Manski worst-case bounds (full interface)

Alias of morie_bndest exposing both the partially-observed-mean and the average-treatment-effect routes. Mirrors Python morie.fn.manskif.

Usage

morie_manskif(y, observed, support, treatment = NULL)

Arguments

y

Outcomes.

observed

Logical vector of observed outcomes; ignored (and may be NULL) when treatment is supplied.

support

Length-2 vector c(K0, K1) with K0 < K1, the assumed outcome support.

treatment

Optional binary 0/1 treatment.

Value

The list returned by morie_bndest.

References

Manski, C. F. (2007). Identification for Prediction and Decision. Harvard University Press.

Examples

set.seed(12)
y <- runif(60); D <- rbinom(60, 1, 0.5)
morie_manskif(y, observed = NULL, support = c(0, 1), treatment = D)
#> $ate_lower
#> [1] -0.5095442
#> 
#> $ate_upper
#> [1] 0.4904558
#> 
#> $ate_width
#> [1] 1
#> 
#> $y1_bounds
#> [1] 0.2254570 0.7421237
#> 
#> $y0_bounds
#> [1] 0.2516679 0.7350012
#> 
#> $p_treated
#> [1] 0.4833333
#> 
#> $contains_zero
#> [1] TRUE
#> 
#> $width_identity
#> [1] "the ATE bounds always have width exactly K1 - K0, so they always contain zero: no-assumption bounds never sign an effect on their own"
#> 
#> $identified
#> [1] FALSE
#> 
#> $n
#> [1] 60
#> 
#> $method
#> [1] "Manski (1990) worst-case bounds on the average treatment effect"
#>