siunitx multicolumn problem

Question

In addition to Right-centered columns in LyX, I have another problem, now with multicolumns. I want to center the number "98" (which spans two columns) above the two other columns. These shall be aligned "right-centered" with S[table-format=2.0]) and "nearer" to each other. See the code and image above.

\begin{table}[H]
\centering
\begin{threeparttable}
\caption{\textbf{\label{tab:ms-soziodemographische-parameter}blabla}}
\begin{tabular}{lS[table-format=2.0]S[math-rm=\mathit,table-format=2.1,input-symbols=()]cc}
\toprule 
 & \multicolumn{2}{>{\centering}p{0.2\columnwidth}}{\textbf{Pat. mit Kopfschmerzen}} &     \textbf{Pat. ohne Kopfschmerzen} & \textbf{p-Wert}\tabularnewline
\midrule
\midrule 
\textbf{Anzahl} & \multicolumn{2}{c}{98} & 82 & \textbf{\textit{0,001}}\tabularnewline
\midrule 
\textbf{Alter: Jahre (SD)} & 41 & (12,9) & 47,4 (12,6) &     \textbf{\textit{0,001}}\tabularnewline
\midrule 
\textbf{Geschlecht: n(\%)} &  &  &  & \textbf{\textit{0,001}}\tabularnewline
- Weiblich & 78 & (79,6) & 47 (57,3) & \tabularnewline
- Männlich & 20 & (20,4) & 35 (42,7) & \tabularnewline
\midrule 
\textbf{KD der MS: Jahre (SD)} & 11,2 & (8,6) & 13,7 (8,9) & 0,056\tabularnewline
\midrule 
\textbf{Verlaufsform: n(\%)} &  &  &  & \textbf{\textit{0,042}}\tabularnewline
- RRMS & 56 & (57,1) & 30 (36,6) & \tabularnewline
- SCP & 29 & (29,6) & 38 (46,3) & \tabularnewline
- PPMS & 11 & (11,2) & 13 (15,9) & \tabularnewline
- CIS & 2 & (2,0) & 1 (1,2) & \tabularnewline
\midrule 
\end{tabular}
\end{threeparttable}
\end{table}

Output looks like:

.

Answer 1

By adding table-number-alignment=center (left, right and the default center-decimal-marker are the other choices) in the column specification, and removing the table-format=x.y specifiers, the 98 is more centered above the columns.

(I don't exactly understand what's going on, as the table-format option should set table-number-alignment to center, according to the manual.)

\documentclass{article}
\usepackage{threeparttable}
\usepackage{siunitx}
\usepackage{booktabs}
\usepackage{array}
\begin{document}
\begin{table}
\centering
\begin{threeparttable}
\caption{\textbf{\label{tab:ms-soziodemographische-parameter}blabla}}
\begin{tabular}{%
  l
  S[table-number-alignment=center]
  S[math-rm=\mathit,input-symbols=(),table-number-alignment=center]
  cc}
\toprule 
 & \multicolumn{2}{>{\centering}p{0.2\columnwidth}}{\textbf{Pat. mit Kopfschmerzen}} &     \textbf{Pat. ohne Kopfschmerzen} & \textbf{p-Wert}\tabularnewline
\midrule
\midrule 
\textbf{Anzahl} & \multicolumn{2}{c}{98} & 82 & \textbf{\textit{0,001}}\tabularnewline
\midrule 
\textbf{Alter: Jahre (SD)} & 41 & (12,9) & 47,4 (12,6) &     \textbf{\textit{0,001}}\tabularnewline
\midrule 
\textbf{Geschlecht: n(\%)} &  &  &  & \textbf{\textit{0,001}}\tabularnewline
- Weiblich & 78 & (79,6) & 47 (57,3) & \tabularnewline
- Männlich & 20 & (20,4) & 35 (42,7) & \tabularnewline
\midrule 
\textbf{KD der MS: Jahre (SD)} & 11,2 & (8,6) & 13,7 (8,9) & 0,056\tabularnewline
\midrule 
\textbf{Verlaufsform: n(\%)} &  &  &  & \textbf{\textit{0,042}}\tabularnewline
- RRMS & 56 & (57,1) & 30 (36,6) & \tabularnewline
- SCP & 29 & (29,6) & 38 (46,3) & \tabularnewline
- PPMS & 11 & (11,2) & 13 (15,9) & \tabularnewline
- CIS & 2 & (2,0) & 1 (1,2) & \tabularnewline
\midrule 
\end{tabular}
\end{threeparttable}
\end{table}
\end{document}

Answer 2

Personally, I'd have to say that changing the alignment of the "98" item and the inter-column width is less pressing than avoiding the current thorough mix-up of the contents of the columns: depending on the row, the column entries either denote years or sample sizes, and/or standard deviations and percentages. I assume that confusing your readers is not your objective. :-)

I've tried to disentangle the contents in the following MWE.

\documentclass{article}
\usepackage[ngerman]{babel}
\usepackage{lmodern,booktabs,threeparttable,siunitx}
\sisetup{locale=DE,table-format=2.1}
\begin{document}

\begin{table}
\centering
\begin{threeparttable}
\caption{blabla}
\label{tab:ms-soziodemographische-parameter}
\begin{tabular}{@{} l *{2}{S S[math-rm=\mathit]} r @{}}
\toprule 
 & \multicolumn{2}{>\centering p{0.18\columnwidth}}{\textbf{Pat. mit Kopfschmerzen}} 
 & \multicolumn{2}{>\centering p{0.18\columnwidth}}{\textbf{Pat. ohne Kopfschmerzen}} 
 & \textbf{p-Wert}\\
\midrule
 & \multicolumn{1}{c}{Jahre}
 & \multicolumn{1}{c}{\textit{SD}}
 & \multicolumn{1}{c}{Jahre}
 & \multicolumn{1}{c}{\textit{SD}} \\
\cmidrule(lr){2-3} \cmidrule(lr){4-5} 
\textbf{Alter} & 41 & 12,9 & 47,4 & 12,6 & 0,001\\    
\textbf{KD der MS} & 11,2 & 8,6 & 13,7 & 8,9 & 0,056\\
\addlinespace
\midrule
 & \multicolumn{1}{l}{\ n}
 & \multicolumn{1}{c}{\textit{\%}}
 & \multicolumn{1}{l}{\ n}
 & \multicolumn{1}{c}{\textit{\%}} \\    
\cmidrule(lr){2-3} \cmidrule(lr){4-5} 
\textbf{Anzahl} & 98 & & 82  & & 0,001  \\[1ex]
\textbf{Geschlecht} &  &  &  & & 0,001  \\[0.25ex]
- Weiblich   & 78 & 79,6 & 47 & 57,3 &  \\    
- M\"annlich & 20 & 20,4 & 35 & 42,7 &  \\[1ex]
\textbf{Verlaufsform} &  &  &  & & 0,042\\[0.25ex]
- RRMS & 56 & 57,1 & 30 & 36,6 & \\    
- SCP  & 29 & 29,6 & 38 & 46,3 & \\    
- PPMS & 11 & 11,2 & 13 & 15,9 & \\    
- CIS  &  2 &  2,0 &  1 &  1,2 & \\    
\midrule 
\end{tabular}
\end{threeparttable}
\end{table}
\end{document}