NEWS

surveybootstrap 0.2.1

Fixed a warning ("Unknown or uninitialised column: weight.scale") that appeared when using bootstrap.estimates() with rescaled.bootstrap.sample(): the index column was being extracted as a one-column tibble instead of a plain vector, causing tibble's $ accessor to misreport the column as missing
Fixed a deprecation warning from rlang ("Unquoting language objects with !!! is deprecated"): replaced !!!psu.vars with across(all_of(...)) / all_of(...) throughout rescaled.bootstrap.sample() and get.rescaled.bootstrap.weights()
Fixed rescaled.bootstrap.weights() producing column names boot_weight_rep.1, boot_weight_rep.2, … instead of the documented boot_weight_1, boot_weight_2, …
Added vignette "The rescaled bootstrap" illustrating the main workflow with the MU284 example data
Added pkgdown site
rescaled.bootstrap.sample() and get.rescaled.bootstrap.weights() now issue a warning when any stratum contains fewer than 2 PSUs (variance cannot be estimated in that case; bootstrap weights for the stratum will be 0)
Added regression tests covering the above bug fixes and the single-PSU warning

Added get.rescaled.bootstrap.weights(): new function that returns rescaled bootstrap weights (and optionally cluster counts and scaling factors) in a wide data frame, making it straightforward to use the rescaled bootstrap for jackknife-after-bootstrap (JAB) variance estimation
rescaled.bootstrap.sample() gains an include_cc argument; when TRUE it returns per-rep cluster inclusion counts alongside the weight-scaling factors, which are needed for JAB calculations
Fixed a bug where bootstrap.estimates() failed silently when used with rescaled.bootstrap.sample(): the weight-scaling column was named weight_scale (underscore) in the sampler but read as weight.scale (dot) by the estimator, causing weights to collapse to numeric(0) and all estimates to be NaN / Inf
Expanded test suite: added structural and semantic tests for rescaled.bootstrap.sample(), srs.bootstrap.sample(), get.rescaled.bootstrap.weights(), and rescaled.bootstrap.weights(); added a regression test that directly catches the weight.scale naming bug; added tests for the weighted.mean() helper

fixed some dependency errors that suddenly arose (explicitly use plyr:: in a few places)
dropped now-deprecated group_indices_() in favor of cur_group_index()
changed docs to use Markdown roxygen
added MU284 dataset for examples and tests
added examples to exported function docs

moved unit tests for goc and helper functions in from networkreporting
added rescaled.bootstrap.weights helper function for case when a tibble with all of the rescaled bootstrap weights is useful
added dependencies to tidyverse, magrittr, and purrr