A bootstrap method for estimating bias and variance in statistical multispecies models using highly disparate data sets

TL;DR

This paper introduces a bootstrap method for estimating bias and variance in multispecies marine ecosystem models, effectively handling complex data correlations and improving uncertainty quantification over traditional Hessian-based methods.

Contribution

A novel bootstrap approach for variance estimation in multispecies models that accounts for spatial and temporal correlations, reducing reliance on model assumptions.

Findings

Effective in real data applications

Identifies estimation bias and data aggregation effects

Provides more reliable uncertainty estimates

Abstract

Statistical multispecies models of multiarea marine ecosystems use a variety of data sources to estimate parameters using composite or weighted likelihood functions with associated weighting issues and questions on how to obtain variance estimates. Regardless of the method used to obtain point estimates, a method is needed for variance estimation. A bootstrap technique is introduced for the evaluation of uncertainty in such models, taking into account inherent spatial and temporal correlations in the data sets thus avoiding many model--specification issues, which are commonly transferred as assumptions from a likelihood estimation procedure into Hessian--based variance estimation procedures. The technique is demonstrated on a real data set and used to look for estimation bias and the effects of different aggregation levels in population dynamics models.