I'm trying to automate the production of .tex tables and I want to add in the add.lines section of stargazer a value called mean_outcome but I'm having trouble producing this value, I think I'm dealing wrong with tidy eval. I have tried quo, enquo, !!, `` but nothing is working.
df <- tibble::tribble(~hhid, ~treatment_arm, ~round_pooled, ~tot_lstock_count, ~tot_animal_tlu,
"1020201023",1,1,10,1,
"1020201023",1,11,6,0.600000023841858,
"1020201036",1,1,10,0.100000001490116,
"1020201036",1,11,4,0.400000005960464,
"1020201039",1,1,9,0.0900000035762787,
"1020201057",1,1,12,0.480000019073486,
"1020201068",1,1,2,0.200000002980232,
"1020201095",1,1,6,0.330000013113022,
"1020201095",1,11,7,0.400000005960464,
"1020201103",1,1,2,0.200000002980232,
"1020201116",1,1,7,0.0599999986588955,
"1020201116",1,11,2,0.0199999995529652,
"1020201139",1,11,12,0.480000019073486,
"1020201144",1,1,5,0.5,
"1020201144",1,11,8,0.800000011920929,
"1020201146",1,1,6,0.0599999986588955,
"1020201146",1,11,6,0.0599999986588955,
"1020201159",1,11,15,1.5,
"1020201180",1,11,11,0.560000002384186,
"1020201205",1,1,2,0.0199999995529652)
reg_coef_tables <- function(df, y) {
outcome = eval(`i`)
mean_outcome <- df %>%
filter(round_pooled == 11 & treatment_arm == 1) %>%
summarise(mean = mean(!!outcome, na.rm = T)) %>%
pull(mean)
reg1 <- felm(outcome ~ treatment_arm, data = df)
stargazer(reg1,
align = TRUE,
dep.var.labels = c("(1)"),
omit.stat = c("f", "adj.rsq", "ser"),
no.space = TRUE,
digits = 3, # number of decimals
add.lines = list(c('Mean outcome', rep(`mean_outcome`, 1)),
dep.var.caption = "", # remove dep var header
out = paste0(outcome, ".tex"))
}
liv_vars = c("tot_lstock_count", "tot_animal_tlu")
liv_out_list = vector("list", 2)
for(i in liv_vars) {
liv_out_list[[i]] <- reg_coef_tables(df, i)
}
Leaving aside stargazer (because, IMHO, it's horrible) and felm (because I don't know which package it's from) and because neither are relevant to your problem with tidyverse, try this
reg_coef_tables <- function(df, y) {
outcome = as.name(y)
mean_outcome <- df %>%
filter(round_pooled == 11 & treatment_arm == 1) %>%
summarise(mean = mean(!!outcome, na.rm = T)) %>%
pull(mean)
}
liv_vars = c("tot_lstock_count", "tot_animal_tlu")
liv_out_list = list()
for(i in liv_vars) {
liv_out_list[[i]] <- reg_coef_tables(df, i)
}
liv_out_list
$tot_lstock_count
[1] 7.888889
$tot_animal_tlu
[1] 0.5355556
Or, perhaps more succinctly:
mean_outcome <- df %>%
filter(round_pooled == 11 & treatment_arm == 1) %>%
summarise(
across(
c(tot_lstock_count, tot_animal_tlu),
\(x) mean(x, na.rm = TRUE)
)
)
mean_outcome
# A tibble: 1 × 2
tot_lstock_count tot_animal_tlu
<dbl> <dbl>
1 7.89 0.536
You can find a useful primer on NSE here.