How to use R package modelsummary to produce a factanal summary table

Question

The modelsummary::supported_models() includes factanal objects, to purportedly produce a summary table of a factor analysis model.

However, the following two attempts fail

library(modelsummary)
library(tidyverse)

iris %>% 
  select(-Species) %>% 
  factanal(factors = 1) %>% 
  modelsummary()

iris %>% 
  select(-Species) %>% 
  factanal(factors = 1) %>% 
  tidy %>% 
  modelsummary()

How do I produce a summary table of a factor analysis fit (ie eventually making a LaTeX object)?

Answer 1

This one is kind of tricky because broom::tidy() does not return a data frame with a term column. So you have two basic options.

library(modelsummary)
library(tinytable)
library(broom)

mod <- factanal(mtcars, factors = 2)

Option 1: modelsummary magic

Define a custom class and extractor for your model:

class(mod) <- c("custom", class(mod))

tidy.custom <- function(x, ...) {
  out <- broom:::tidy.factanal(x, ...)
  colnames(out)[colnames(out) == "variable"] = "term"
  return(out)
}

modelsummary(mod,
  estimate = "uniqueness",
  output = "tinytable",
  statistic = c("fl1", "fl2"),
  shape = term ~ statistic) |> print("markdown")

# +----------------+------------------+--------+--------+
# |                | (1)                                |
# +----------------+------------------+--------+--------+
# |                | Est.             | fl1    | fl2    |
# +================+==================+========+========+
# | mpg            | 0.167            | 0.686  | -0.602 |
# +----------------+------------------+--------+--------+
# | cyl            | 0.070            | -0.629 | 0.731  |
# +----------------+------------------+--------+--------+
# | disp           | 0.096            | -0.730 | 0.609  |
# +----------------+------------------+--------+--------+
# | hp             | 0.143            | -0.337 | 0.862  |
# +----------------+------------------+--------+--------+
# | drat           | 0.298            | 0.807  | -0.225 |
# +----------------+------------------+--------+--------+
# | wt             | 0.168            | -0.810 | 0.420  |
# +----------------+------------------+--------+--------+
# | qsec           | 0.150            | -0.162 | -0.908 |
# +----------------+------------------+--------+--------+
# | vs             | 0.256            | 0.291  | -0.812 |
# +----------------+------------------+--------+--------+
# | am             | 0.171            | 0.907  | 0.081  |
# +----------------+------------------+--------+--------+
# | gear           | 0.246            | 0.860  | 0.125  |
# +----------------+------------------+--------+--------+
# | carb           | 0.386            | 0.031  | 0.783  |
# +----------------+------------------+--------+--------+
# | Num.Obs.       | 32               |        |        |
# +----------------+------------------+--------+--------+
# | n.factors      | 2                |        |        |
# +----------------+------------------+--------+--------+
# | total.variance | 0.80459022172898 |        |        |
# +----------------+------------------+--------+--------+
# | n              | 32               |        |        |
# +----------------+------------------+--------+--------+
# | method         | mle              |        |        |
# +----------------+------------------+--------+--------+

Option 2: broom + tinytable

broom can extract info from the model as a data.frame, then use tinytable to create a table:

s <- tidy(mod)
tt(s, digits = 2)

term	uniqueness	fl1	fl2
mpg	0.167	0.686	-0.602
cyl	0.07	-0.629	0.731
disp	0.096	-0.73	0.609
hp	0.143	-0.337	0.862
drat	0.298	0.807	-0.225
wt	0.168	-0.81	0.42
qsec	0.15	-0.162	-0.908
vs	0.256	0.291	-0.812
am	0.171	0.907	0.081
gear	0.246	0.86	0.125
carb	0.386	0.031	0.783