Abstract

In the wildlife literature there has been some recent criticism of statistical significance testing. In the past few years, both the Journal of ·Wildlife Management and the Wildlife Society Bulletin have published articles criticizing the overuse and misuse of hypothesis tests. One alternative to using hypothesis tests for model selection is the information-theoretic approach, proposed by Burnham and Anderson (1998). This technique uses values such as the Akaike Information Criterion and others to choose a set of plausible models from a set of a prioTi candidate models. Inferences are based on the set of plausible models, rather than on a single selected best model, and model-averaged point estimates of parameters may be used for prediction. The Burnham and Anderson method is gaining popularity in the wildlife science community, and statisticians who work with wildlife scientists should be aware of this analysis technique and how to use it properly. This paper will introduce statisticians to the information-theoretic approach to model selection and the statistical theory underlying it, as well as demonstrate the technique using data on bird species richness and abundance in riparian areas in southeastern Nebraska.

Keywords

Akaike information criterion, Schwarz's Bayesian Information Criterion, Kullback-Liebler distance

Creative Commons License


This work is licensed under a Creative Commons Attribution-Noncommercial-No Derivative Works 4.0 License.

Share

COinS
 
Apr 28th, 10:30 AM

THE INFORMATION-THEORETIC APPROACH TO MODEL SELECTION: DESCRIPTION AND CASE STUDY

In the wildlife literature there has been some recent criticism of statistical significance testing. In the past few years, both the Journal of ·Wildlife Management and the Wildlife Society Bulletin have published articles criticizing the overuse and misuse of hypothesis tests. One alternative to using hypothesis tests for model selection is the information-theoretic approach, proposed by Burnham and Anderson (1998). This technique uses values such as the Akaike Information Criterion and others to choose a set of plausible models from a set of a prioTi candidate models. Inferences are based on the set of plausible models, rather than on a single selected best model, and model-averaged point estimates of parameters may be used for prediction. The Burnham and Anderson method is gaining popularity in the wildlife science community, and statisticians who work with wildlife scientists should be aware of this analysis technique and how to use it properly. This paper will introduce statisticians to the information-theoretic approach to model selection and the statistical theory underlying it, as well as demonstrate the technique using data on bird species richness and abundance in riparian areas in southeastern Nebraska.