Premium
How to walk on statistical mandalas as a population ecologist
Author(s) -
Toquenaga Yukihiko
Publication year - 2016
Publication title -
population ecology
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.819
H-Index - 59
eISSN - 1438-390X
pISSN - 1438-3896
DOI - 10.1007/s10144-015-0532-z
Subject(s) - statistical inference , population , compromise , bayesian probability , inference , bayesian inference , statistics , statistical model , ecology , feature (linguistics) , econometrics , data science , artificial intelligence , machine learning , computer science , biology , mathematics , sociology , demography , philosophy , social science , linguistics
Abstract We population ecologists who are believed to be good at dealing with statistics often get confused about what kinds of statistical methods we should apply to our nuisance data. There are a couple of conflicting paradigms and many associated methods in statistics. Classical frequentists’ approaches that have dominated in science have been severely criticized by the newcomers: Bayesian and evidential statistics. But, both newcomers also have weak points. Researchers devoted to different statistical approaches are seeking soft landing places where they can compromise each other. Key aspects of statistical inference are discriminating model selection and parameter estimation. Likelihood and Fisher information play important roles in both processes. As an overview of the compromise processes, here I will introduce three contributing papers by M. L. Taper, J. M. Ponciano, R. M. Dorazio, and K. Yamamura for the special feature entitled “Bayesian, Fisherian, error, and evidential statistical approaches for population ecology.” This special feature is based on a symposium held in Tsukuba, Japan, on 11 October 2014