[Tex/LaTex] Help with regression table

tables

I need some help with building regression tables in Latex.

What I wanted to do is to create three new columns in the right, just like the three ones that are already there with the content for a "scenario 2", just like the "scenario 1" as you will be able to see in the table, with the same dependent variables, same specifications and same covariates.

I send the code for the table I've been working on below:

\documentclass{article}

\usepackage{float}              
\floatstyle{plaintop}           
\restylefloat{table}

\usepackage{booktabs}
\usepackage{dcolumn}
\newcolumntype{d}[1]{D..{#1}}
\newcommand{\mc}[1]{\multicolumn{1}{c@{}}{#1}}

\begin{document}

\begin{table}[H]
\centering 
  \caption{} 
  \label{}

\smallskip
\begin{tabular}{@{} l *{3}{d{2.6}} @{}} 
\toprule
& \multicolumn{3}{c@{}}{Scenario 1} \\ 
\cmidrule(l){2-4} 
& \multicolumn{3}{c@{}}{\textit{test}} \\[1ex] 
& \mc{only treat} & \mc{treat + coduf} & \mc{all covs} \\ 
\midrule
 $treat$ & 0.023 & 0.027 & 0.046 \\ 
  & (0.023) & (0.027) & (0.046) \\ [1ex]
 $coduf$ & & 0.002 & 0.003 \\ 
  & & (0.003) & (0.003) \\ [1ex]
 $mat\_casa$ &  &  & 0.043^{**} \\ 
  &  &  & (0.019) \\  [1ex]
 $abast\_agua2$ &  &  & -0.004 \\ 
  &  &  & (0.008) \\ [1ex]
 $trat\_agua$ &  &  & -0.004 \\ 
  &  &  & (0.012) \\ [1ex]
 $esgoto$ &  &  & 0.015 \\ 
  &  &  & (0.018) \\ [1ex]
 $sexo$ &  &  & 0.193^{***} \\ 
  &  &  & (0.042) \\ [1ex]
 $idade$ &  &  & 0.006^{***} \\ 
  &  &  & (0.001) \\ [1ex]
 $cor\_raca$ &  &  & 0.037^{**} \\ 
  &  &  & (0.015) \\ [1ex]
 $escol$ &  &  & -0.009 \\ 
  &  &  & (0.009) \\ [1ex]
 $comodos$ &  &  & -0.008 \\ 
  &  &  & (0.102) \\ [1ex]
 \textit{constant} & 0.230^{***} & 0.174^{**} & -0.506^{***} \\ 
  & (0.034) & (0.092) & (0.172) \\ [1ex]
\midrule
   F-test & \multicolumn{1}{c}{0.2661} & \multicolumn{1}{c}{0.2122} & \multicolumn{1}{c}{$7.6729^{***}$} \\
   AIC & \multicolumn{1}{c}{37{,}997.73} & \multicolumn{1}{c}{38{,}099.51} & \multicolumn{1}{c}{33{,}477.07} \\
   N & \multicolumn{1}{c}{624} & \multicolumn{1}{c}{624} & \multicolumn{1}{c}{624} \\
\bottomrule
\multicolumn{3}{@{}l@{}}{\footnotesize Nota: $^{*}\, p<0.1$; $^{**}\, p<0.05$; $^{***}\, p<0.01$}
\end{tabular} 
\end{table}

\end{document}

Best Answer

Most likely I do not understand the question, but what prevents you from just adding the columns? Of course, I do not have the data that is to be filled in, so I left it blank. Let me also remark that I would not use $ to make text italic, use \textit{...} instead.

\documentclass{article}
\usepackage[margin=1in]{geometry}
\usepackage{float}              
\floatstyle{plaintop}           
\restylefloat{table}

\usepackage{booktabs}
\usepackage{dcolumn}
\newcolumntype{d}[1]{D..{#1}}
\newcommand{\mc}[1]{\multicolumn{1}{c@{}}{#1}}

