Search code examples
latexstatastata-macros

Outputting multiple regressions into a LaTeX document

Outreg2 is a community-contributed command, that helps us easily output the results of regressions run on Stata into a clean table, which can then be viewed in text, Word documents, or in LaTeX.

Using the auto.dta dataset, I run the following regression:

sysuse auto.dta, clear
ssc install outreg2
gen  wtsq  = weight^2
foreach s in price headroom trunk{ 
    xi: reg `s' weight wtsq, vce(robust)
    outreg2 weight wtsq using tab_base_`s'_j, keep(weight wtsq) bdec(3) nocons tex(nopretty) replace
    xi: reg `s' weight wtsq foreign, vce(robust)
    outreg2 weight wtsq foreign using tab_base_`s'_j, keep(weight wtsq foreign) bdec(3) nocons tex(nopretty) append
    xi: reg `s' weight wtsq foreign length, vce(robust)
    outreg2 weight wtsq foreign length using tab_base_`s'_j, keep(weight wtsq foreign length) bdec(3) nocons tex(nopretty) append
}

I get output three .tex files named tab_base_price_j, tab_base_trunk_j, and so on. When I open the .tex files in LaTeX and run them, I obtain the regression tables in PDF in a perfect format, just as I want. However, each of these files in LaTeX has the following format:

\documentclass[]{article}
\setlength{\pdfpagewidth}{8.5in} \setlength{\pdfpageheight}{11in}
\begin{document}
\begin{tabular}{lccc} \hline
 & (1) & (2) & (3) \\
*** ALL THE TABLE VALUES - DELETED from this illustration ***
\end{tabular}
\end{document}

If I want to create a new document (as a journal article or paper format), and I want to input one of these .tex files using \input{tab_base_price_j.tex} in LaTeX, I receive this error: ! LaTeX Error: Can be used only in preamble.

How do I output the regression tables from Stata in a way that the output .tex files do not have \begin{document}, and just start with:

\begin{tabular}{lccc} \hline
 & (1) & (2) & (3) \\
*** ALL THE TABLE VALUES - DELETED from this illustration ***
\end{tabular}
Solution

  • You just need to use the tex(fragment) option:

    sysuse auto.dta, clear
generate  wtsq  = weight^2

foreach s in price headroom trunk { 
    regress `s' weight wtsq, vce(robust)
    outreg2 weight wtsq using tab_base_`s'_j.tex, keep(weight wtsq) bdec(3) nocons tex(fragment)
    regress `s' weight wtsq foreign, vce(robust)
    outreg2 weight wtsq foreign using tab_base_`s'_j.tex, keep(weight wtsq foreign) bdec(3) nocons tex(fragment)
    regress `s' weight wtsq foreign length, vce(robust)
    outreg2 weight wtsq foreign length using tab_base_`s'_j.tex, keep(weight wtsq foreign length) bdec(3) nocons tex(fragment)
}

    You can then input these as parts of a larger document as follows:

    \documentclass[10pt]{article}
\begin{document}
... text before inclusion of table tab_base_price_j.tex ...
\input{tab_base_price_j.tex}
... text after inclusion of table tab_base_price_j.tex ...
\input{tab_base_headroom_j.tex}
... text after inclusion of table tab_base_headroom_j.tex ...
\input{tab_base_trunk_j.tex}
... text after inclusion of table tab_base_trunk_j.tex ...
\end{document}