# Creating a statistical table

I will be grateful if you can help in creating a statistical table. Actually my problem is mainly in the editing. Here is the table I am trying to create:

In the table, there should be C and C&T columns under both the ADF test and KPSS test columns. However, I cannot center the ADF test and KPSS columns, and I cannot place the values under each column exactly. C and C&T columns also do not fit under the At Level and First Difference columns.

The codes I use are as follows:

\documentclass{article}
\usepackage{booktabs}
\begin{document}
\begin{table}[h!]
\caption{Summary statistics of the variables}
\bigskip
\begin{tabular}{@{}l ccccp{1.5cm} p{0.5cm} ccccp{1.5cm}@{}}
\toprule
& \multicolumn{2}{c}{ADF Test} & \phantom{ab} && \multicolumn{2}{c}{KPSS Test} & \phantom{ab}\\ \cmidrule{2-5}\cmidrule{6-9}
& At Level & & First \newline Difference && At Level & & First \newline  Difference \\ \cmidrule{2-3}\cmidrule{4-5} \cmidrule{6-7} \cmidrule{8-9}
Variables & C & C\&T & C & C\&T && C & C\&T & C & C\&T \\ \midrule
vix & -59.907*** & -59.903*** & -4.382*** & -4.460 && 0.73 & 0.13 & 0.07 & -1.05 \\
NSP & -0.74 & -2.16 & -6.98*** & -1.05 && 0.77 & 0.16 & 0.1 & -1.05\\
\bottomrule
\end{tabular}
\end{table}
\end{document}

• Welcome to TeX.SE.
– Mico
Oct 2, 2021 at 10:27
• Welcome to TSE. Please post a Minimal Working Example, instead of a code snippet. Oct 2, 2021 at 10:31
• I've taken the liberty of augmenting your code (added three lines at the top, and one at the bottom) in order to make it minimally compilable. Feel free to revert.
– Mico
Oct 2, 2021 at 11:10

Like this?

(red lines indicate part of page layout)

Edit: MWE now consider @Mico suggestions (thank you very much, Mico!)

With use of tabularray package and its libraries booktabs and siunitx. Since table is quite wide, reduced are font size and column separation:

\documentclass{article}
\usepackage[skip=1ex]{caption}
\usepackage{tabularray}
\UseTblrLibrary{booktabs, siunitx}

\begin{document}
\begin{table}[ht]
round-precision=2,
tight-spacing,
}
\caption{ADF and KPSS test of (non)stationarity}
\centering
\small                 % <---
\begin{tblr}{colsep=4pt,   % <---
colspec={@{} l *{2}{Q[c,si={table-format=-2.2{***} }]} % <--- changed
Q[c,si={table-format=-1.2{***} }]  % <--- changed
*{5}{Q[c,si={table-format=-1.2}]}
@{} },
row{1}={font=\bfseries}
}
\toprule
& \SetCell[c=4]{c} {{{ADF Test}}}    &   &   &
& \SetCell[c=4]{c}  {{{KPSS Test}}}    &   &    &   \\
\cmidrule[r]{2-5}  \cmidrule[l]{6-9}
& \SetCell[c=2]{c} {{{At Level}}}    &
& \SetCell[c=2]{c} {{{1st Diff.}}}  &
& \SetCell[c=2]{c} {{{At Level}}}  &
& \SetCell[c=2]{c} {{{1st Diff.}}}  \\
\cmidrule[r]{2-3}   \cmidrule[lr]{4-5}
\cmidrule[lr]{6-7}  \cmidrule[lr]{8-9}
Variables   & {{{C}}} & {{{C\&T}}}
& {{{C}}} & {{{C\&T}}}
& {{{C}}} & {{{C\&T}}}
& {{{C}}} & {{{C\&T}}}\\
\midrule
vix         & -59.907***
& -59.903***
& -4.382***
& -4.460
& 0.73
& 0.13
& 0.07
& -1.05 \\
NSP         & -0.74
& -2.16
& -6.98***
& -1.05
& 0.77
& 0.16
& 0.1
& -1.05\\
\bottomrule
\end{tblr}
\end{table}
\end{document}

• See my comment under one of the other answers: It's not good form to write "Difference", as in these tests one can be dealing with both first and higher-order differences. Also, I think there's no reason for showing third decimals; 2 decimals should be more than enough (and eliminating them simplifies/streamlines the table noticeably).
– Mico
Oct 2, 2021 at 13:18
• @Mico, you are right. For my excuse, I just tried to reproduce desired form of table and consequently I didn't dig into the meaning of its content. I will edit my answer ASAP (hopefully site will be available next few hours) and correct it considering your advice. Oct 2, 2021 at 17:07
• Thank you very much indeed for your answer Zarko, it really helped me a lot. Oct 3, 2021 at 12:29

I would use 8 c but no p columns for the 8 tests. I would also strive to use the same number of significant digits throughout. If need be, consider aligning the numbers on the decimal markers.

Oh, and do try to provide a caption that's a tad more informative than "Summary statistics of the variables". In the code below, I suggest "ADF and KPSS tests of (non)stationarity". [Aside: I'm an econometrician. I'm rather familiar with both ADF and KPSS tests, in no small part because the "P" in "KPSS" denotes the first letter of my thesis adviser's surname...)

