NEWS

metaDyn 1.0.4 (2026-09-05)

Improved convergence assessment and Hessian-rescue behavior, including more robust selection of the best candidate fit.
Treats lower-bound solutions for softplus covariance-diagonal parameters as admissible rather than automatically classifying them as convergence failures.
Hardened parameter-bound detection, nudging, and bound relaxation.

Revised the model-implied mean vector and covariance matrix for distal-outcome models so that distal outcomes are predicted by latent effect sizes rather than observed effect-size estimates.
Added support for treating covariates as either fixed definition variables or stochastic variables modeled jointly with effect sizes and distal outcomes.
Added Monte Carlo method confidence intervals.