Bayes factor
Definition
Bayes factor refers to a statistical measure used in Bayesian graphical modeling to test for conditional independence between pairs of variables by comparing the relative evidence for competing hypotheses about network structure. Specifically, the edge inclusion Bayes factor is employed within Bayesian model averaging techniques to assess whether an edge should be present or absent in a graphical model. The inclusion Bayes factor requires specification of two sets of prior distributions: one for the network structure (the configuration of edges) and another for the edge weight parameters indicating the strength of connections. The sensitivity of the inclusion Bayes factor to these prior specifications—particularly the scale of the prior on partial correlations—is substantial, with even small variations in prior choice capable of altering the Bayes factor's sensitivity and its ability to distinguish between the presence and absence of edges.
Sources: Sekulovski et al. (2024)
Related Terms
Applications
Bayes Factor and Prior Distributions
The choice and scale of prior distributions for both network structure and edge weight parameters have a significant impact on the values and reliability of the inclusion Bayes factor in Bayesian graphical modeling. Applied researchers must make informed prior choices to prevent misleading conclusions and ensure robust evidence for or against specific network structures.
Sources: Sekulovski et al. (2024)
Bayes Factor and Conditional Independence Testing
The edge inclusion Bayes factor is a methodological tool for testing conditional independence between variable pairs in graphical models. This approach provides a solution for guiding informed prior choices in conditional independence testing.
Sources: Sekulovski et al. (2024)
