Exporting LaTeX regression table with 2 group classifications, using esttab

[Paula de Souza Leao Spinola]

I am using the command esttab in Stata to export a regression table. I would like to label my models with two different levels of group classification.

So far I have the Stata code below:

sysuse auto, clear
gen lowcost = (price<6000)

reg price weight mpg if foreign==0 & lowcost==0
est sto reg1
reg price weight mpg if foreign==0 & lowcost==1
est sto reg2
reg price weight mpg if foreign==1 & lowcost==0
est sto reg3
reg price weight mpg if foreign==1 & lowcost==1
est sto reg4

local regressions reg1 reg2 reg3 reg4
esttab `regressions' using "Latex/test.tex", replace mgroups("Domestic" "Foreign", pattern(1 0 1 0) prefix(\multicolumn{@span}{c}{) suffix(}) span) mtitles("High-cost" "Low-cost" "High-cost" "Low-cost") nonumbers

I would like to add another group classification for all the 4 models with the following label: "Outcome: Price". In my understanding, this would require us to add the following option to the esttab command above:

mgroups("Outcome: Price", pattern(1 0 0 0) prefix(\multicolumn{@span}{c}{) suffix(}) span)

The problem is that esttab doesn't allow us to use the option -mgroups()- more than once. Does anyone have a solution to this?

To see the tabel in LaTeX:

\documentclass{article}
\begin{document}
    \input{test.tex}
\end{document}

[Andrew Musau]

estout is from SSC (FAQ Advice #12). You can attempt to include this as a -prehead-

sysuse auto, clear
gen lowcost = (price<6000)

reg price weight mpg if foreign==0 & lowcost==0
est sto reg1
reg price weight mpg if foreign==0 & lowcost==1
est sto reg2
reg price weight mpg if foreign==1 & lowcost==0
est sto reg3
reg price weight mpg if foreign==1 & lowcost==1
est sto reg4

local regressions reg1 reg2 reg3 reg4

esttab `regressions', replace prehead(&\multicolumn{4}{c}{Outcome: Price} \\) ///
mgroups("Domestic" "Foreign", pattern(1 0 1 0) prefix(\multicolumn{@span}{c}{) suffix(}) span) ///
mtitles("High-cost" "Low-cost" "High-cost" "Low-cost") nonumbers tex

Res.:

. esttab `regressions', replace prehead(&\multicolumn{4}{c}{Outcome: Price} \\) ///
> mgroups("Domestic" "Foreign", pattern(1 0 1 0) prefix(\multicolumn{@span}{c}{) suffix(}) span) ///
> mtitles("High-cost" "Low-cost" "High-cost" "Low-cost") nonumbers tex

&\multicolumn{4}{c}{Outcome: Price} \\
            &\multicolumn{2}{c}{Domestic}               &\multicolumn{2}{c}{Foreign}                \\
            &\multicolumn{1}{c}{High-cost}&\multicolumn{1}{c}{Low-cost}&\multicolumn{1}{c}{High-cost}&\multicolumn{1}{c}{Low-cost}\\
\hline
weight      &       3.885         &       0.828\sym{*}  &       4.763\sym{*}  &       1.870         \\
            &      (1.71)         &      (2.39)         &      (3.36)         &      (2.02)         \\
[1em]
mpg         &       175.7         &       23.98         &       11.64         &      -8.623         \\
            &      (0.56)         &      (0.53)         &      (0.07)         &     (-0.26)         \\
[1em]
\_cons      &     -8192.2         &      1502.7         &     -3848.0         &      1000.1         \\
            &     (-0.62)         &      (0.77)         &     (-0.54)         &      (0.39)         \\
\hline
\(N\)       &          14         &          38         &           9         &          13         \\
\hline\hline
\multicolumn{5}{l}{\footnotesize \textit{t} statistics in parentheses}\\
\multicolumn{5}{l}{\footnotesize \sym{*} \(p<0.05\), \sym{**} \(p<0.01\), \sym{***} \(p<0.001\)}\\
\end{tabular}
}

[Paula de Souza Leao Spinola]

Many thanks Andrew Musau!

I don't know why something seems wrong with the LaTeX code that is outputed from your code. For instance, I don't see

\begin{tabular}

in the beginning of the tex code. Am I missing something?

Many thanks again.

[Nick Cox]

Cross-posted at https://stackoverflow.com/questions/...s-using-esttab

Please note our cross-posting policy, which is that you are asked to tell us about it. https://www.statalist.org/forums/help#crossposting

[Andrew Musau]

May be a bug. Try adding:

esttab `regressions', replace prehead("\begin{tabular}{l*{@M}{r}}" "\hline \hline" "&\multicolumn{4}{c}{Outcome: Price} \\") ///
mgroups("Domestic" "Foreign", pattern(1 0 1 0) prefix(\multicolumn{@span}{c}{) suffix(}) span) ///
mtitles("High-cost" "Low-cost" "High-cost" "Low-cost") nonumbers tex

[Steph Owusu]

Andrew Musau Thanks for your answers here! I have a quick clarification question; Say, I would like to run four more models for outcome "Tax" in addition to the previous Outcome "Price". Assume that all other things remain the same. How do I get four more columns to the right side of "Outcome: Price" that span under say "Outcome: Tax". I would still like to have mgroups("Domestic" "Foreign", pattern(1 0 1 0)) under "Outcome: Tax" and also still like to have mtitles("High-cost" "Low-cost" "High-cost" "Low-cost") under "Outcome: Tax". So I will have eight columns in total.
Thanks a lot!

