APA formatted tables with easystats

[DominiqueMakowski]

Arguably, this is outside of easystats' scope.
But given that we already have export_table and format_table and some table utilities, I wonder how much work would it be to add a style="apa" thing to consistently (LateX + html) style it in APA style. Probably a feature that so many people need to use, and yet the landscape of current solutions is scattered and frankly not satisfactory

[strengejacke]

I don't know how APA style looks like.

[DominiqueMakowski]

[strengejacke]

I mean, who formats table when submitting a paper? It's done at a later stage by the publisher, anyway?

[strengejacke]

That doesn't mean we should not support APA style, but I don't know how the formatting looks like... so...

[DominiqueMakowski]

who formats table when submitting a paper?

Well technically you are right, but I use apaquarto to format my preprints, and it's nice to have nice tables

Apa-format is pretty much just like simple with only horizontal lines and no colours

Perhaps instead of an actual feature, a section in our tables vignette touching on that would be great

[strengejacke]

Maybe it's easy to do? We can export both into {gt} and {tinytable}, and I know that tinytable offers theming... pinging @vincentarelbundock

[vincentarelbundock]

well, I'm sure it's possible (maybe even "easy"), but I also have no idea what APA looks like.

[DominiqueMakowski]

[vincentarelbundock]

Here you go:

library(tinytable)

dat <- data.frame(
  Year = c("2014-15", "2015-16", "2016-17", "2017-18", "2018-19", "2019-20"),
  Bachelors = c(9771, 10068, 10187, 10930, 11084, 11825),
  Masters = c(6582, 7582, 7756, 8030, 8349, 8931),
  Doctors = c(3065, 3725, 3893, 4085, 4125, 4524)
)

colnames(dat) <- c(
  "Year",
  "Bachelor's\nDegrees",
  "Master's\nDegrees",
  "Doctor's\nDegrees"
)

n <- nrow(dat)

caption <- "**Table 1** <br> *Degrees in Modern Art in United States*"

note <- "*Note*. table 300 in National Education Statistics; 2018 ed., United States Dept. of Education, Institution of Education Science, National Center for Education Statistics, June 2018, nces.ed.gov/programs/usde/d08/tables.asp."

tab <- tt(dat, caption = caption, notes = note) |>
  format_tt(j = 2:4, digits = 0, num_fmt = "decimal", num_mark_big = ",") |>
  # Center all header cells
  style_tt(i = 0, align = "c") |>
  # First column: left-aligned body cells
  style_tt(j = 1, align = "l") |>
  # Remaining columns: centered body cells
  style_tt(j = 2:4, align = "c") |>
  # APA dividing lines: above header, below header, below last row
  style_tt(i = 0, line = "t", line_width = 0.1) |>
  style_tt(i = 0, line = "b", line_width = 0.1) |>
  style_tt(i = n, line = "b", line_width = 0.1) |>
  format_tt("notes", markdown = TRUE) |>
  format_tt("caption", markdown = TRUE) |>
  format_tt(linebreak = "<br>")

print(tab, "html")

[bwiernik]

Seems like maybe report is good home for this?

[bhattmaulik]

There is a package "apa7". Can it be helpful? Also, I believe that instead of "report" package, the other packages might be more appropriate to host this format. While extracting the model performance or model parameters, can an extra optional argument be added for this? For example, tbl_format = "apa7".

[vincentarelbundock]

I would caution against adding even more complexity to packages like insight, parameters, and performance.

[bwiernik]

Exactly, I think things designed for making publication level outputs belong in report

[kleinlennart]

Hi all, I was looking into whether someone had already solved this, and was happy to find this thread.

Here is a small motivating example. I was imagining a workflow along these lines:

item_scale |>
  correlation::correlation(
    method = "polychoric"
  ) |> 
  report::report_table(format = "apa")

Which could then potentially result in something like this:

\begin{table}[!h]
\centering
\begin{threeparttable}
\caption{Descriptive Statistics and Polychoric Correlations for Example Scale}
\label{tab:example_polychoric_correlations}

\begin{tabular}[t]{lccc>{\centering\arraybackslash}p{0.9cm}>{\centering\arraybackslash}p{0.9cm}>{\centering\arraybackslash}p{0.9cm}>{\centering\arraybackslash}p{0.9cm}>{\centering\arraybackslash}p{0.9cm}}
\toprule
Variable & $n$ & $M$ & $SD$ & 1 & 2 & 3 & 4 & 5\\
\midrule
1. Item 1 & 1,000 & 2.88 & 0.73 & --- &  &  &  & \\
2. Item 2 & 1,000 & 2.71 & 0.79 & $.71^{***}$ & --- &  &  & \\
3. Item 3 & 1,000 & 2.82 & 0.75 & $.69^{***}$ & $.71^{***}$ & --- &  & \\
4. Item 4 & 1,000 & 2.84 & 0.79 & $.67^{***}$ & $.62^{***}$ & $.73^{***}$ & --- & \\
5. Item 5 & 1,000 & 2.91 & 0.77 & $.58^{***}$ & $.56^{***}$ & $.58^{***}$ & $.55^{***}$ & ---\\
\bottomrule
\end{tabular}

\begin{tablenotes}[flushleft]
\item \textit{Note}. Variables refer to responses to five ordinal survey items. Polychoric correlations are reported below the diagonal. Asterisks indicate Holm-adjusted $p$ values. $^{*}p < .05$. $^{**}p < .01$. $^{***}p < .001$.
\end{tablenotes}
\end{threeparttable}
\end{table}

Which would be in line with the APA style for correlation tables

See here: https://apastyle.apa.org/style-grammar-guidelines/tables-figures/sample-tables

Huge fan of easystats, thank you guys! 🙏