Export summary stats table by group variable to LaTex

[Noor Sethi]

Dear Statalisters,

Is there a way to output a summary stats table (like the one below) to LaTex using Stata 17? Essentially, the table must summarize a list of variables by a group variable (i.e. Female dummy). So far, I've tried using

tabstat

with no luck.

I apologize for not pasting any state code that I've tried. Nothing has come close to producing such a table.

Thanks in advance!

{
\def\sym#1{\ifmmode^{#1}\else\(^{#1}\)\fi}
\begin{tabular}{l*{1}{rrrr}} \hline \hline
                                                &        Mean&   Std. Dev.&         Min&         Max\\ \hline
Female                                  &            \\[.25cm]
Earnings, in thousands of US dollars            &      243.89&      293.30&          25&        3000\\
\hspace{.2in} Female                         &      196.23&      209.93&          25&        3000\\
\hspace{.2in} Male                         &      263.87&      319.80&          25&        3000\\ [.25cm]
Age                                             &       46.73&        8.45&          29&          75\\
\hspace{.2in} Female                         &       45.89&        9.17&          29&          75\\
\hspace{.2in} Male                         &       47.08&        8.11&          29&          68\\[.25cm]
Average number of hours worked per week    &       50.20&       11.79&          10&         100\\
\hspace{.2in} Female                     &       47.47&       11.13&          10&         100\\
\hspace{.2in} Male                         &       51.35&       11.87&          10&         100\\[.25cm]
At least one child                          &        0.78&        0.41&           0&           1\\
\hspace{.2in} Female                     &        0.69&        0.46&           0&           1\\
\hspace{.2in} Male                         &        0.82&        0.38&           0&           1\\[.25cm]
At least one child below the age of 3            &        0.13&        0.33&           0&           1\\
\hspace{.2in} Female                     &        0.12&        0.33&           0&           1\\
\hspace{.2in} Male                         &        0.13&        0.34&           0&           1\\[.25cm]
Married at time of survey (2022)            &        0.78&        0.42&           0&           1\\
\hspace{.2in} Female                      &        0.68&        0.47&           0&           1\\
\hspace{.2in} Male                         &        0.82&        0.39&           0&           1\\
\hline \hline
\end{tabular}
}

[Jeff Pitblado (StataCorp)]

Yes, in Stata 17 table was re-implemented using the new collect commands.
Here is an example using the auto data.

sysuse auto

* list of variables to summarize
local vlist mpg turn trunk displ

* compute overall statistics and store them in a collection named 'overall'
table (var) (result), ///
    stat(mean `vlist') ///
    stat(sd `vlist') ///
    stat(min `vlist') ///
    stat(max `vlist') ///
    name(overall) ///
    nototal
collect addtags foreign[_hide]

* compute statistics within levels of variable 'foreign' and store them in a 
* collection named 'by'
table (var foreign) (result), ///
    stat(mean `vlist') ///
    stat(sd `vlist') ///
    stat(min `vlist') ///
    stat(max `vlist') ///
    name(by) ///
    nototal

* get levels of the 'by' variable 'foreign' and create a macro to later use
* to set the automatic levels for 'foreign'
collect levels foreign
local auto _hide `s(levels)'
di "`auto'"

* combine the collections
collect combine both = by overall

* fix the autolevels of 'foreign'
collect style autolevels foreign `auto', clear

* change the result labels to match the original LaTeX example
collect label levels result sd "Std. Dev." min "Min" max "Max", modify

* other style changes
collect style cell result[mean sd], nformat(%18.2fc)
collect style header foreign, title(hide)

* set the layout and verify the table look like we want
collect layout (var#foreign) (result)

* export to LaTeX file
collect export table.tex, replace

Here is the resulting table.

. collect layout (var#foreign) (result)

Collection: both
      Rows: var#foreign
   Columns: result
   Table 1: 12 x 4

--------------------------------------------------------
                       |    Mean   Std. Dev.   Min   Max
-----------------------+--------------------------------
Mileage (mpg)          |   21.30        5.79    12    41
  Domestic             |   19.83        4.74    12    34
  Foreign              |   24.77        6.61    14    41
Turn circle (ft.)      |   39.65        4.40    31    51
  Domestic             |   41.44        3.97    31    51
  Foreign              |   35.41        1.50    32    38
Trunk space (cu. ft.)  |   13.76        4.28     5    23
  Domestic             |   14.75        4.31     7    23
  Foreign              |   11.41        3.22     5    16
Displacement (cu. in.) |  197.30       91.84    79   425
  Domestic             |  233.71       85.26    86   425
  Foreign              |  111.23       24.88    79   163
--------------------------------------------------------