[Andrew Musau]

We can recycle the same set of estimates in #2 to illustrate this. The following is not compiled, so you may have to adjust the column positions of the model groups. Otherwise, this should do it for what you want. Changes are highlighted.

sysuse auto, clear
gen lowcost = (price<6000)

reg price weight mpg if foreign==0 & lowcost==0
est sto reg1
reg price weight mpg if foreign==0 & lowcost==1
est sto reg2
reg price weight mpg if foreign==1 & lowcost==0
est sto reg3
reg price weight mpg if foreign==1 & lowcost==1
est sto reg4

local regressions reg1 reg2 reg3 reg4

esttab `regressions' `regressions' using myfile.tex, replace prehead(&\multicolumn{4}{c}{Outcome: Price} &\multicolumn{8}{c}{Outcome: Tax}\\) ///
mgroups("Domestic" "Foreign" "Domestic" "Foreign", pattern(1 0 1 0 1 0 1 0) prefix(\multicolumn{@span}{c}{) suffix(}) span) ///
mtitles("High-cost" "Low-cost" "High-cost" "Low-cost" "High-cost" "Low-cost" "High-cost" "Low-cost") nonumbers tex

Res.:

&\multicolumn{4}{c}{Outcome: Price} &\multicolumn{8}{c}{Outcome: Tax}\\

&\multicolumn{2}{c}{Domestic} &\multicolumn{2}{c}{Foreign} &\multicolumn{2}{c}{Domestic} &\multicolumn{2}{c}{Foreign} \\

&\multicolumn{1}{c}{High-cost}&\multicolumn{1}{c}{Low-cost}&\multicolumn{1}{c}{High-cost}&\multicolumn{1}{c}{Low-cost}&\multicolumn{1}{c}{High-cost}&\multicolumn{1}{c}{Low-cost}&\multicolumn{1}{c}{High-cost}&\multicolumn{1}{c}{Low-cost}\\

\hline

weight & 3.885 & 0.828\sym{*} & 4.763\sym{*} & 1.870 & 3.885 & 0.828\sym{*} & 4.763\sym{*} & 1.870 \\

& (1.71) & (2.39) & (3.36) & (2.02) & (1.71) & (2.39) & (3.36) & (2.02) \\

[1em]

mpg & 175.7 & 23.98 & 11.64 & -8.623 & 175.7 & 23.98 & 11.64 & -8.623 \\

& (0.56) & (0.53) & (0.07) & (-0.26) & (0.56) & (0.53) & (0.07) & (-0.26) \\

[1em]

\_cons & -8192.2 & 1502.7 & -3848.0 & 1000.1 & -8192.2 & 1502.7 & -3848.0 & 1000.1 \\

& (-0.62) & (0.77) & (-0.54) & (0.39) & (-0.62) & (0.77) & (-0.54) & (0.39) \\

\hline

\(N\) & 14 & 38 & 9 & 13 & 14 & 38 & 9 & 13 \\

\hline\hline

\multicolumn{9}{l}{\footnotesize \textit{t} statistics in parentheses}\\

\multicolumn{9}{l}{\footnotesize \sym{*} \(p<0.05\), \sym{**} \(p<0.01\), \sym{***} \(p<0.001\)}\\

\end{tabular}

}

[Steph Owusu]

Thank you Andrew Musau. The only change I had to make to get me what I wanted is change &\multicolumn{8}{c}{Outcome: Tax} to &\multicolumn{4}{c}{Outcome: Tax}. This is how my code looks

 sysuse auto, clear
gen lowcost = (price<6000)

reg price weight mpg if foreign==0 & lowcost==0
est sto reg1
reg price weight mpg if foreign==0 & lowcost==1
est sto reg2
reg price weight mpg if foreign==1 & lowcost==0
est sto reg3
reg price weight mpg if foreign==1 & lowcost==1
est sto reg4

local regressions reg1 reg2 reg3 reg4

esttab `regressions' `regressions' using myfile.tex, replace  prehead("\begin{tabular}{l*{@M}{r}}\hline \hline" "&\multicolumn{4}{c}{Outcome: Price} &\multicolumn{4}{c}{Outcome: Tax}\\") mgroups("Domestic" "Foreign" "Domestic" "Foreign", pattern(1 0 1 0 1 0 1 0) prefix(\multicolumn{@span}{c}{) suffix(}) span) mtitles("High-cost" "Low-cost" "High-cost" "Low-cost" "High-cost" "Low-cost" "High-cost" "Low-cost") nonumbers tex