APPLICATION OF A GENERALIZED LOGIT MODEL TO CONDITION DATA FOR TRAWL-CAUGHT PACIFIC HALIBUT, HIPPOGLOSSUS-STENOLEPIS

Fisheries data commonly relate a categorical response variable to one or more continuous explanatory variables. We propose a generalized log it model for the analysis of such data. Our model differs from the sta ndard generalized logit model in that transformation parameters for th e explanatory variables are estimated simultaneously with other model parameters. We also employ likelihood inference and generic minimizati on techniques. Our model formulation leads to response surfaces with d istinctly different shapes for different levels of the response variab le. We illustrate our model with data on the condition of trawl-caught Pacific halibut, Hippoglossus stenolepis, as assessed by observers on commercial trawl vessels. In this data set, the probability of assign ing a particular condition level to a fish depends significantly on th e time the fish spent on the vessel's deck, fish length, total weight of the catch, tow depth, and tow duration.