.by argument for the group_*() functions

[tvqt]

I was reading through this StackOverflow question from a few years ago, and learned of the group_map function. It seems like it might be the ugly duckling of the dplyr function family.

It would be neat if the group_*() functions (group_map(), group_modify() and group_walk()) included the .by syntax similar to the other dplyr functions!

For example, in the case of:

library(estimatr)

mtcars %>%
  group_by(am, vs) %>%
  group_map(~tidy(lm_robust(mpg ~ hp, data = .x))) %>%
  ungroup() # question asker forgot to include this (!)

It could instead be:

library(estimatr)

mtcars |> group_map(~tidy(lm_robust(mpg ~ hp, data = .x)), .by = c(am, vs))

Keep up the great work! :-)

[DavisVaughan]

These are still a bit experimental as they don't quite feel exactly right, so we probably won't add the .by argument to them anytime soon

A lot of times you can accomplish the same thing with reframe(), i.e.

library(estimatr)
library(dplyr)

mtcars |> 
  as_tibble() |>
  reframe(
    tidy(lm_robust(mpg ~ hp, data = pick(mpg, hp))), 
    .by = c(am, vs)
  )
#> # A tibble: 8 × 11
#>      am    vs term       estimate std.error statistic p.value conf.low conf.high
#>   <dbl> <dbl> <chr>         <dbl>     <dbl>     <dbl>   <dbl>    <dbl>     <dbl>
#> 1     1     0 (Intercep…  26.4      2.14        12.3  2.49e-4  20.5     32.4    
#> 2     1     0 hp          -0.0368   0.00909     -4.05 1.54e-2  -0.0621  -0.0116 
#> 3     1     1 (Intercep…  37.3      6.09         6.13 1.67e-3  21.7     53.0    
#> 4     1     1 hp          -0.111    0.0866      -1.28 2.55e-1  -0.334    0.111  
#> 5     0     1 (Intercep…  30.9      1.57        19.7  6.18e-6  26.9     34.9    
#> 6     0     1 hp          -0.0994   0.0157      -6.33 1.45e-3  -0.140   -0.0590 
#> 7     0     0 (Intercep…  23.2      3.27         7.10 3.28e-5  15.9     30.5    
#> 8     0     0 hp          -0.0422   0.0164      -2.56 2.82e-2  -0.0788  -0.00552
#> # ℹ 2 more variables: df <dbl>, outcome <chr>

I'm going to close this one because it is something we are passively thinking about, but not anything we plan to change in the medium term