Model averaging and muddled multimodel inferences.

By Brian S. Cade. Ecology, 2015

605 citations recorded by the source corpus.

Subfields: Ecological Modeling, Environmental Engineering, Global and Planetary Change

View Paper: https://www.semanticscholar.org/paper/2d4533a0e908d072dab7a5c17b3ee16a691e4afc

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Ecological Modeling

In Environmental Engineering

In Global and Planetary Change

Citations in the corpus

Quantifying variable importance in a multimodel inference framework (2016, background): “Whereas SW is a faithful to its definition of variable importance, it may be less useful and less intuitive than a metric that quantifies some effect size measure (Cade 2015).”
The consequences of checking for zero‐inflation and overdispersion in the analysis of count data (2019, background): “Indeed, Cade (2015) warns of ‘seriously compromised statistical interpretations’.”
Model averaging in ecology: a review of Bayesian, information-theoretic, and tactical approaches for predictive inference (2018, background): “…summarized in Banner and Higgs (2017): parameters (such as partial regression coefficients) are estimated conditional on the model structure; as the model structure changes, parameters may become incommensurable (see Posada and Buckley 2004, Cade 2015, Banner and Higgs 2017, and Appendix S1.”
A farewell to the sum of Akaike weights: The benefits of alternative metrics for variable importance estimations in model selection (2017, methodology): “That is why, contrary to assessments of absolute importance, predictor variables need to be standardised for accurate assessments of their relative ef - fect sizes based on model-averaged parameter estimates (Cade, 2015; Nakagawa & Cuthill, 2007; Schielzeth, 2010).”
MMI: Multimodel inference or models with management implications? (2015, methodology): “Because model weights are associated with models, not individual explanatory variables in the models, effect sizes are best described using model-averaged predictions rather than model-averaged coefficients, particularly for non-linear models (Cade 2015).”