Closed lucasmation closed 5 years ago
I would tend to recommend a joint test (like the omnibus ri f-test in the examples, see ?conduct_ra
) over a one-by-one balance table, since all those tests are correlated with each other. But that's not exactly your question!
how about something like this:
library(tidyverse)
library(ri2)
N <- 10
declaration <- declare_ra(N = N, m = 5)
dat <- tibble(X1 = rnorm(N),
X2 = rnorm(N),
X3 = rnorm(N),
Z = conduct_ra(declaration))
covs <- c("X1", "X2", "X3")
balance <-
covs %>%
map( ~ paste0(., " ~ Z")) %>%
map( ~ conduct_ri(
formula = .,
assignment = "Z",
declaration = declaration,
data = dat
)) %>%
map_df(tidy) %>%
mutate(covariate = covs)
balance
@acoppock , nice package, tks!
Do have any simple pipeline to produce a Balancing table based on conduct_ri for a (potentially long) set of covariates?