textables: Customized LaTeX tables in R

Created by Thibaut Lamadon and Bradley Setzler, University of Chicago

Overview

This package produces highly-customized LaTeX tables in R. The broad organization of functions is as follows:

• tt_numeric_*: Functions to add numbers to tables;
• tt_text_*: Functions to add text to tables;
• tt_rule_*: Functions to add rules to tables;
• tt_spacer_*: Functions to add spacing to tables;
• tt_tabularize and tt_save: Functions to output the table to LaTeX.

It supports building a table in blocks, in the spirit of ggplot2, using the + and %&% operators for concatenation.

Here is an example of the type of table that this package can easily construct:

The package can be installed with the command devtools::install_github("setzler/textables").

Details

Numeric Columns and Rows: tt_numeric_*

The functions for constructing numeric columns and rows are as follows:

• tt_numeric_column: a numeric column; and,
• tt_numeric_row: a numeric row, with argument cspan that allows the numbers to span multiple columns.

Numeric formatting options include:

• dec: control decimal places, for example, dec=3 displays 3 decimal places;
• se: surround numbers with parenthesis with se=TRUE;
• percentage: add a percentage sign to each number with percentage=TRUE;
• pvalues: use p-values to add stars to indicating significance, for example, pvalues=c(0.005,0.05) would add 3 stars and 2 stars, respectively.

Text Columns and Rows: tt_text_*

The functions for constructing text columns and rows are:

• tt_text_column: a text column.
• tt_text_row: a text row, with argument cspan that allows the text to span multiple columns.

Rules: tt_rule_*

• tt_rule_top: add a top-rule;
• tt_rule_mid: add a mid-rule;
• tt_rule_bottom: add a bottom-rule;
• tt_midrule_partial: add a partial mid-rule (user must supply begin and end points in a list of vectors, e.g., tt_midrule_partial = list(c(1,2),c(3,4))).

Spacers: tt_spacer_*

• tt_spacer_row: adds vertical space between rows, for example, tt_spacer_rows(4.5) adds 4.5pt of vertical space before the next row begins.

Concatenation: + and %&% operators

• +: binds rows together vertically; and,
• %&%: binds columns together horizontally.

The output from tt_numeric_column and tt_text_column can be combined into a single table with +, while the output from tt_numeric_row and tt_text_row could be combined into a single table with %&%.

Finishing and Exporting: tt_tabularize and tt_save

• tt_tabularize: Converts a tt object into a tabular by collapsing into TeX code with begin/end tabular commands;
• tt_save: save as a .tex file. The stand_alone=T option makes it a document that can be compiled directly by LaTeX. The tabularize_output=T option runs tabularize on the tt object before exporting.

Example

This example demonstrates the construction of the example table seen at the beginning of this document.

1. Install and load the package

devtools::install_github("setzler/textables")

library(textables)

2. Construct example regression output

library(data.table)

## Warning: package 'data.table' was built under R version 3.3.2

dd <- data.table(
    sample = c("Full Sample", "Full Sample", "Subsample", "Subsample"), 
    controls = c("No", "Yes", "No", "Yes"), 
    coef = c(1.17, 1.59, 1.105, 1.69),
    SEs = c(.6, .481, .789, .8), 
    pvals = c(.051, .001, .16, .091), 
    N = c(1234567, 1234567, 891011, 891011)
  )

print(dd)

##         sample controls  coef   SEs pvals       N
## 1: Full Sample       No 1.170 0.600 0.051 1234567
## 2: Full Sample      Yes 1.590 0.481 0.001 1234567
## 3:   Subsample       No 1.105 0.789 0.160  891011
## 4:   Subsample      Yes 1.690 0.800 0.091  891011

3. Text rows, rules, and spacers

We start by creating a text row indicating whether or not controls are included:

tt <- with(dd, tt_text_row(controls)) 

print(tt)

## \begin{tabular}{r}
## No & Yes & No & Yes \\
## \end{tabular}

Note that, when printed to the console, a tt object will automatically include the begin/end tabular lines as a convenience in case the user wishes to test out the appearance of the table without relying on textable's exporting methods described below.

