Teaching Inference for Capture-Recapture Data: Populations, Samples, Probabilities, Parameters and Models
Kenneth P. Burnham
Fish, Widlife, and Conservation Biology, Colorado State University, Fort Collins, CO, United States

The initial literature on capture-recapture theory was often statistical nonstandard and different models (e.g., open vs. closed models, live capture vs. dead recoveries) were developed separately, hence with different notation and appearance. Capture-recapture theory and application has matured and increased dramatically since 1990, has been unified, and has entered the statistical mainstream. This rapid development poised a challenge for teaching a graduate level course on the subject. At Colorado State University the challenge was met by focusing on the unified theory of the subject with simplified notation, creation of a single computer program for numerically fitting all models, abandoning closed-form solutions for estimators, emphasizing that the models were generalized linear models, and that basic ideas of sampling theory were also useful for developing capture-recapture models. Only by focusing on a unified approach, with standard statistical theory, can we effectively teach of the vast array of capture-recapture models and statistical inference based on these models.

Keywords: Capture-recapture; Teaching inference

Biography: At the time of his nominal retirement (2 Jan. 2009) Ken was a Senior Scientist employed by the United States Geological Survey, in the Biological Resources section, as an Assistant Unit Leader, Colorado Cooperative Fish and Wildlife Research Unit, and Professor at Colorado State University. His formal degrees: a B.S. in Biology, and a M.S. and Ph.D. in statistics. His expertise and interests: study design (statistical aspects) and specialized sampling and data analysis methodology for wildlife and ecology, in particular population estimation, such as from capture-recapture, bird banding, distance sampling, population monitoring, and inference about population dynamics. Ken's recent research emphasis has been on theory and application of information-theoretic data-based model selection and multimodel inference.