Simulate a randomized experiment

Recreates the process of randomly assigning each unit to a treatment group and observing the corresponding potential outcome, given a complete schedule of potential outcomes.

Usage

sim_experiment(schedule, reps = 1, response_name = NULL)

Arguments

schedule

A data frame of class schedule with no missing potential outcomes, as returned by impute_unobserved().

reps

A positive integer giving the number of simulations to run. Results from each simulation are stacked row-wise and identified by the replicate column. Defaults to 1.

response_name

A character string giving the name of the response column in the output. For schedules created with experiment_to_schedule(), this defaults to the original response column name. For schedules created with as_schedule() (wide format), there is no original response column name, so this defaults to "Y".

Value

A data frame with the treatment column, the response column (named according to response_name), and a replicate integer column indicating which simulation each row belongs to.

Examples

data <- data.frame(
  treatment = factor(c("control", "control", "treatment", "treatment", "control", "treatment")),
  response  = c(10, 12, 15, 18, 11, 17)
)
schedule <- experiment_to_schedule(data, treatment, response)
imputed  <- impute_unobserved(schedule, tau = 0)
sim_experiment(imputed, reps = 5)
#> # A tibble: 30 × 3
#>    replicate treatment response
#>        <int> <fct>        <dbl>
#>  1         1 treatment       10
#>  2         1 control         12
#>  3         1 treatment       15
#>  4         1 treatment       18
#>  5         1 control         11
#>  6         1 control         17
#>  7         2 treatment       10
#>  8         2 control         12
#>  9         2 treatment       15
#> 10         2 control         18
#> # ℹ 20 more rows