Graphics, Figures & Tables ⇒ dcolumn | Combination of Digits and Characters

[sjkiss]

Hi,

I've generated the following table using apsrtable (R) and it works excellent, except for the table cells that mix characters with digits (e.g. the tables cells in the "Wald" row).

The code is generated by a call to sprintf in R, but that does not seem to be the problem. I've used the same function to generate latex code that works with regular text. The problem appears to be mixing characters and and digits in a single table cell. I've tried hard-encoding white space with curly braces, but that doesn't work. I think the problem might be that the table is constructed in dcolumn.
Can any one offer any suggestions?
Simon Kiss

Code: Select all

\documentclass{article}
\usepackage{dcolumn}
\begin{document}
\begin{table}[!ht]
\caption{Comparing Model Fit Of Non vs. Repeated EVents}
\label{} 
\begin{tabular}{ l D{.}{.}{2}D{.}{.}{2} } 
\hline 
  & \multicolumn{ 1 }{ c }{ Non-repeated Events } & \multicolumn{ 1 }{ c }{ Repeated Events } \\ \hline
 %                    & Non-repeated Events  & Repeated Events     \\ 
Environmentalism     & 2.46 ^{***}          & 2.18 ^{**}          \\ 
                     & (0.73)               & (0.70)              \\ 
Partisanship         & -0.36                & -0.31               \\ 
                     & (0.37)               & (0.36)              \\ 
Professionalism      & 0.91                 & 0.75                \\ 
                     & (0.94)               & (0.59)              \\ 
Avg. Health Stories  & 0.66 ^{**}           & 0.50 ^*             \\ 
                     & (0.21)               & (0.21)               \\
 $N$                  & 468                  & 468                 \\ 
Wald                 & 28 {on 4 df}, p = 0.00 & 20 on 4 df, p = 0.00\\ 
Log-Likelihood       & -354.79              & -311.02             \\ 
$R^2$                & 0.11 (Max 0.80)    & 0.07  (Max  0.75)    \\ \hline
 \multicolumn{3}{l}{\footnotesize{Robust standard errors in parentheses}}\\
\multicolumn{3}{l}{\footnotesize{$^\dagger$ significant at $p<.10$; $^* p<.05$; $^{**} p<.01$; $^{***} p<.001$}} 
\end{tabular} 
 \end{table}
\end{document}

[localghost]

The excellent siunitx package is preferable in this case. The manual shows in detail how to handle tabular material like this.


Thorsten

[sjkiss]

That would be a possibility, except that this code is generated via an R function that takes the regression coefficients from a fitted model. So it would be quite extensive to modify that R function to use an entirely different Latex package. It would be useful if there was some simple, minimal change I could put into to the Latex code to accomplish this.

[localghost]

Define a new column type in the preamble of your document which is flexible regarding the format of the digits.

Code: Select all

\newcolumntype{d}[1]{D{.}{.}{#1}}

For details about this please refer to the dcolumn manual.

Modify your table preamble with respect to this new column type and insert the necessary number of integers and decimals before and after the period, respectively. For example

Code: Select all

\begin{tabular}{ld{4.5}d{4.5}} \hline
% table body
\end{tabular}

This reserves the necessary space for four integers before and five decimals behind the period. It serves only as an idea and probably needs some adjustment.