An Initial Response to Legault: “The ability of two age composition error distributions to estimate selectivity and spawning stock biomass in simulated stock assessments”
The basis of inferences drawn from the simulation studies of Legault (2013), which indicated positive bias in estimates spawning stock biomass by certain SCAA models, together with their tendency to estimate stronger domes in selectivity than actually present, is examined. Legault’s results are indicated as likely to be an artefact of inappropriate specifications for his simulation study, which led to the generation of some datasets with unrealistically low proportions of older fish. For the more realistic of his datasets, median bias in the estimation of spawning biomass is smaller, and becomes virtually zero if a constraint on selectivity imposed by the estimation procedure which Legault applied is removed. Accordingly it seems that Legault’s concerns, including about certain SCAA models having a tendency to estimate stronger domes in selectivity than are actually present, are likely unfounded.