From Harley (2002), for a given gear and sex category, the robust normal log-likelihood for proportions at length in the catch is:
 ,
where Y is the number with observed proportions at length, A is the number of length categories,  is the sample size and  is the variance. We set the maximum sample size at 3,000 to keep a few years with very large samples from dominating the fit, and used the variance of the predicted proportions (Fournier et al. 1990):
Share with your friends: |
