Skip to contents

Declare a variable of the data-generating process

Source: R/var.R
pa_var.Rd

Describes the marginal distribution of one regressor (the focal variable or a covariate) for ape_dgp(). Dependence between variables is set via the correlation argument of ape_dgp() (a Gaussian copula).

Usage

pa_var(
  name,
  type = c("binary", "normal"),
  p = NULL,
  mean = 0,
  sd = 1,
  icc = NULL
)

Arguments

name

Variable name (a string; used in printing and error messages).

type

"binary" (Bernoulli) or "normal".

p

Success probability, strictly between 0 and 1 (binary only).

mean, sd

Mean and standard deviation (normal only; sd > 0).

icc

Panel designs only (ape_dgp_panel()): within-unit correlation of the variable over time, via a shared-component device on the latent normal scale (u_it = sqrt(icc) v_i + sqrt(1-icc) e_it) before the marginal transform. icc = 0 redraws each period; icc = 1 makes the variable time-constant (e.g., a unit-level treatment). For binary variables the realized outcome-scale correlation is attenuated relative to icc, as with the Gaussian-copula correlation. Ignored by the cross-sectional constructors.

Value

An object of class pa_var.

Examples

pa_var("treat", "binary", p = 0.5)
#> $name
#> [1] "treat"
#> 
#> $type
#> [1] "binary"
#> 
#> $p
#> [1] 0.5
#> 
#> $mean
#> [1] 0
#> 
#> $sd
#> [1] 1
#> 
#> $icc
#> NULL
#> 
#> attr(,"class")
#> [1] "pa_var"
pa_var("age", "normal", mean = 45, sd = 12)
#> $name
#> [1] "age"
#> 
#> $type
#> [1] "normal"
#> 
#> $p
#> NULL
#> 
#> $mean
#> [1] 45
#> 
#> $sd
#> [1] 12
#> 
#> $icc
#> NULL
#> 
#> attr(,"class")
#> [1] "pa_var"