\begin{document}

\begin{table}[H]
\centering 
  \caption{} 
  \label{}

\smallskip
\begin{tabular}{@{} l *{6}{d{5.6}} @{}} 
\toprule
& \multicolumn{3}{c@{}}{Scenario 1} & \multicolumn{3}{c@{}}{Scenario 2} \\ 
\cmidrule(l){2-7} 
& \multicolumn{3}{c@{}}{\textit{test}} \\[1ex] 
& \mc{only treat} & \mc{treat + coduf} & \mc{all covs} 
& \mc{only treat} & \mc{treat + coduf} & \mc{all covs}\\ 
\midrule
 $treat$ & 0.023 & 0.027 & 0.046 \\ 
  & (0.023) & (0.027) & (0.046) \\ [1ex]
 $coduf$ & & 0.002 & 0.003 \\ 
  & & (0.003) & (0.003) \\ [1ex]
 $mat\_casa$ &  &  & 0.043^{**} \\ 
  &  &  & (0.019) \\  [1ex]
 $abast\_agua2$ &  &  & -0.004 \\ 
  &  &  & (0.008) \\ [1ex]
 $trat\_agua$ &  &  & -0.004 \\ 
  &  &  & (0.012) \\ [1ex]
 $esgoto$ &  &  & 0.015 \\ 
  &  &  & (0.018) \\ [1ex]
 $sexo$ &  &  & 0.193^{***} \\ 
  &  &  & (0.042) \\ [1ex]
 $idade$ &  &  & 0.006^{***} \\ 
  &  &  & (0.001) \\ [1ex]
 $cor\_raca$ &  &  & 0.037^{**} \\ 
  &  &  & (0.015) \\ [1ex]
 $escol$ &  &  & -0.009 \\ 
  &  &  & (0.009) \\ [1ex]
 $comodos$ &  &  & -0.008 \\ 
  &  &  & (0.102) \\ [1ex]
 \textit{constant} & 0.230^{***} & 0.174^{**} & -0.506^{***} \\ 
  & (0.034) & (0.092) & (0.172) \\ [1ex]
\midrule
   F-test & \multicolumn{1}{c}{0.2661} & \multicolumn{1}{c}{0.2122} & \multicolumn{1}{c}{$7.6729^{***}$} \\
   AIC & \multicolumn{1}{c}{37{,}997.73} & \multicolumn{1}{c}{38{,}099.51} & \multicolumn{1}{c}{33{,}477.07} \\
   N & \multicolumn{1}{c}{624} & \multicolumn{1}{c}{624} & \multicolumn{1}{c}{624} \\
\bottomrule
\multicolumn{3}{@{}l@{}}{\footnotesize Nota: $^{*}\, p<0.1$; $^{**}\, p<0.05$; $^{***}\, p<0.01$}
\end{tabular} 
\end{table}
\end{document}

Related Solutions

[Tex/LaTex] Regression Table Alignment Problem

As exercise for use of threeparttable and siunitx packages. Considered is natural width of table (to my taste it gives more nice result):

\documentclass[11pt,a4paper]{article}
\usepackage[utf8]{inputenc}
\usepackage{amsmath,amsfonts,amssymb}
\usepackage{booktabs}

\usepackage{siunitx}                        % added
\usepackage{threeparttable}                 % added
%\newcommand\mc[1]{\multicolumn{1}{c}{#1}}  % not used now

\begin{document}
    \begin{table}[htb]
    \centering
\sisetup{table-number-alignment = center, % <-- added/changed
         table-space-text-pre ={(},
         table-space-text-post={\textsuperscript{***}},
         input-open-uncertainty={[},
         input-close-uncertainty={]},
         table-align-text-pre=false,
         table-align-text-post=false}
\begin{threeparttable}
    \caption{IV Analysis: Investment on Bribes table}
    \label{tab1}
\begin{tabular}{r 
                S[table-format=-2.3] % <-- adopted to number of digits in numbers in cells
                S[table-format=-1.4] % <-- adopted ...
                S[table-format=-1.5] % <-- adopted ...
                 }
\toprule
                    & \multicolumn{3}{c}{ln\_invest}                                \\
\cmidrule(lr){2-4}
                    &   {(1)}           &   {(2)}           &  {(3)}                \\
\midrule
Bribe               &  -0.167           &  -0.0617          &   0.00939             \\
                    & (-0.90)           & (-0.44)           &  (0.13)               \\
\addlinespace
medium (20-99)      &   0.339           &   0.298           &   1.024\tnote{***}    \\
                    &  (0.90)           &  (0.73)           &  (8.74)               \\
\addlinespace
large (100 and over)&   2.426\tnote{***}&   1.578\tnote{**} &   2.654\tnote{***}    \\
                    &  (4.31)           &  (2.79)           & (17.07)               \\
\addlinespace
Govt=1              &   0.764           &   0.735           &  -0.112               \\
                    &  (1.31)           &  (1.42)           & (-0.63)               \\
\addlinespace
For=1               &  -0.341           &  -0.486           &   0.326\tnote{*}      \\
                    & (-0.95)           & (-1.41)           &  (2.14)               \\
\addlinespace
Obstacle            &   0.111           &   0.0340          &   0.114               \\
                    &  (0.64)           &  (0.20)           &  (1.13)               \\
\addlinespace
Agree               &  -0.123           &  -0.0445          &   0.0196              \\
                    & (-0.66)           & (-0.25)           &  (0.18)               \\
\addlinespace
lnAge               &   0.155           &   0.734\tnote{**} &   0.122               \\
                    &  (0.62)           &  (2.63)           &  (1.90)               \\
\addlinespace
Constant            &  10.13\tnote{***} &   8.684\tnote{***}&   8.062\tnote{***}    \\
                    &  (12.78)          &  (9.98)           & (23.18)               \\
\midrule
Observations        & {6041}            & {4724}            & {1317}                \\
\bottomrule
\end{tabular}
    \smallskip
    \footnotesize
$t$ statistics in parentheses\par
\begin{tablenotes}[para,flushleft]
    \item[*]    $p < 0.10$,
    \item[**]   $p < 0.05$,
    \item[***]  $p < 0.01$
    \end{tablenotes}\par
Note: Robust standard errors in parentheses
\end{threeparttable}
    \end{table}
\end{document}

If you persist to have table width equal to text width, than replace

\begin{tabular}{r ... }
...
\end{tabular}

with

\begin{tabular*}{\linewidth}{@{\hspace{\tabcolsep}\extracolsep{\fill}} r ...}
...
\end{tabular*}

Edit: meanwhile I learned (thanks Heiko Oberdiek, see his answer) how to better define column properties. Considering it I change columns definition, see \sisetup{...} in MWE above. Now for centering of column headers the command \mc{...} is not needed anymore.

