Philosophy and the practice of Bayesian statistics

Abstract

A substantial school in the philosophy of science identifies Bayesian inference with inductive inference and even rationality as such, and seems to be strengthened by the rise and practical success of Bayesian statistics. We argue that the most successful forms of Bayesian statistics do not actually support that particular philosophy but rather accord much better with sophisticated forms of hypothetico-deductivism. We examine the actual role played by prior distributions in Bayesian models, and the crucial aspects of model checking and model revision, which fall outside the scope of Bayesian confirmation theory. We draw on the literature on the consistency of Bayesian updating and also on our experience of applied work in social science. Clarity about these matters should benefit not just philosophy of science, but also statistical practice. At best, the inductivist view has encouraged researchers to fit and compare models without checking them; at worst, theorists have actively discouraged practitioners from performing model checking because it does not fit into their framework.

36 pages, 5 figures. v2: Fixed typo in caption of figure 1. v3: Further typo fixes. v4: Revised in response to referees

References in corpus (7)

• A weakly informative default prior distribution for logistic and other regression models
• Misspecification in infinite-dimensional Bayesian statistics
• Frequentist statistics as a theory of inductive inference
• Relaxation Penalties and Priors for Plausible Modeling of Nonidentified Bias Sources
• Bayesian Checking of the Second Levels of Hierarchical Models
• Comment: Bayesian Checking of the Second Levels of Hierarchical Models
• Model Assessment Tools for a Model False World

Cited by in corpus (22)

• The prior can generally only be understood in the context of the likelihood
• Bayesian inference of network structure from unreliable data
• Belief and Truth in Hypothesised Behaviours
• From Probability to Consilience: How Explanatory Values Implement Bayesian Reasoning
• Improved information criteria for Bayesian model averaging in lattice field theory
• Reconciling Kinetic and Equilibrium Models of Bacterial Transcription
• Statistical reform and the replication crisis
• Posterior predictive checks to quantify lack-of-fit in admixture models of latent population structure
• On Model Selection in Cosmology
• Bayesian Renormalization
• Intrinsic and extrinsic thermodynamics for stochastic population processes with multi-level large-deviation structure
• Models, measurement and inference in epithelial tissue dynamics
• Frequentist tests for Bayesian models
• Investigating subphotospheric dissipation in gamma-ray bursts using joint Fermi-Swift observations
• Moving Beyond Chi-Squared in Nuclei and Neutron Stars
• The Posterior Predictive Null
• Data-Space Validation of High-Dimensional Models by Comparing Sample Quantiles
• Irrational Exuberance: Correcting Bias in Probability Estimates
• Probability Models in Statistical Data Analysis: Uses, Interpretations, Frequentism-As-Model
• Accelerated Parallel Non-conjugate Sampling for Bayesian Non-parametric Models
• Over-representation of phonological features in basic vocabulary doesn't replicate when controlling for spatial and phylogenetic effects
• Bayesian Unbiasing of the Gaia Space Mission Time Series Database