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

st: RE: Stata Estimates to LaTeX Tables

From   "Joao Pedro W. de Azevedo" <>
To   <>
Subject   st: RE: Stata Estimates to LaTeX Tables
Date   Tue, 11 May 2004 09:53:46 +0100

This is very nice idea.
I couple of suggestions are:
A package that would export marginal effect estimations. I've been doing
with some simple matrix commands but I'm sure it can be incorporated into a
proper ado file.
The option of saving the Std. Erros on the side instead of under the

-----Original Message-----
[] On Behalf Of Fredrik
Sent: Tuesday, May 11, 2004 1:24 AM
Subject: st: Stata Estimates to LaTeX Tables

I assume there are many users on this list that regularly convert Stata
estimates into LaTeX tables and wanted to get some input from you...

There are a number of solutions available to get a stata regression output
into LaTeX. If you only want one single regression, outtex provides a nicely
formatted, if somewhat old, LaTeX code with sufficient features for most
users. However, I almost exclusively report multiple versions/models which
means I want to report them in multiple columns, one model per column. There
are, to the best of my knowledge, two ways to do that. For the direct stata
to LaTeX route there is the est2vec+est2tex package. The other option is to
use outreg to get the output into Excel (actually it is only a tab separated
text file) and then use the Excel2TeX macro to output into LaTeX.

While est2vec/tex is the best route as far as being direct, it lacks certain
features that makes it less useful (and/or adds significant post-processing

My goal is to extend est2vec and est2tex to provide (myself with) a better
solution. Before I get started I wanted to see if I could get some input on
this (and maybe even some help :-)

If I get enough people to show interest I would consider setting up a
sourceforge project for this... if anyone has experience with that, let me

Est2Vec (storing the relevant information in matrices for later export to

Exponential Output
One limitation I've found here so far is the ability to report exponential
forms of the estimates and standard deviations. Lifting code from -outreg- I
have already added the option -eform- to -est2vec- and it seems to be
working (but needs more testing).

Manual Variables
I find that I frequently want to report a table with several control
variables (such as state dummies) suppressed). I am considering allowing the
user to add one (or more) lines to the end of the variable lists. For
example EXTRAVAR(State Dummies)(yes) would add a new variable named "State
Dummies" with a value "yes" to the coefficient matrix (and probably an empty
observation in the errors matrix).

Is there anything else missing??

Est2Tex (create and save LaTeX files from information stored in matrices by

Significance Marks
My first goal here is to allow the use of "fancy" significance marks. Both
-est2tex- and -outreg- simply use *, **, *** (or +,*,**). -Outtex- however
defines \onepc, \fivepc and \tenpc for a much more elegant output. Further,
-outtex- also (depending on options) puts the estimates and the significance
in separate columns (r @{} l). There are two ways to implement the latter,
one is using the same approach as -outtex-, the other is to use the package
-dcolumns-. The latter would be nicer if I could get it to work (so far it
doesn't like macros in the columns and I might have to use the -outtex-
approach). An extension would be to suppress the definition of the
significance terms (\onepc etc) in each table and rely on those being
defined in the header of the document. A similar flag would probably be
added for -dcolumn- column types (if I decide to use -dcolumn-).
Implementation through a new FANCY2 option (or something like that).

Standard Error formatting
-Est2tex- currently does not allow me to put standard errors within brackets
(rather than parenthesis) or include them without a parenthesis. The former
is an option for -outreg- and the latter is an option for -outtex- but I
don't know if it should be a priority... let me know if this is important to
include. Implementation options: NOPAR BRACKET

Long table
-Outtex- allows the use of the -longtable- package and I find that a
necessity for many of my long regression results. This would be implemented
with LONG_TABLE option.

-Est2tex- produces -tabular- output that the user then has to put into a
-table- wrapper should (s)he want a float and/or want a caption and label
for the table. I am considering adding a TA_BLE option that would also be
assumed if the user adds either CAP_TION() or LA_BEL()

I just learned about the package -booktabs- that makes the table conform
closer to predominant publishing standards. I would consider enabling it as
part of the FANCY2 option.

*   For searches and help try:

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