How Can We Display F-Statistic / Degrees of Freedom in two lines

Question

I am using Stargazer to output a regression table with multiple lm models together. The problem is that Stargazer outputs the F-Statistic and degrees of freedom in a single line by default. If you have 3 or 4 models, that means the output in latex is unnecessarily wide and extends past the page.

There is an option to hide the degrees of freedom using:

stargazer(lm1,lm2,lm3,lm4, df = F)

But, there is no way to get it to display on two rows.

Here's a reproducible example of the bad outcome:

lm.out.1 <- lm(stack.loss ~ Air.Flow , data=stackloss)
lm.out.2 <- lm(stack.loss ~ Water.Temp, data=stackloss)
lm.out.3 <- lm(stack.loss ~ Acid.Conc., data=stackloss)
lm.out.4 <- lm(stack.loss ~ Air.Flow + Water.Temp, data=stackloss)
lm.out.5 <- lm(stack.loss ~ Air.Flow + Water.Temp + Acid.Conc., data=stackloss)

stargazer(lm.out.1, 
          lm.out.2,
          lm.out.3,
          lm.out.4,
          lm.out.5)

When you render the latex output, you get an unnecessarily wide table:

Which is outside the margins when your render to print:

It is possible to see how this table could be neatly and efficiently rendered in a printable manner using the df=F option:

stargazer(lm.out.1, 
          lm.out.2,
          lm.out.3,
          lm.out.4,
          lm.out.5, df = F)

Which produces the following:

BUT, now we have lost the information about the degrees of freedom.

Is there any way to print that information with a line-break so that it prints on two rows of the table?

There seems to be something helpful posted by the package author here - about adding degrees of freedom: Displaying degrees of freedom in stargazer table

I am not sure how to take that and get the desired output of F-Stat and DF on two lines...

Daren

Answer 1

You have to hack the output of stargazer() to make a table that displays F-statistics and dfs in distinct lines. A user-defined function in this answer (show_F_in_two_lines()) will produce a table as shown below.

library(stringr)

show_F_in_two_lines <- function(stargazer) {
  # `Stringr` works better than base's regex 
  require(stringr)

  # If you remove `capture.output()`, not only the modified LaTeX code 
  # but also the original code would show up
  stargazer <- stargazer |>
    capture.output()

  # Reuse the index in which F-statistics are displayed
  position_F <- str_which(stargazer, "F Statistic")

  # Extract only F-statistics
  Fs <- stargazer[position_F] |>
    str_replace_all("\\(.*?\\)", "")

  # Extract only df values and make a new line for them
  dfs <- stargazer[position_F] |>
    str_extract_all("\\(.*?\\)") |>
    unlist() |>
    (
      \(dfs)
      paste0(" & ", dfs, collapse = "")
    )() |>
    paste0(" \\\\")

  # Reuse table elements that are specified
  # after the index of F-statistics
  after_Fs <- stargazer[-seq_len(position_F)]

  c(
    stargazer[seq_len(position_F - 1)],
    Fs,
    dfs,
    after_Fs
  ) |>
    cat(sep = "\n")
}

stargazer(
  header = FALSE,
  lm.out.1,
  lm.out.2,
  lm.out.3,
  lm.out.4,
  lm.out.5
) |>
  show_F_in_two_lines()

Mckissick