Admissible back-door adjustment sets, smallest first

The empty set is checked too: if it is admissible there is no confounding to adjust for, and adjusting anyway costs precision.

Usage

morie_backdoor_sets(adj, treatment, outcome, max_size = NULL)

Arguments

adj

Logical adjacency matrix.

treatment, outcome

Node indices (1-based).

max_size

Optional cap on set size.

Value

A list of integer vectors.

Examples

A <- matrix(FALSE, 3, 3)
morie_backdoor_sets(A, 2, 3)
#> [[1]]
#> integer(0)
#> 
#> [[2]]
#> [1] 1
#>