Bayesian Model Comparison for Time-Varying Parameter VARs with Stochastic Volatility

In most applications of Bayesian model comparison or Bayesian hypothesis testing, the results are reported in terms of the Bayes factor only, not in terms of the posterior probabilities of the models. Posterior model probabilities are not reported because researchers are reluctant to declare prior model probabilities, which in turn stems from uncertainty in the prior. Fortunately, Bayesian formalisms are designed to embrace prior uncertainty, not ignore it. This article provides a novel derivation of the posterior distribution of model probability, and shows many examples. The posterior distribution is useful for making decisions taking into account the uncertainty of the posterior model probability. Benchmark Bayes factors are provided for a spectrum of priors on model probability. R code is posted at https://osf.io/36527/. This framework and tools will improve interpretation and usefulness of Bayes factors in all their applications.

Download Full-text

Combined-chain nested sampling for efficient Bayesian model comparison

Digital Signal Processing ◽

10.1016/j.dsp.2017.07.021 ◽

2017 ◽

Vol 70 ◽

pp. 84-93 ◽

Cited By ~ 2

Author(s):

R. Wesley Henderson ◽

Paul M. Goggans ◽

Lei Cao

Keyword(s):

Bayesian Model ◽

Model Comparison ◽

Nested Sampling ◽

Bayesian Model Comparison

Download Full-text

Capturing Ordinal Theoretical Constraint in Psychological Science

10.31234/osf.io/a4xu9 ◽

2018 ◽

Cited By ~ 1

Author(s):

Julia M. Haaf ◽

Fayette Klaassen ◽

Jeffrey Rouder

Keyword(s):

Social Sciences ◽

Linear Model ◽

Bayesian Model ◽

Model Comparison ◽

Social Scientists ◽

Psychological Science ◽

Bayesian Model Comparison ◽

Comparison Approach ◽

The Social ◽

Ordinal Level

Most theories in the social sciences are verbal and provide ordinal-level predictions for data. For example, a theory might predict that performance is better in one condition than another, but not by how much. One way of gaining additional specificity is to posit many ordinal constraints that hold simultaneously. For example a theory might predict an effect in one condition, a larger effect in another, and none in a third. We show how common theoretical positions naturally lead to multiple ordinal constraints. To assess whether multiple ordinal constraints hold in data, we adopt a Bayesian model comparison approach. The result is an inferential system that is custom-tuned for the way social scientists conceptualize theory, and that is more intuitive and informative than current linear-model approaches.

Download Full-text