Final Report: Joint investigation into statistical methodologies underpinning the derivation of toxicant guideline values in Australia and New Zealand

This report summarises the outcomes of a one-year joint collaborative research project involving Australian and Canadian researchers who undertook extensive investigations into methodologies and tools associated with SSD modelling, focusing on Burrlioz and ssdtools.

Research undertaken included exploring and resolving issues with numerical instability and developing alternative solutions for estimating hazard concentrations (HCx) and/or associated 95% confidence intervals of the Burr III and inverse Pareto distributions; comparisons between Burrlioz and ssdtools using collated benchmark and synthetic datasets; the development and evaluation of mixture distributions as a potential candidate for accommodating bimodal data; and an assessment of bias and coverage across a range of candidate distribution sets used in model averaging. Based on the findings, the project recommends that the Australian-New Zealand and Canadian jurisdictions adopt the R-package ssdtools using an expanded default set of distributions.

The report was prepared for the Department of Agriculture, Water and the Environment.