We append to this a row label, add 1pt of space, and follow it with a midrule:

tt <- tt_text_row("Controls") %&% tt + tt_spacer_row(1) + tt_rule_mid()

print(tt)

## \begin{tabular}{rr}
## Controls & No & Yes & No & Yes \\[1.000000pt]
## \midrule 
## \end{tabular}

We add a row above with labels that span multiple columns with matching partial midrules:

tt = tt_text_row(" ") %&% with(dd, tt_text_row(unique(sample), cspan=c(2, 2))) + 
  tt_rule_mid_partial(list(c(2,3),c(4,5)))  + tt

print(tt)

## \begin{tabular}{rr}
##   & \multicolumn{2}{c}{Full Sample} & \multicolumn{2}{c}{Subsample} \\
##  \cmidrule(lr){2-3} \cmidrule(lr){4-5} 
## Controls & No & Yes & No & Yes \\[1.000000pt]
## \midrule 
## \end{tabular}

4. Constructing numeric rows

We create a row of coefficient estimates with stars indicating significance and rounding to the third decimal place:

tt <- tt +  tt_text_row("Coefficient ($\\tilde{\\beta}_\\nu$)") %&% 
      with(dd, tt_numeric_row(coef, pvalues=pvals, dec=2))

print(tt)

## \begin{tabular}{rr}
##   & \multicolumn{2}{c}{Full Sample} & \multicolumn{2}{c}{Subsample} \\
##  \cmidrule(lr){2-3} \cmidrule(lr){4-5} 
## Controls & No & Yes & No & Yes \\[1.000000pt]
## \midrule 
## Coefficient ($\tilde{\beta}_\nu$) & 1.17* & 1.59*** & 1.10 & 1.69* \\
## \end{tabular}

We add to this a row of standard errors:

tt <- tt +  tt_text_row("Std. Error") %&% 
      with(dd, tt_numeric_row(SEs, se=T, dec=2))

Finally, we add a row of sample size integers (note the automatic commas, which can be disabled with big.mark=""). We separate the sample size with a midrule:

tt <- tt + tt_rule_mid() + tt_text_row("Sample Size") %&% 
      with(dd, tt_numeric_row(unique(N), cspan=c(2,2), dec=0))

print(tt)

## \begin{tabular}{rr}
##   & \multicolumn{2}{c}{Full Sample} & \multicolumn{2}{c}{Subsample} \\
##  \cmidrule(lr){2-3} \cmidrule(lr){4-5} 
## Controls & No & Yes & No & Yes \\[1.000000pt]
## \midrule 
## Coefficient ($\tilde{\beta}_\nu$) & 1.17* & 1.59*** & 1.10 & 1.69* \\
## Std. Error & (0.60) & (0.48) & (0.79) & (0.80) \\
## \midrule 
## Sample Size & \multicolumn{2}{c}{1,234,567} & \multicolumn{2}{c}{891,011} \\
## \end{tabular}

5. Finishing and saving

We convert the tt object into a LaTeX tabular using tt_tabularize. We can make the tabular prettier with top and bottom rules with the pretty_rules=T option:

tab <- tt_tabularize(tt, header=c("l",rep("r",4)), pretty_rules=T)

print(tab)

##  [1] "\\begin{tabular}{lrrrr}"                                                         
##  [2] "\\toprule "                                                                      
##  [3] "\\midrule "                                                                      
##  [4] "  & \\multicolumn{2}{c}{Full Sample} & \\multicolumn{2}{c}{Subsample} \\\\"      
##  [5] " \\cmidrule(lr){2-3} \\cmidrule(lr){4-5} "                                       
##  [6] "Controls & No & Yes & No & Yes \\\\[1.000000pt]"                                 
##  [7] "\\midrule "                                                                      
##  [8] "Coefficient ($\\tilde{\\beta}_\\nu$) & 1.17* & 1.59*** & 1.10 & 1.69* \\\\"      
##  [9] "Std. Error & (0.60) & (0.48) & (0.79) & (0.80) \\\\"                             
## [10] "\\midrule "                                                                      
## [11] "Sample Size & \\multicolumn{2}{c}{1,234,567} & \\multicolumn{2}{c}{891,011} \\\\"
## [12] "\\midrule "                                                                      
## [13] "\\bottomrule "                                                                   
## [14] "\\end{tabular}"

We can use tt_save to save the tabular to a .tex file. If we want to be able to compile the .tex file directly, we can use the stand_alone=TRUE option, then use R's system command to compile to PDF:

setwd("inst")
tt_save(tab,filename='example.tex',stand_alone=T)
system("pdflatex example.tex")

The resulting table is displayed at the beginning of this document.