\documentclass{article}
\usepackage{booktabs}
\newcommand\sym[1]{\ifmmode{^{#1}}\else{$^{#1}$}\fi}
\begin{document}

\begin{table}[h!]
\setlength\tabcolsep{0pt} % let LaTeX figure out amount of intercol. whitespace
\caption{ADF and KPSS tests of (non)stationarity\strut}
\begin{tabular*}{\textwidth}{@{\extracolsep{\fill}} l *{8}{c}}
\toprule
& \multicolumn{4}{c}{ADF Test} & \multicolumn{4}{c}{KPSS Test} \\
\cmidrule{2-5} \cmidrule{6-9}
& \multicolumn{2}{c}{At Level} & \multicolumn{2}{c}{First Diff.}
& \multicolumn{2}{c}{At Level} & \multicolumn{2}{c}{First Diff.} \\
\cmidrule{2-3} \cmidrule{4-5} \cmidrule{6-7} \cmidrule{8-9}
Variables & C & C\&T & C & C\&T & C & C\&T & C & C\&T \\
\midrule
VIX & $-$59.91\sym{***} & $-$59.90\sym{***}
&  $-$4.38\sym{***} & $-$4.46
& 0.73 & 0.13 & 0.07 & $-$1.05 \\
NSP & \hphantom{0}$-$0.74\hphantom{\sym{***}}
& \hphantom{0}$-$2.16\hphantom{\sym{***}}
& $-$6.98\sym{***} & $-$1.05
& 0.77 & 0.16 & 0.10 & $-$1.05\\
\bottomrule
\end{tabular*}
\end{table}

\end{document}

• @Dr.ManuelKuehner - The macro Isn't a test for math mode, it just starts with a test for math mode. If the condition is true, ^{#1} is executed; if it's not, $^{#1}$ is executed. I suppose the outer curly braces aren't necessary. Writing newcommand\sym[1]{\ifmmode^{#1}\else$^{#1}$\fi} would have been more parsimonious -- though not easier for humans to read.
– Mico
Oct 2, 2021 at 20:59
• @Dr.ManuelKuehner - Many thanks for my aside being a "cool remark". :-) My adviser (Peter Phillips) was one of the early researchers of the properties of various types of non-stationarity. My thesis didn't involve the ADF and KPSS tests, though. Instead, I looked into the properties of tests of constancy of (unconditional) second moments if the observed time series are leptokurtic ("heavy-tailed"). The asymptotic theory I developed relied heavily on the properties of stable distributions. I've pursued a non-academic career, but I've never regretted having written a "far-out" thesis. :-)
– Mico
Oct 2, 2021 at 21:17
• Thanks a lot. I will have a look at the references. I really enjoy the community here, I learn about TeX and beyond. Oct 5, 2021 at 14:20
• @Dr.ManuelKuehner - You're most welcome. A minor correction: The research conference in Eltville, Germany was held in late 2005, not late 2006.
– Mico
Oct 5, 2021 at 15:58
• I just now actually see that you wrote that you have worked for "The Mandelbrot"! Sometimes the text does not go from the eyes to the brain :). That's a cool side note. Oct 5, 2021 at 22:11

Edited to take into account @Mico 's comment

\documentclass{article}
\usepackage{booktabs}
\begin{document}
\begin{table}[h!]
\caption{Summary statistics of the variables}
\bigskip
\begin{tabular}{l *8c}
\toprule
& \multicolumn{4}{c}{ADF Test} & \multicolumn{4}{c}{KPSS Test} \\ \cmidrule(rl){2-5}\cmidrule(rl){6-9}
& \multicolumn{2}{c}{At Level }& \multicolumn{2}{c}{Difference} & \multicolumn{2}{c}{At Level }& \multicolumn{2}{c}{Difference} \\ \cmidrule(rl){2-3}\cmidrule(rl){4-5} \cmidrule(rl){6-7} \cmidrule{8-9}
Variables & C & C\&T & C & C\&T & C & C\&T & C & C\&T \\ \midrule
vix & -59.907*** & -59.903*** & -4.382*** & -4.460 & 0.73 & 0.13 & 0.07 & -1.05 \\
NSP & -0.74 & -2.16 & -6.98*** & -1.05 & 0.77 & 0.16 & 0.1 & -1.05\\
\bottomrule
\end{tabular}
\end{table}

\end{document}


• For these tests, you can't just write "difference", as (depending on the circumstance) it could be the first, second or very (rarely) third difference. A separate issue: Your setup offers no assurance that the table will fit inside the text block. (In fact, when I compile your code, LaTeX complains that the table is ca 1.5cm too wide.)
– Mico
Oct 2, 2021 at 12:01
• Hi @Mico. I have delete inadvertently the word #First" before "Difference". You are write the table is too wide and we can fix it as you did by \setlength\tabcolsep{0pt}  Oct 2, 2021 at 13:07
• One respect in which the table in "Zarko"'s answer is better than the one is this answer is that it uses genuine minus signs rather than hyphens. I presume you could get the same effect here by setting the numbers (including, of course, the minus signs) in math mode. Oct 2, 2021 at 20:26
• Thank you @Mico Oct 2, 2021 at 21:25