lavinteract NEWS

Version 0.1.0

• Initial release.
• lav_slopes(): computes conditional (simple) slopes from a fitted ‘lavaan’ model.

Version 0.1.1

• CRAN-requested correction: quote software names in Title and Description (‘lavaan’).

Version 0.2.1

• lav_vif(): variance inflation factors for structural predictors.
• Examples for lav_slopes() and lav_vif() now fully runnable and fast.
• Package help cleaned.
• Added helper %||%.

Version 0.2.2

• CRAN compliance: documented S3 method args, fixed Rd usage, and made R code ASCII-only.

Version 0.3.2

• lav_cv(): repeated holdout (Monte Carlo) cross-validation of R^2 for ‘lavaan’ models.

Version 0.3.3

• lav_cv(): CRAN-requested fixes for ASCII-only code, explicit external function calls, and S3 method documentation.

Version 0.3.4

• lav_cv(): further CRAN-requested fixes for ASCII-only code and explicit external function calls.

Version 0.3.5

• Fixed CITATION document.

Version 0.4.5

• lav_fdr(): false discovery rate (Benjamini-Yekutieli by default) correction for selected ‘lavaan’ parameter p-values.

Version 0.4.6

• lav_fdr(): fixed table appearance for long predictor or outcome names.

Version 0.5.0

• lav_vif(): added support for higher-order latent predictors, fixed multigroup handling, improved automatic data recovery.
• lav_slopes(): improved automatic data recovery, added automatic probe values for latent moderators in single-group models, added automatic plotting ranges.
• lav_jn(): Johnson-Neyman regions of significance for continuous moderators.
• lav_deltaR2(): incremental effect sizes (part R^2 and Cohen’s f^2) for structural predictors via reduced-model comparisons.
• lav_localfit(): residual-based local fit diagnostics.

Version 0.5.1

• lav_localfit(): fixed import of utils::head().

Version 0.5.2

• lav_cv(): factor-score method is now selectable via scores, and Bartlett scoring is the new default (previously regression).
• lav_cv(): added aggregation to choose the median (default) or mean split-wise summary.
• lav_cv(): the proportional overfitting index (percent drop from in-sample to cross-validated R^2) is now the primary reported metric, shown next to the absolute index (in-sample minus cross-validated R^2).
• lav_cv(): splits are retained only if the training-set solution converges and passes the post-fit admissibility check.
• lav_cv(): added an internal cap on split attempts, with a warning when the requested number of usable splits cannot be reached; the number of attempted splits is now returned.

Version 0.5.3

• lav_cv(): wrapped the example in \donttest{} to keep check time within CRAN limits.