Compute minimum test statistic

Calculates the minimum value of the rank sum statistic under the null hypothesis, assuming at most n-k treated units can have effects greater than c.

Usage

min_stat(Z, Y, k, c, method.list = NULL, score = NULL, ind.sort.treat = NULL)

Arguments

Z

An n-dimensional binary treatment assignment vector.

Y

An n-dimensional observed outcome vector.

k

Quantile index (between 1 and n).

c

Threshold for the null hypothesis.

method.list

A list specifying the scoring method.

score

Optional pre-computed score vector.

ind.sort.treat

Optional pre-computed sorted treatment indices.

Value

The minimum test statistic value.

References

Caughey, D., Dafoe, A., Li, X., & Miratrix, L. (2023).

Examples

Z <- c(0, 0, 0, 1, 1, 1)
Y <- c(1, 2, 3, 10, 20, 30)
score <- rank_score(6, list(name = "Wilcoxon", scale = FALSE))
# k = 6 exempts no treated unit, so this is the treated rank sum: 4 + 5 + 6
min_stat(Z, Y, k = 6, c = 0, score = score)
#> [1] 15