Nine new functions extend the analysis and planning workflow:
check_gstudy_design() diagnoses invalid scores, missing
IDs, duplicate cells, missing crossed cells, insufficient levels and
simple nested imbalance.error_budget() shows each component’s contribution to
D-study error.sem_gtheory() reports relative and absolute standard
errors of measurement.score_interval() constructs approximate normal
measurement intervals.dstudy_grid() compares candidate facet counts in a
single table.optimize_dstudy() finds minimum-cost candidates meeting
a G or Phi target.dstudy_sensitivity() compares gains from increasing
each facet separately.simulate_gstudy() generates balanced Gaussian crossed
data.bootstrap_gstudy() estimates Gaussian parametric
percentile intervals for crossed-design variance components and D-study
quantities.New planning functions reject negative variance estimates unless the caller explicitly requests truncation, and record adjusted components. Existing D-study functions retain their original raw-estimate behavior.
Fixed numeric ID indexing in persons-by-items and nested estimators, handled unused factor levels, added validation to all G-study entry points, and rejected ambiguous facet labels. ANOVA extraction now uses the aov summary rather than requesting F-tests from a saturated model. No new package dependencies.
See inst/doc/planning-guide.md for assumptions, sources
and a worked example.