Skip to contents

Changelog

Source: NEWS.md

twoPhaseSens 0.1.0

  • Initialized package development scaffold.

  • Generalized the validated six-method engine to use internal canonical variable roles rather than MIDUS-specific variable names.

  • Added initial input validation, S3 result object, diagnostics, and a provisional sensitivity-table formatter.

  • Added initial tests for input structure and Naive/CCA equivalence.

  • Finalized the initial public API for analysis settings: users can analyze one or multiple outcomes, choose a subset of methods, set the confidence level, control FCS/JM-MI computation, and supply either one reproducible seed or outcome-specific seeds.

  • Polished tbl_sensitivity() with user-controlled decimal places, outcome/method labels, optional standard-error columns, captions, and gtsummary/data-frame output modes.

  • Added plot_sensitivity() and plot() support for forest plots of method-specific estimates and confidence intervals, including outcome/method filtering, custom labels, configurable reference lines, and fixed or free facet scales.

  • Upgraded plot_sensitivity() to the manuscript publication style, including the validated six-method palette, open method-specific symbols, horizontal confidence intervals, a dotted null line, and right-side coefficient/CI and p-value annotations.

  • Completed numerical validation against the manuscript implementation for both MIDUS outcomes across all six methods.

  • Added a synthetic example-data generator and a full Getting Started vignette.

  • Refined publication forest plots for single- and multiple-outcome analyses, including portable font handling, outcome facet strips, customizable estimate labels, and manuscript-size export guidance.

  • Updated README and package metadata for release-readiness review.

  • Added runnable examples for the public API and focused regression tests for FCS-MI, JM-MI, IPW, and AIPW.

  • Expanded R CMD check to Ubuntu, Windows, and macOS on current R, plus R-devel and R-oldrel-1 on Ubuntu.

  • Verified installation and end-to-end execution from the built source tarball in a fresh R library and process.

  • Prepared version 0.1.0 for public release, including the public GitHub repository, pkgdown documentation website, full R-devel source-package checks, and Windows checks on R-devel, R-release, and R-oldrelease.

  • Standardized displayed estimates and confidence limits in sensitivity tables and forest-plot annotations to three decimal places by default.