Should we change the ordering of the grouping variables?

[ddsjoberg]

In the example below, the group1 is the outermost variable, but really should be next to variable.

But in the hierarchical ARD, we'd need to figure out the ordering....

library(cards)

ard_strata(
  ADSL,
  .by = AGEGR1,
  .f = ~ cardx::ard_stats_kruskal_test(.x, by = ARM, variables = AGE)
)
#> {cards} data frame: 12 x 11
#>    group1 group2 group2_level variable stat_name stat_label      stat
#> 1     ARM AGEGR1          <65      AGE statistic  Kruskal-…     7.381
#> 2     ARM AGEGR1          <65      AGE   p.value    p-value     0.025
#> 3     ARM AGEGR1          <65      AGE parameter  Degrees …         2
#> 4     ARM AGEGR1          <65      AGE    method     method Kruskal-…
#> 5     ARM AGEGR1          >80      AGE statistic  Kruskal-…     0.938
#> 6     ARM AGEGR1          >80      AGE   p.value    p-value     0.626
#> 7     ARM AGEGR1          >80      AGE parameter  Degrees …         2
#> 8     ARM AGEGR1          >80      AGE    method     method Kruskal-…
#> 9     ARM AGEGR1        65-80      AGE statistic  Kruskal-…     1.146
#> 10    ARM AGEGR1        65-80      AGE   p.value    p-value     0.564
#> 11    ARM AGEGR1        65-80      AGE parameter  Degrees …         2
#> 12    ARM AGEGR1        65-80      AGE    method     method Kruskal-…
#> ℹ 4 more variables: context, fmt_fn, warning, error

^{Created on 2024-12-18 with reprex v2.1.1}