Closed
Description
The following code generates a symmetric design.
> (gs_design_ahr(analysis_time = 36, info_frac = 1:3/3,
+ upper = gs_spending_bound,
+ upar = list(sf = sfLDOF, total_spend = 0.025),
+ lower = gs_spending_bound,
+ lpar = list(sf = sfLDOF, total_spend = 0.025),
+ binding = TRUE, h1_spending = FALSE))$bound
# A tibble: 6 × 7
analysis bound probability probability0 z `~hr at bound` `nominal p`
<int> <chr> <dbl> <dbl> <dbl> <dbl> <dbl>
1 1 upper 0.00450 0.000104 3.71 0.476 0.000104
2 1 lower 0.00000100 0.000104 -3.71 2.10 1.00
3 2 upper 0.382 0.00605 2.51 0.701 0.00601
4 2 lower 0.00000213 0.00605 -2.51 1.43 0.994
5 3 upper 0.900 0.0250 1.99 0.795 0.0231
6 3 lower 0.00000218 0.0250 -1.99 1.26 0.977
After applying to_integer() to the above design, the boundaries are not symmetrical anymore.
> (gs_design_ahr(analysis_time = 36, info_frac = 1:3/3,
+ upper = gs_spending_bound,
+ upar = list(sf = sfLDOF, total_spend = 0.025),
+ lower = gs_spending_bound,
+ lpar = list(sf = sfLDOF, total_spend = 0.025),
+ binding = TRUE, h1_spending = FALSE) %>%
+ to_integer())$bound
# A tibble: 6 × 7
analysis bound probability probability0 z `~hr at bound` `nominal p`
<int> <chr> <dbl> <dbl> <dbl> <dbl> <dbl>
1 1 upper 0.00439 0.000101 3.72 0.473 0.000101
2 1 lower 0.000109 0.00405 -2.65 1.70 0.996
3 2 upper 0.378 0.00596 2.52 0.698 0.00593
4 2 lower 0.00597 0.378 -0.311 1.05 0.622
5 3 upper 0.901 0.0250 1.99 0.793 0.0232
6 3 lower 0.0250 0.900 1.28 0.861 0.0998
Thanks @yihui, for discovering this issue during the shiny app development.