Multimodel approaches are not the best way to understand multifactorial systems

This is a Preprint and has not been peer reviewed. This is version 3 of this Preprint.

Add a Comment

You must log in to post a comment.


Comments

There are no comments or no comments have been made public for this article.

Downloads

Download Preprint

Authors

Benjamin Bolker 

Abstract

Information-theoretic (IT) and multi-model averaging (MMA) statistical approaches are widely used but suboptimal tools for pursuing a multifactorial approach (also known as the method of multiple working hypotheses) in ecology. (1) Conceptually, IT encourages ecologists to perform tests on sets of artificial models. (2) MMA improves on IT model selection by implementing a simple form of *shrinkage estimation* (a way to make accurate predictions from a model with many parameters, by "shrinking" parameter estimates toward zero). However, other shrinkage estimators such as penalized regression or Bayesian hierarchical models with regularizing priors are more computationally efficient and better supported theoretically. (3) In general the procedures for extracting confidence intervals from MMA are overconfident, giving overly narrow intervals. If researchers want to accurately estimate the strength of multiple competing ecological processes along with reliable confidence intervals, the current best approach is to use full (maximal) statistical models (possibly with Bayesian priors) after making principled, a priori decisions about which predictors to include.

DOI

https://doi.org/10.32942/X2Z01P

Subjects

Applied Statistics, Ecology and Evolutionary Biology

Keywords

multi-model averaging, model selection, coverage, AIC, inference, shrinkage

Dates

Published: 2023-07-23 11:14

Last Updated: 2024-03-01 03:41

Older Versions
License

CC-By Attribution-ShareAlike 4.0 International

Additional Metadata

Language:
English

Conflict of interest statement:
None

Data and Code Availability Statement:
Not applicable