2019 STAR Panel: Selectivity sahpe

[chantelwetzel-noaa]

Reconsider spline or some other non-parametric selectivity forms, as there are unexpected behaviors observed in estimating double normal selectivities for the fixed gear fishery. The requirement (apparently) to have to fix the “P6” parameter related to old sablefish selectivity and some other interactions seems unusual, and several of the convergence issues were related to selectivity parameters for some gear types/surveys. The 2015 model used a different form for selectivity and the document provided little justification to choose one over the other. “Fewer parameters” is a poor reason if performance overall is so much worse. The convergence problems did seem to be largely resolved after the STAT fixed the CV of young ages.