Model Comparison and the Principle of Parsimony

By Joachim Vandekerckhove, D. Matzke, Eric-Jan Wagenmakers. 2015

338 citations recorded by the source corpus.

How researchers see this paper

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

In Artificial Intelligence

Citations in the corpus

Balancing Type I Error and Power in Linear Mixed Models (2015, methodology): “rz,1978)andtheaforementionedlikelihood ratio test. For our simulations, we use the AIC and LRT criteria. The BIC is known to put a strong penalty on the model complexity for small sample sizes (e.g., Vandekerckhove et al., 2015). IncontrasttotheAIC,whichallowsustocompareseveralmodelsatonce,theLRTcan”
bridgesampling: An R Package for Estimating Normalizing Constants (2017, neutral): “The marginal likelihood is obtained by integrating out the model parameters with respect to their prior distribution: The marginal likelihood implements the principle of parsimony also known as Occam’s razor (e.g., Jeﬀerys and Berger 1992; Myung and Pitt 1997; Vandekerckhove et al. 2015).”
Robust Modeling in Cognitive Science (2019, neutral): “The canonical Bayesian approach to model selection is based on the Bayes factor, which is a ratio expressing how much evidence the data provide for one model over another (Kass and Raftery 1995; Lee and Wagenmakers 2013, Chapter 7; Vandekerckhove et al. 2015).”
Safe models for risky decisions (2017, methodology): “A suitable importance density should (1) be easy to evaluate; (2) have the same domain as the posterior distribution; (3) closely resemble the posterior distribution, and (4) have fatter tails than the posterior distribution (Neal, 2001; Vandekerckhove et al., 2015).”
Model selection for spectral parameterization (2024, unknown): “With ms-specparam , we performed Bayes factor analyses (Vandekerckhove et al., 2015) as an objective measure of evidence for the presence of rhythmic activity in the neurophysiological power spectrum.”