Any idea what the problem might be?

[Avery Cohn]

Sorry. I got an error message while posting. This is a duplicate thread. Please reply here - http://www.statalist.org/forums/foru...git-and-mlogit .

[Andrea Maier]

Is there a fix of this problem by now?

[Richard Williams]

Did you check the thread that was linked to?

http://www.statalist.org/forums/foru...git-and-mlogit

[Andrea Maier]

Yes, I did. Thanks anyway.

However I am not really sure what to do after
estadd r(table). Because esttab than does not return me anything of the list coeftable. An example would be: mlogit rep78 c.price##i.foreign listcoef estadd r(table) esttab it only returns values from the mlogit, that I haven't even saved.

[Richard Williams]

esttab wasn't part of the original question, but yes, you often would want to use it or a similar command. My inclination would be to use matrix commands to extract the part of r(table) that you wanted (e.g. e^b) and then use estadd with that matrix.

[Richard Williams]

OK, you could do something like this (requires that spost13 be installed):

webuse nhanes2f, clear
logit diabetes weight height
listcoef
mat list r(table)
matrix table = r(table)
matrix bStdX = table[1..2, 5]'
estadd matrix bStdX = bStdX
esttab , main(bStdX)

. esttab , main(bStdX)

----------------------------
                      (1)   
                 diabetes   
----------------------------
diabetes                    
weight              1.589***
                  (10.66)   

height              0.633***
                  (-9.06)   

_cons                    ***
                   (3.31)   
----------------------------
N                   10335   
----------------------------
bStdX coefficients; t statistics in parentheses
* p<0.05, ** p<0.01, *** p<0.001

[Andrea Maier]

Thank you very much.

Actually, this goes in the right direction, thanks so much.

However, I am having a little problem and dare to ask for help again, please excuse me.

sysuse auto, clear

gen price_cat = 1 if price <= 5000

replace price_cat = 2 if inrange(price, 5001, 10000)

replace price_cat = 3 if price > 10000

mlogit price_cat c.length##i.foreign, baseoutcome(2)

listcoef, lt

mat list r(table)

But I want longer labels for the names in the r(table) matrix and I would wish to only select the rows (every third), that contain the comparision between the groups 1 and 3.

I am not really sure how to work with the matrices, so I really appreciate every help. Thanks so much in advance.

[Brian Yalle]

Hi,
I'm using Stata 16 and I want to present a table result that include standardized coefficients using estout command (esttab, estadd).

I write here because I think post # 7 is somehow related to my question.

Particularly, I want a table that shows y-standardized coefficient (bStdY) and fully standardized coefficient (bStdXY) for logit and probit model. Ben Jann, the autor of estout command, provides a page which give instructions to leverage its potential.

In the final part of "estadd" section (http://repec.sowi.unibe.ch/stata/estout/estadd.html), there is an example how to obtain a table result for 2 models that contains "y-standardized".

. capture program drop estadd_bstdy
. program estadd_bstdy, eclass
1. tempname bstdy
2. matrix `bstdy' = e(b)
3. quietly summarize `e(depvar)' if e(sample)
4. matrix `bstdy' = `bstdy' / r(sd)
5. ereturn matrix bstdy = `bstdy'

sysuse auto
eststo: quietly regress price weight mpg
eststo: quietly regress price weight mpg foreign
estadd bstdy : *
estout, cells(b bstdy(par))
--------------------------------------
                     est1 est2
                  b/bstdy b/bstdy
--------------------------------------
weight  1.746559    3.464706
        (.0005922)  (.0011747)
mpg     -49.51222   21.8536
        (-.0167867) (.0074093)
foreign 3673.06
        (1.245318)
_cons   1946.069    -5853.696
        (.659797) (-1.984643)
--------------------------------------

The codes above were applied in a linear regression.
My goal is to compare bStdy and bStdXY coefficients from logit and probit model. The way I made some modifications to obtain standardized coefficients is:

. capture program drop estadd_bstdy
. program estadd_bstdy, eclass
  1. tempname bstdy
  2. matrix `bstdy' = e(b)
  3. quietly summarize `e(depvar)' if e(sample)
  4. matrix `bstdy' = `bstdy' / r(sd)
  5. ereturn matrix bstdy = `bstdy' 
  6. end

. eststo stdy_logit: logit sick tp t2m ro
. eststo stdy_probit: probit sick tp t2m ro
. estadd bstdy : stdy_logit stdy_probit
. estout, cells(bstdy(fmt(4)))
--------------------------------------
               stdy_logit  stdy_probit
                    bstdy        bstdy
--------------------------------------
sick                               
tp                 0.0011       0.0006
t2m                0.0030       0.0016
ro                 0.0007       0.0005
_cons             -6.3265      -3.7224
--------------------------------------

Although, the results were not as expected and were different from those obtained by listcoef command:

eststo stdy_logit: logit sick tp t2m ro
listcoef, std help

eststo sstdy_probit: probit sick tp t2m ro
listcoef, std help

So my questions are what is wrong in the modified codes? How to obtain bStdY and bStdXY coefficients from logit and probit model in order to present in a table?

Any help will be apreciated.