Arni Magnusson
UW School of Aquatic & Fishery Sciences
April 27, 2005

Uncertainty is a fundamental part of fisheries stock assessment, that needs to be quantified to successfully manage the resource. Among the statistical methods that are used to measure uncertainty are Markov chain Monte Carlo (MCMC) simulations, bootstrap, and Hessian delta-method approximation. In this study, a large number of stochastic datasets is generated, where the true parameter values are known. Confidence bounds are then estimated using the different methods, and the claimed uncertainty is compared with how often they contain the true value. The findings from this simulation study are reviewed, as well as theoretical and practical differences between the methods.