Tables – Getting Help with Complicated Table Conversion Using excel2latex

A solution with {NiceTabular} of nicematrix.

The rules of each sub-table are drawn by a command \Block (with the key hvlines) inserted in top-left cell of the sub-table.

The thick rule forming right angle is drawn with Tikz in the \CodeAfter after the construction of the tabular (using PGF/Tikz nodes created by nicematrix under the rows, columns and cells of the array).

\documentclass[10pt]{article}

\usepackage{nicematrix,tikz}

\begin{document}

\sffamily
\small
\renewcommand{\arraystretch}{1.8}
\setlength{\tabcolsep}{2pt}
\newcolumntype{R}{w{r}{8mm}}

\begin{NiceTabular}{clRw{c}{4mm}RRRRRcRl}[cell-space-top-limit=2pt]
\Block{5-1}{\rotate some long text here}
& \Block[hvlines]{5-2}{}
  qwerty 1 & 1 & & \Block[hvlines]{5-5}{}
                   1 & 2 & 3 & 4 & 5 & & \Block[hvlines]{4-2}{}
                                         4 & text \\
& qwerty 2 & 2 & & 1 & 2 & 3 & 4 & 5 & & 3 & text text \\
& qwerty 3 & 3 & & 1 & 2 & 3 & 4 & 5 & & 2 & text text text \\
& qwerty 4 & 4 & & 1 & 2 & 3 & 4 & 5 & & 1 & text \\
& qwerty 5 & 5 & & 1 & 2 & 3 & 4 & 5 \\
\\
&          &   & & \Block[hvlines]{2-5}{}
                   1 & 2 & 3 & 4 & 5 & & \Block[C,draw]{2-2}{centered text }\\
\RowStyle{\rotate}
&          &   & & abc & abc abc & abc & abc abc & abc abc abc \\  
\CodeAfter
  \tikz \draw [thick] (1-|4.5) |- (6.5-|10) ;
\end{NiceTabular}

\end{document}

Answer to the question in the comment

I have non easy solution. I have changed the columns w{r}{8mm} in columns p[r]{8mm} because, in the columns p, it's possible to change the horizontal alignement with \RowStyle{\centering}. However, that creates a problem for the rotatede elements which are now wrapped at 8 mm (maybe I should change that in nicematrix). That's why I have use \makebox...

\documentclass{article}
\usepackage{nicematrix,tikz}

\begin{document}

\sffamily
\small
\renewcommand{\arraystretch}{1.8}
\setlength{\tabcolsep}{2pt}
\newcolumntype{R}{p[r]{8mm}}

\begin{NiceTabular}{clRw{c}{4mm}RRRRRcRl}[cell-space-top-limit=2pt]
\Block{5-1}{\rotate some long text here}
& \Block[hvlines]{5-2}{}
  qwerty 1 & 1 & & \Block[hvlines]{5-5}{}
                   1 & 2 & 3 & 4 & 5 & & \Block[hvlines]{4-2}{}
                                         4 & text \\
& qwerty 2 & 2 & & 1 & 2 & 3 & 4 & 5 & & 3 & text text \\
& qwerty 3 & 3 & & 1 & 2 & 3 & 4 & 5 & & 2 & text text text \\
& qwerty 4 & 4 & & 1 & 2 & 3 & 4 & 5 & & 1 & text \\
& qwerty 5 & 5 & & 1 & 2 & 3 & 4 & 5 \\
\\
\RowStyle{\rotate\centering}
&          &   & & \Block[hvlines]{2-5}{}
                   1 & 2 & 3 & 4 & 5 & & \Block[C,draw]{2-2}{centered text }\\
\noalign{\vskip 11mm}
\RowStyle{\rotate\centering}
&          &   & & \makebox[18mm][r]{abc} & \makebox[18mm][r]{abc abc} & \makebox[18mm][r]{abc} & \makebox[18mm][r]{abc abc} & \makebox[18mm][r]{abc abc abc} \\  
\CodeAfter
  \tikz \draw [thick] (1-|4.5) |- (6.5-|10) ;
\end{NiceTabular}

\end{document}

Best Answer

Related Solutions

[Tex/LaTex] Regression Table Alignment Problem

Tables – Getting Help with Complicated Table Conversion Using excel2latex

Related Question