Add freq_xtab() for contingency count matrices and grouped arrays, with per-matrix margins, stable category axes, explicit missing values, and globally configurable factor-level retention.
Breaking API: freq_table() accepts one outcome with groups supplied through dplyr::group_by(); legacy two-variable calls now error with migration guidance. Proportions and compact columns replace verbose percentage output by default.
Export freq_tbl(), wald_ci(), and logit_ci(), with consistent grouped and ungrouped schemas, named options, explicit missing-data handling, and optional percentages, overall results, generic headers, standard errors, and t values.
Preserve the existing t-based interval convention; within-group standard errors and critical values now both use n_group - 1. Logit bounds at zero/one and intervals with denominators <= 1 are explicitly missing.
Keep freq_test() and freq_format() compatible with compact/generic results. Construct Fisher matrices by category identities for rectangular tables, validate complete test inputs, and reject unsupported extra arguments.
Fix get_group_n() to work without attaching dplyr; preserve formatting prefixes and safely format empty tables and character categories.
Add six comprehensive vignettes, a function/argument coverage inventory, migration guidance, and independent numerical/integration tests.
Require R >= 4.1.0 and dplyr >= 1.0.0 for the documented pipeline syntax. This is an unpublished development version, not a CRAN release.
Add built-in example study data to freqtables