No abstract

No abstract

No abstract

No abstract

An analysis of a trouser trawl experiment with multiple subsampled hauls is presented. The standard method of analysis involves scaling catches by subsampling fractions and combining the catches for each haul to obtain an estimate of the size selectivity of the test gear. This procedure results in invalid statistical inferences and is not appropriate if the split of fish is not the same in all hauls or if subsampling fractions have measurement error. Modifications of the selectivity model are proposed that significantly improve the analysis of this type of data (e.g. to estimate selectivity without subsampling fractions). Likelihood ratio tests are developed to test which modified selectivity model best fits the data. Confidence bounds are developed as diagnostic tools. The procedures are illustrated with a data analysis.