Here is the resulting LaTeX file

\documentclass{article}
\usepackage{multirow}
\usepackage{amsmath}
\usepackage{ulem}
\usepackage[table]{xcolor}
\begin{document}
\begin{table}[!h]
\centering
\begin{tabular}{lllll}
\cline{1-5}
\multicolumn{1}{c}{} &
  \multicolumn{1}{|r}{Mean} &
  \multicolumn{1}{r}{Std. Dev.} &
  \multicolumn{1}{r}{Min} &
  \multicolumn{1}{r}{Max} \\
\cline{1-5}
\multicolumn{1}{l}{Mileage (mpg)} &
  \multicolumn{1}{|r}{21.30} &
  \multicolumn{1}{r}{5.79} &
  \multicolumn{1}{r}{12} &
  \multicolumn{1}{r}{41} \\
\multicolumn{1}{l}{\hspace{1em}Domestic} &
  \multicolumn{1}{|r}{19.83} &
  \multicolumn{1}{r}{4.74} &
  \multicolumn{1}{r}{12} &
  \multicolumn{1}{r}{34} \\
\multicolumn{1}{l}{\hspace{1em}Foreign} &
  \multicolumn{1}{|r}{24.77} &
  \multicolumn{1}{r}{6.61} &
  \multicolumn{1}{r}{14} &
  \multicolumn{1}{r}{41} \\
\multicolumn{1}{l}{Turn circle (ft.)} &
  \multicolumn{1}{|r}{39.65} &
  \multicolumn{1}{r}{4.40} &
  \multicolumn{1}{r}{31} &
  \multicolumn{1}{r}{51} \\
\multicolumn{1}{l}{\hspace{1em}Domestic} &
  \multicolumn{1}{|r}{41.44} &
  \multicolumn{1}{r}{3.97} &
  \multicolumn{1}{r}{31} &
  \multicolumn{1}{r}{51} \\
\multicolumn{1}{l}{\hspace{1em}Foreign} &
  \multicolumn{1}{|r}{35.41} &
  \multicolumn{1}{r}{1.50} &
  \multicolumn{1}{r}{32} &
  \multicolumn{1}{r}{38} \\
\multicolumn{1}{l}{Trunk space (cu. ft.)} &
  \multicolumn{1}{|r}{13.76} &
  \multicolumn{1}{r}{4.28} &
  \multicolumn{1}{r}{5} &
  \multicolumn{1}{r}{23} \\
\multicolumn{1}{l}{\hspace{1em}Domestic} &
  \multicolumn{1}{|r}{14.75} &
  \multicolumn{1}{r}{4.31} &
  \multicolumn{1}{r}{7} &
  \multicolumn{1}{r}{23} \\
\multicolumn{1}{l}{\hspace{1em}Foreign} &
  \multicolumn{1}{|r}{11.41} &
  \multicolumn{1}{r}{3.22} &
  \multicolumn{1}{r}{5} &
  \multicolumn{1}{r}{16} \\
\multicolumn{1}{l}{Displacement (cu. in.)} &
  \multicolumn{1}{|r}{197.30} &
  \multicolumn{1}{r}{91.84} &
  \multicolumn{1}{r}{79} &
  \multicolumn{1}{r}{425} \\
\multicolumn{1}{l}{\hspace{1em}Domestic} &
  \multicolumn{1}{|r}{233.71} &
  \multicolumn{1}{r}{85.26} &
  \multicolumn{1}{r}{86} &
  \multicolumn{1}{r}{425} \\
\multicolumn{1}{l}{\hspace{1em}Foreign} &
  \multicolumn{1}{|r}{111.23} &
  \multicolumn{1}{r}{24.88} &
  \multicolumn{1}{r}{79} &
  \multicolumn{1}{r}{163} \\
\cline{1-5}
\end{tabular}
\end{table}
\end{document}

Here is a screenshot of the resulting PDF.

[Jeff Pitblado (StataCorp)]

That felt like too much work. I was trying to reproduce the original example layout. You can also do this with a single call to table.

The following reproduces the above table, with a little less effort.

sysuse auto

local vlist mpg turn trunk displ

* compute summary statistics
table (var foreign) (result), ///
    stat(mean `vlist') ///
    stat(sd `vlist') ///
    stat(min `vlist') ///
    stat(max `vlist') ///
    name(by)

* reorder the levels of 'foreign' and hide the 'Total' level
collect levelsof foreign
collect style autolevels foreign .m `s(levels)', clear
collect style header foreign[.m], level(hide)

* change the result labels to match the original LaTeX example
collect label levels result sd "Std. Dev." min "Min" max "Max", modify

* other style changes
collect style cell result[mean sd], nformat(%18.2fc)
collect style header foreign, title(hide)

* review table look
collect preview

* export to LaTeX file
collect export table.tex, replace

[Noor Sethi]

Beautiful! Thank you, Jeff.

I cannot believe I didn't know about this set of commands- game changer. I'm pasting your code with the modifications I made in bold to get the precise output I wanted.

sysuse auto
local vlist mpg turn trunk displ  

* compute summary statistics
table (var foreign) (result), ///
stat(mean `vlist') ///    
stat(sd `vlist') ///    
stat(min `vlist') ///    
stat(max `vlist') ///    
name(by)  

* reorder the levels of 'foreign' and hide the 'Total' level
collect levelsof foreign
collect style autolevels foreign .m `s(levels)', clear
collect style header foreign[.m], level(hide)
 
* change the result labels to match the original LaTeX example
collect label levels result sd "Std. Dev." min "Min" max "Max", modify  

* other style changes
collect style cell result[mean sd], nformat(%18.2fc)
collect style header foreign, title(hide)
collect style row stack, nobinder spacer // add space between each variable being summarized
collect style cell border_block, border(right, pattern(nil)) // remove the border that appeared to the right of the row labels

* review table look
collect preview  

* export to LaTeX file
collect title "Descriptive Statistics" // add title
collect export table.tex, tableonly replace // leave out Latex preamble in export