A comparison of statistical methods for fitting population models to data

A comparison was made of 3 statistical methods for fitting population model s to abundance data: (1) Neyman-Pearson (frequentist), (2) Bayesian, and (3 ) Likelihood. Each school of statistical inference was briefly described, a long with a description of the most common techniques used in each school f or estimating the parameters of complex models, such as in fitting populati on models to abundance data. An example of each type of inference was shown by using the same data to estimate the depletion level of a population of spotted dolphins. The 3 specific techniques used were maximum likelihood es timation with a non-parametric bootstrap, approximation of the Bayesian pos terior distribution using the sampling-importance-resampling numerical inte gration routine, and profile likelihood to represent likelihood inference f or the parameter of interest. Point estimates from each technique were simi lar, but intervals varied substantially. Strengths and weaknesses of each a pproach were illustrated through the example.