@@ -984,6 +984,35 @@ directly wrapped for use with `graph_test_shortcut_gsd()`. However,
984984` gsDesign ` and ` rpact ` implement the same standard spending function
985985formulas, so ` gsDesign ` wrappers can be used to achieve equivalent results.
986986
987+ However, it would be possible to use ` rpact ` if the ` graphicalMCP ` package functions interface
988+ differently with ` rpact ` compared to ` gsDesign ` . In particular, this code snippet shows how to obtain the
989+ repeated p-values from ` rpact ` shown in the above repeated p-values example:
990+
991+ ``` {r rpact-snippet}
992+ rpact::setLogLevel("DISABLED")
993+ repP <- function(pVals){
994+ cum_n <- seq_along(pVals) * 2
995+ design <- rpact::getDesignGroupSequential(typeOfDesign = "asOF", kMax = 3)
996+ data <- rpact::getDataset(
997+ cumMeans = c(qnorm(1 - pVals) / sqrt(cum_n)),
998+ cumStDevs = rep(1, length(pVals)),
999+ cumN = cum_n
1000+ )
1001+ stage_res <- rpact::getStageResults(design, data, normalApproximation = TRUE)
1002+ rpact::getRepeatedPValues(stage_res)
1003+ }
1004+
1005+ repP(c(0.0062, 0.0002))
1006+ repP(c(0.0170, 0.0035))
1007+ repP(c(0.0090, 0.0020))
1008+ repP(c(0.1300, 0.0600))
1009+ ```
1010+
1011+ It would be more beneficial to use the built-in ` rpact ` integration routines to obtain these
1012+ repeated p-values, because this would provide an alternative computation method.
1013+ Merely supplying another implementation of the same simple spending function formulas
1014+ would not provide a meaningful alternative.
1015+
9871016## Summary
9881017
9891018The ` graph_test_shortcut_gsd() ` function performs multiple testing in group
0 commit comments