Add NEWS.md and cran-comments.md; use seq_len() in loops

- NEWS.md: document the development-version changes (svydesign_build, MOVER
  RERI CIs, the interaction Scale column, svykmplot se=, the getting-started
  vignette, the svygof/svyAUC/svycoxph_CE_mi fixes, self-contained examples,
  and the new test suite).
- cran-comments.md: map each of the three CRAN reviewer points to its fix and
  record the check status, for the eventual resubmission.
- Replace 1:nrow()/1:n loop counters with seq_len() in jointeffects(),
  svycoxph_CE(), and svytable1() to avoid the latent 1:0 zero-length bug.

R CMD check: 0 errors, 0 warnings; remaining NOTEs are environmental (the
"0.14.1.9000" development-version note and a transient "unable to verify current
time"). 86 tests pass.

Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>

Author: ehsanx

NEWS.md

@@ -0,0 +1,41 @@
+# svyTable1 0.14.1.9000
+
+## New features
+
+* `svydesign_build()` builds a complex-survey design from plain column names with
+  input validation and clear error messages, and supports subpopulation
+  ("domain") analysis the correct way -- by subsetting the design rather than
+  pre-filtering rows, which would break subgroup standard errors.
+* `addint()` and `addintlist()` gain `ci_method = c("delta", "mover")`. The new
+  `"mover"` option uses the MOVER interval of Zou (2008) <doi:10.1093/aje/kwn104>
+  for RERI, which generally has better coverage than the symmetric Wald
+  interval. The default (`"delta"`) is unchanged.
+* `addintlist()` now returns a `Scale` column ("OR (logistic)", "HR (Cox)",
+  "RR (log-binomial)", "RR (Poisson)", or "ratio") that makes the effect-measure
+  scale of RERI/AP/S explicit.
+* `svykmplot()` gains an `se` argument. `se = FALSE` draws the survival curves
+  without the (slow) survey-weighted confidence bands.
+* New "Getting Started" vignette giving an end-to-end SPPH 604 workflow, and a
+  package-level help page (`?svyTable1`) indexing the functions by purpose.
+
+## Improvements and bug fixes
+
+* `svygof()` now attaches the decile groups to the existing design via
+  `update()` instead of rebuilding a bare `svydesign`, so it preserves finite
+  population corrections and calibration and works on replicate-weights designs
+  (it previously errored on them).
+* `svyAUC()` builds its confidence interval on the logit scale (so it stays
+  within `[0, 1]`) and uses the design degrees of freedom instead of a fixed
+  critical value; removed redundant internal computation.
+* `svycoxph_CE_mi()` default `time_var`/`status_var` corrected to
+  `"stime"`/`"status"`.
+* `svytable1()` now errors on a zero-row design instead of returning a
+  placeholder table, and no longer over-suppresses cells whose design effect is
+  below 1.
+* Documentation: method references with DOIs added to `DESCRIPTION`; every
+  exported function's `\value` now names its output class and meaning; the
+  additive-interaction help notes that RERI/AP/S approximate risk-ratio measures
+  only when the outcome is rare.
+* Examples and vignettes are now self-contained on the bundled `nhanes_mortality`
+  data and no longer require the `NHANES` package.
+* Added a real test suite (replacing placeholder smoke tests).

R/jointeffects.R

@@ -133,7 +133,7 @@ jointeffects <- function(interaction_model,
   results <- list()
 
   # --- 4. Loop through each combination ---
-  for (i in 1:nrow(all_combinations)) {
+  for (i in seq_len(nrow(all_combinations))) {
     lvl1 <- all_combinations$Level1[i]
     lvl2 <- all_combinations$Level2[i]
 

R/svycoxph_CE.R

@@ -156,7 +156,7 @@ svycoxph_CE <- function(formula_rhs,
     if(length(ids) == nrow(data_full)) ids <- ids[data_full[[time_var]] > 0]
     data_clean$internal_ids_auto <- ids
   } else {
-    data_clean$internal_ids_auto <- 1:nrow(data_clean)
+    data_clean$internal_ids_auto <- seq_len(nrow(data_clean))
   }
 
   if(!is.null(design$strata) && length(design$strata) > 0) {
@@ -209,7 +209,7 @@ svycoxph_CE <- function(formula_rhs,
   time_intervals <- c(0, cuts, max(split_data[[time_var]]))
   loop_formula <- as.formula(paste0("Surv(tstart, ", time_var, ", ", status_var, ") ~ ", formula_rhs))
 
-  for (i in 1:n_intervals) {
+  for (i in seq_len(n_intervals)) {
     fit <- tryCatch({
       # Subset the design for the specific time group
       d_sub <- suppressMessages(

R/svytable1.R

@@ -306,7 +306,7 @@ svytable1 <- function(design, strata_var, table_vars,
         means_by_strata_w <- svyby(var_formula, strata_formula, design, svymean, na.rm = TRUE)
         vars_by_strata_w <- svyby(var_formula, strata_formula, design, svyvar, na.rm = TRUE)
 
-        for(i in 1:nrow(means_by_strata_w)){
+        for(i in seq_len(nrow(means_by_strata_w))){
           s_lvl <- as.character(means_by_strata_w[i, 1])
           mean_val <- means_by_strata_w[i, 2]
           sd_val <- sqrt(vars_by_strata_w[i, 2])
@@ -353,7 +353,7 @@ svytable1 <- function(design, strata_var, table_vars,
 
         missing_by_strata <- svyby(~is_missing, strata_formula, design_temp, svymean, na.rm = TRUE)
 
-        for (i in 1:nrow(missing_by_strata)) {
+        for (i in seq_len(nrow(missing_by_strata)) ) {
           s_lvl <- as.character(missing_by_strata[i, 1])
           uw_n_strata_missing <- sum(is.na(current_var_vector[design$variables[[strata_var]] == s_lvl]))
           missing_row[[s_lvl]] <- sprintf("%s (%.1f%%)", format_num(uw_n_strata_missing, FALSE), missing_by_strata[i, 2] * 100)

cran-comments.md

@@ -0,0 +1,37 @@
+## Resubmission
+
+This is a resubmission. In response to the previous review, the three requested
+changes have been made:
+
+1. **References in the `Description` field.** Method references are now given in
+   the `Description` field in the requested form, e.g. Lumley (2004)
+   <doi:10.18637/jss.v009.i08>, Archer and Lemeshow (2006)
+   <doi:10.1177/1536867X0600600106>, van Buuren and Groothuis-Oudshoorn (2011)
+   <doi:10.18637/jss.v045.i03>, with no space after `doi:` and angle brackets
+   for auto-linking.
+
+2. **`\value` tags.** `pipe.Rd` now has a `\value` entry, and every exported
+   function's `\value` describes the structure (class) and meaning of the
+   returned object.
+
+3. **`\dontrun{}`.** All previously `\dontrun{}` examples have been converted to
+   runnable examples or `\donttest{}` and rewritten onto the bundled
+   `nhanes_mortality` dataset so they execute without the suggested `NHANES`
+   package.
+
+## Test environments
+
+* Local: Windows 11, R 4.5.1
+* (Before submission, also run: win-builder devel and release; R-hub.)
+
+## R CMD check results
+
+`R CMD check --as-cran --run-donttest`: 0 errors | 0 warnings | 1 note.
+
+The single NOTE is the expected "New submission" note. (During development the
+version carries a `.9000` development suffix; set a normal release version,
+e.g. `0.15.0`, before submitting.)
+
+All examples and vignettes run on the bundled `nhanes_mortality` dataset and do
+not require any Suggests package, so the package checks cleanly with the
+Suggests packages absent as well as present.