Adding multicolumns to my texreg output

Question

I want to make a table with grouped columns via texreg. I can only see options for grouped rows (groups).

Here's an example:

set.seed(01349)
DF <- data.frame(y = rnorm(100), x1A = rnorm(100), x2A = rnorm(100),
                 x1B = rnorm(100), x2B = rnorm(100))
regs <- lapply(paste0("x", 1:2, c("A", "A", "B", "B")), function(x)
          lm(paste0("y ~ ", x), data = DF))

Here's as close as I can get with plain texreg:

texreg(regs, custom.coef.names = c("Intercept", rep("x", 4)),
       custom.model.names = c("1", "2", "1", "2"))

With LaTeX output:

\begin{table}
\begin{center}
\begin{tabular}{l c c c c }
\hline
           & 1 & 2 & 1 & 2 \\
\hline
Intercept  & $-0.13$  & $-0.13$  & $-0.11$  & $-0.11$  \\
           & $(0.12)$ & $(0.12)$ & $(0.12)$ & $(0.12)$ \\
x          & $0.02$   & $0.07$   & $0.13$   & $-0.11$  \\
           & $(0.13)$ & $(0.12)$ & $(0.12)$ & $(0.13)$ \\
\hline
R$^2$      & 0.00     & 0.00     & 0.01     & 0.01     \\
Adj. R$^2$ & -0.01    & -0.01    & 0.00     & -0.00    \\
Num. obs.  & 100      & 100      & 100      & 100      \\
RMSE       & 1.18     & 1.17     & 1.17     & 1.17     \\
\hline
\multicolumn{5}{l}{\scriptsize{$^{***}p<0.001$, $^{**}p<0.01$, $^*p<0.05$}}
\end{tabular}
\caption{Statistical models}
\label{table:coefficients}
\end{center}
\end{table}

I'd prefer an extra line (highlighted with % comments):

\begin{table}
\begin{center}
\begin{tabular}{l c c c c }
\hline
%*************A HEADER LINE HERE*********************
 & \multicolumn{2}{c}{A} & \multicolumn{2}{c}{B} \\ %
%****************************************************
           & 1 & 2 & 1 & 2 \\
\hline
Intercept  & $-0.13$  & $-0.13$  & $-0.11$  & $-0.11$  \\
           & $(0.12)$ & $(0.12)$ & $(0.12)$ & $(0.12)$ \\
x          & $0.02$   & $0.07$   & $0.13$   & $-0.11$  \\
           & $(0.13)$ & $(0.12)$ & $(0.12)$ & $(0.13)$ \\
\hline
R$^2$      & 0.00     & 0.00     & 0.01     & 0.01     \\
Adj. R$^2$ & -0.01    & -0.01    & 0.00     & -0.00    \\
Num. obs.  & 100      & 100      & 100      & 100      \\
RMSE       & 1.18     & 1.17     & 1.17     & 1.17     \\
\hline
\multicolumn{5}{l}{\scriptsize{$^{***}p<0.001$, $^{**}p<0.01$, $^*p<0.05$}}
\end{tabular}
\caption{Statistical models}
\label{table:coefficients}
\end{center}
\end{table}

Am I missing something, or is there no built-in way to do this?

My workaround is:

x <- capture.output(texreg(
  regs, custom.coef.names = c("Intercept", rep("x", 4)),
  custom.model.names = c("1", "2", "1", "2")))

x[6] <- paste0("& \\multicolumn{2}{c}{A} & \\multicolumn{2}{c}{B} \\\\ \n", x[6])

cat(x, sep = "\n")

But that's obviously a bit duct-tape-y.

Answer 1

This may be late but still useful.

A new version of texreg (1.36.28) just came out on GitHub (not on CRAN yet, May 22nd, 2020). It adds the option custom.header to create groups of regressions.

Applied to your example, it creates:

library(texreg)

set.seed(01349)
DF <- data.frame(y = rnorm(100), x1A = rnorm(100), x2A = rnorm(100),
                 x1B = rnorm(100), x2B = rnorm(100))
regs <- lapply(paste0("x", 1:2, c("A", "A", "B", "B")), function(x)
  lm(paste0("y ~ ", x), data = DF))

screenreg(
  regs, 
  custom.header = list("A" = 1:2, "B" = 3:4),
  custom.coef.names = c("Intercept", rep("x", 4)),
  custom.model.names = c("1", "2", "1", "2"),
)

=============================================
                   A                 B       
           ----------------  ----------------
           1        2        1        2      
---------------------------------------------
Intercept   -0.13    -0.13    -0.11    -0.11 
            (0.12)   (0.12)   (0.12)   (0.12)
x            0.02     0.07     0.13    -0.11 
            (0.13)   (0.12)   (0.12)   (0.13)
---------------------------------------------
R^2          0.00     0.00     0.01     0.01 
Adj. R^2    -0.01    -0.01     0.00    -0.00 
Num. obs.  100      100      100      100    
=============================================
*** p < 0.001; ** p < 0.01; * p < 0.05

I used screenreg() to show the output more easily but it works with texreg() too.