Stabilized inverse-probability-of-censoring weights

The same product form as the treatment weight, with the probability of remaining uncensored in place of the treatment probability. Uncensored subjects stand in for those who were censored.

Usage

Stbciw(C, H = NULL, numerator = NULL)

Arguments

C

Conditional probabilities of remaining uncensored, subjects by time.

H

Accepted for signature compatibility; shape check only.

numerator

Marginal \(P(C_t=0|C_{t-1}=0)\); omitted gives unstabilized weights.

Value

Named list with estimate, weights, unstabilized, mean_weight, max_weight, n, n_times, method.

References

Robins (1993) Proc Biopharm Sect ASA 24-33.

Examples

Stbciw(matrix(c(0.9, 0.8, 0.95, 0.85), nrow = 2))
#> $estimate
#> [1] 1.320089
#> 
#> $weights
#> [1] 1.169591 1.470588
#> 
#> $unstabilized
#> [1] 1.169591 1.470588
#> 
#> $mean_weight
#> [1] 1.320089
#> 
#> $max_weight
#> [1] 1.470588
#> 
#> $n
#> [1] 2
#> 
#> $n_times
#> [1] 2
#> 
#> $method
#> [1] "stabilized censoring weights (Robins 1993)"
#>