Should the ard_*() functions add a strata argument?

[ddsjoberg]

Some ard_*() functions have a by= argument. All stats are calculated for all combinations of variables found in the by= argument (observed and unobserved combinations).

Perhaps all these functions should have a strata= argument as well? The data would first be stratified by the strata= variable(s), then the calculations would occur within each stratum. A few use cases for this:

1. Tabulating AE within SOC.
2. Summarizing items stratified by visit (perhaps not all items collected at all timepoints)

We already have the ard_stratieif_categorical(), but it could perhaps be encompassed in ard_categorical(strata=)

[ddsjoberg]

Here is how the new argument could be documented. I like it together so users will read both descriptions.