NEWS

Version 1.4.1 - 2026-10-01

• Increased the package version for resubmission after CRAN reported that submitted version 1.4 was not newer than existing version 1.4.
• Clarified the descriptive-statistics wording in DESCRIPTION and completed the spelling dictionary for the cited author names.
• Updated application version labels and desktop package metadata to 1.4.1.
• Enabled incoming feasibility checks and required suggested dependencies in the default local package check, using CRAN test settings and checking the PDF manual.
• Replaced obsolete README check failures from version 0.5 with current package-check instructions and synchronized the golem configuration version.

Version 1.4 - 2026-10-01

• Added method references with author, year, and DOI to DESCRIPTION for factor retention, EGA and bootstrap stability, EFA replication, Dynamic Fit Index cutoffs, and item weighting.
• Made the final-root MAP regression checks tolerant of small numerical differences between BLAS/LAPACK implementations, addressing the Fedora and Linux ARM64 test failures. Earlier-root comparisons and the selected factor counts remain strict.
• Replaced interactive-only application startup and opt-in browser checks in the standard test suite with portable application-construction and server-flow checks. The full browser scenario remains available through dev/run_ui_tests.R.
• Fixed Turkish missingness-plot legend translation so it preserves the underlying TRUE/FALSE levels and percentages without scale warnings.
• Removed the archived EFAtools dependency and the Comparison Data factor-retention option. MAP and EKC are now calculated internally, which also removes the transitive EFAtools dependency through EFA.dimensions; EKC now follows the published observed-eigenvalue correction. All other factor-retention methods remain available.
• Legacy project files that selected Comparison Data now reset that selection to Hull Method and show a warning when loaded.

Version 1.3 - 2026-08-04

• Added internal EFA replication analysis following Osborne and Fitzpatrick (2012), including a reproducible random split, automatic factor-label and sign alignment, item-level structural comparisons, the squared loading-difference criterion, decision guidance, CSV output, and an APA 7 Word report.
• Fixed slow first launches of the Windows desktop application by loading analysis-only R packages on demand, extending the cold-start allowance, and recording each R startup stage in the diagnostic log.

Version 1.2 - 2026-08-01

• Added .fafa project files that restore the dataset, preprocessing state, factor dictionary, and interface settings.
• Added reproducible R, standalone HTML, and PDF workflow reports together with a downloadable audit log.
• Added anonymized diagnostic reports that omit data, variable names, model syntax, file paths, and user details.
• Added session-scoped caches for factor retention, EFA, EGA, CFA, reliability, and existing assumption checks.
• Added English and Turkish interface switching without restarting the application or clearing inputs.
• Added opt-in shinytest2 browser tests for the critical upload, reverse-scoring, navigation, and language flow.
• Added a visual dimension builder that generates Stratified Alpha strata automatically from saved factor definitions.
• Updated the Item Weighting attribution to Kılıç (2026) and linked the published article DOI.
• Completed Turkish translations for data selection, missing-data terminology, Item Drop Out messages, EFA reporting, Item Weighting, and the application overview.
• Added APA 7 Word reports for project summaries, assumption checks, EFA, CFA, EGA, and measurement invariance outputs.
• Added model-specific Dynamic Fit Index cutoffs to CFA while retaining conventional cutoffs for comparison.
• Relicensed FAfA under AGPL-3 and added visible source-code and warranty notices to the application.
• Integrated the Dynamic Fit Index simulation into FAfA under AGPL-3 and removed the archived external package dependency.
• Fixed correlation labels in the EFA heatmap when “Show correlation values” is selected.
• Added analysis-safe normalization for Turkish and special characters in imported variable names.
• Added automatic reverse-scoring with per-item scale detection and a visible scoring audit table.
• Updated the missingness workflow so excluded variables are removed from the Missingness Map and downstream analyses.
• Prevented empty CSV downloads when outlier removal has not been applied.
• Added a reproducible random seed setting to dataset splitting.
• Added a shared CFA factor-indicator dictionary for reliability, Stratified Alpha, CR, and AVE workflows.
• Expanded the CFA model builder with second-order, bifactor, and complex cross-loading models; multi-select lists now remain open while items are selected.
• Added lavaan syntax import and export, a larger resizable syntax editor, and the chi-square/df fit index.
• Added PNG (300 dpi), SVG, and JPG (300 dpi) downloads for CFA and EGA plots.
• Applied GUI plot settings to downloaded CFA diagrams and added distinct higher-order CFA filenames.
• Bundled an offline interface font and added a Segoe UI-first font fallback stack.
• Replaced the multi-panel scree output with a single standard eigenvalue plot and added PNG (300 dpi) and SVG downloads.
• Restored selectable colour palettes and PNG (300 dpi) and SVG downloads for the EFA correlation heatmap.
• Fixed APA 7 Word downloads so the generated DOCX files open correctly in Microsoft Word.
• Added Lubbe’s (2019) permutation parallel analysis for categorical items, with reproducible permutation, reference-quantile, and random-seed settings.
• Added Bootstrap Exploratory Graph Analysis (bootEGA) with dimension and item stability results, downloadable tables and plots, and an APA 7 Word report.
• Fixed Measurement Invariance LRT tables so concise model names are shown instead of serialized lavaan objects.
• Added group-wise ordinal category diagnostics and an audited, analysis-only adjacent-category merge when a response category is empty within a group.

Version 0.5 - 2025-12-12

This release introduces major modules for data preprocessing and item analysis, alongside enhancements to the CFA workflow.

• Missing Value Module: Added a new module for comprehensive missing data diagnostics and imputation. Users can now visualize missingness patterns and apply strategies ranging from listwise deletion to advanced imputation methods such as MICE and missForest.
• Item Drop Out: Introduced the Item Drop Out module, formerly Item-Rest, to support automated item elimination strategies for optimizing scale properties and factor structures.
• CFA Module Updates: Enhanced the CFA Model Builder to simplify covariance and model specification.
• Reliability Analysis: Added the Stratified Alpha coefficient for multidimensional reliability analysis.

mirror server hosted at Truenetwork, Russian Federation.