Checks convergence of an assessment model

Utilty function for use within assess_lmm. Checks whether a model has converged. Currently only checks assessments with normal (or lognormal) errors, where it considers whether:

• fixed effect estimates are away from their bounds

• random effect estimates are lower than their upper bounds; they can of course be equal to zero

• standard errors are present for model predictions and fixed effects estimates

• standard errors of the fixed effects estimates are not unrealistically small; the tolerance is chosen to be much smaller than seen in typical OSPAR assessments which have converged

  Model fits based on other distributions are assumed to have converged. Some checking is needed here in future

Usage

check_convergence_lmm(assessment, coeff_se_tol = 0.001)

Arguments

assessment

An assessment from assess_lmm

coeff_se_tol

The tolerance for checking whether standard errors on the fixed effects estimates are unrealistically small; defaults to 0.001

Value

An integer: 0 indicates convergence, 1 indicates an issue