Crosstable with two variables that displays (only) the mean, standard deviation and samplesize of a third variable in R

Question

i'm searching for an easy way to create a multivariat crosstable with two variables that only displays the mean, standard deviation and frequencies/samplesize of a third variable. In the best case with column sums and row sums.

Example with the mtcars dataset (table with "cyl" and "vs"; mean, standard deviation and frequencies of "hp"): In Stata it would be tabulate cyl vs, summarize(hp) freq mean sta .

To make clear how it ideally should look like I've made an example (the results are just made up):

or

Example 2 (results also just made up)

(Finally I have to transfer the table to LaTeX - ideally with stargazer - so it would be best of all if this were possible.)

I found a lot of solutions for crosstables with means, but only for two variables. And certainly not with mean, std. dev. and frequencies in one table. You would be a huge help to me.

edit: Now I tried this, but I don't know how to add standard deviation, column sums and row sums.

    library(tidyr)

ct <- mtcars %>%
  group_by(cyl, vs) %>%
  summarise(hp = mean(hp, na.rm=TRUE), .groups = "drop") %>%
  spread(vs, hp)
ct

ct %>%
  adorn_rounding() %>%
  adorn_ns(
    ns = mtcars %>% # calculate the Ns on the fly by calling tabyl on the original data
      tabyl(cyl, vs)
  ) %>%
  adorn_title("combined", row_name = "Cylinders", col_name = "Is Automatic")

# stargazer(ct)

And I get this (means by group and frequencies in brackets):

  Cylinders/Is Automatic          0          1
1                      4    91  (1)  81.8 (10)
2                      6 131.7  (3) 115.2  (4)
3                      8 209.2 (14)    NA  (0)

edit2:

library(tidyr)
library(arsenal)
tab1 <- tableby(vs ~ hp, data=mtcars, strata = cyl, numeric.stats = c("meansd", "N"), test = FALSE)
summary(tab1, text = TRUE)

This sort of works for me, but to get the exact table - as in my illustration - into LaTeX, I unfortunately have to do a lot of work by hand.

Unfortunately stargazer(tab1) does not work at all, because of

% Error: Unrecognized object type.

.

summary(tab1, text = "latex") does work, but as mentioned to get a more or less beautiful, publication quality table in LaTeX there's a lot of work by hand. Maybe it's too much to ask, but does anyone have any ideas?

Answer 1

Several packages in R could be helpful. The one I have used is arsenal which may be close to Stata tabulate, summarize

library(tidyr)
library(arsenal)
tab1 <- tableby(vs ~ hp, data=mtcars, strata = cyl,
                numeric.stats = c("meansd", "N"), test = FALSE)
summary(tab1, text = TRUE)
#> 
#> 
#> |cyl |             |     0 (N=18)     |    1 (N=14)     |   Total (N=32)   |
#> |:---|:------------|:----------------:|:---------------:|:----------------:|
#> |4   |hp           |                  |                 |                  |
#> |    |-  Mean (SD) |   91.000 (NA)    | 81.800 (21.872) | 82.636 (20.935)  |
#> |    |-  N         |        1         |       10        |        11        |
#> |6   |hp           |                  |                 |                  |
#> |    |-  Mean (SD) | 131.667 (37.528) | 115.250 (9.179) | 122.286 (24.260) |
#> |    |-  N         |        3         |        4        |        7         |
#> |8   |hp           |                  |                 |                  |
#> |    |-  Mean (SD) | 209.214 (50.977) |       NA        | 209.214 (50.977) |
#> |    |-  N         |        14        |        0        |        14        |

^{Created on 2021-03-15 by the reprex package (v1.0.0)}