Ecologists overestimate the importance of predictor variables in model averaging: a plea for cautious interpretations

By M. Galipaud, Mark A. F. Gillingham, M. David, F. Dechaume‐Moncharmont. 2014

147 citations recorded by the source corpus.

Subfields: Ecological Modeling, Environmental Chemistry, Nature and Landscape Conservation

View Paper: https://www.semanticscholar.org/paper/a4bbe070e40d5ae6ac0ce63ff1fd34a60eec286d

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 Chemistry

In Nature and Landscape Conservation

Citations in the corpus

A farewell to the sum of Akaike weights: The benefits of alternative metrics for variable importance estimations in model selection (2017, methodology): “For each sample size, the mean SW (dot), interquartile interval (thick line) and 95% interval (thin line) were calculated from 10,000 simulated independent datasets (Galipaud et al., 2014).”
MMI: Multimodel inference or models with management implications? (2015, background): “Yet, this simple summary is likely to be much less informative than estimates of regression coefficients and their standard errors (Murray and Connor 2009, Galipaud et al. 2014, Cade 2015).”
Identifying Causes of Patterns in Ecological Networks: Opportunities and Limitations (2017, methodology): “Then the statistical support for each predictor variable can be calculated by summing over all models (Schleuning et al. 2011, González-Castro et al. 2015; but see Galipaud et al. 2014).”
A brief introduction to mixed effects modelling and multi-model inference in ecology (2018, background): “Recent work has demonstrated that this approach is flawed because Akaike weights are interpreted as relative model probabilities, and give no information about the importance of individual predictors in a model (Cade, 2015), and fail to distinguish between variables with weak or strong effects (Gali”
Quantifying variable importance in a multimodel inference framework (2016, methodology): “…or model probabilities (SW) for each explanatory variable has been proposed as a measure of relative variable importance (RVI) in linear models (Burnham & Anderson 2002, p. 168), but recent simulation studies concluded that SW is a poor estimate of RVI (Murray & Conner 2009; Galipaud et al. 2014).”