Deviance Information Criterion¶
Compare Bayesian hierarchical models by adding posterior mean deviance to an effective-complexity penalty derived from posterior deviance, using quantities readily estimated from MCMC draws.
Core Idea¶
For data (y), parameter ( heta), and likelihood \(p(y\mid\theta)\), define deviance \(D(\theta)=-2\log p(y\mid\theta)+C\). The deviance information criterion uses posterior draws to estimate fit and effective complexity:
Equivalently, \(\mathrm{DIC}=D(\bar\theta)+2p_D\). Smaller values are preferred among models fit to the same data with the same likelihood and deviance convention. DIC was designed for complex hierarchical models where the nominal parameter count is unclear but posterior deviance is cheap to compute from MCMC output.
Scope of Application¶
DIC is most natural for approximately regular hierarchical models with posterior distributions summarized sensibly by a mean and with a clearly defined likelihood focus. It is easy to compute in BUGS-style workflows.
It can behave poorly under multimodality, skewness, mixtures, weak identification, parameterization changes, negative effective parameter counts, or ambiguous choices about latent variables in the likelihood. The 2014 retrospective emphasizes these limitations.
Clarity¶
Report the deviance definition, which stochastic nodes count as parameters, the likelihood focus, the (p_D) variant, posterior summary, observation granularity, MCMC diagnostics, and comparison set. A DIC value is meaningless in isolation; only differences under a common construction are interpretable.
Manages Complexity¶
DIC compresses a posterior sample into one fit–complexity coordinate without requiring a nominal dimension for a hierarchical model. It also decomposes the score so analysts can see whether a model wins through fit or pays for flexibility. The convenience is computational, not a guarantee that the approximation matches predictive performance.
Abstract Reasoning¶
- Fix comparable data and likelihood definitions.
- Obtain converged posterior draws.
- Compute deviance for every retained draw.
- Average the deviance.
- Compute \(D(\bar\theta)\) under a declared parameterization.
- derive (p_D) and inspect its plausibility.
- Add fit and penalty.
- Compare differences, not absolute scores.
- Cross-check with posterior predictive diagnostics, WAIC, or LOO when irregularity matters.
Knowledge Transfer¶
The portable pattern is estimate flexibility from the gap between average realized fit and fit at a representative state, then penalize average fit by that effective flexibility. The proposed immediate parent is Optimality Criterion.
Relationships to Other Abstractions¶
Current abstraction Deviance Information Criterion Domain-specific
Parents (1) — more general patterns this builds on
-
Deviance Information Criterion is a kind of Optimality criterion Domain-specific
Optimality Criterion is the proposed immediate parent.
Hierarchy paths (2) — routes to 2 parentless roots
- Deviance Information Criterion → Optimality criterion → Evaluation → Comparison → Self Checking
- Deviance Information Criterion → Optimality criterion → Selection
Neighborhood in Abstraction Space¶
Deviance Information Criterion sits in a sparse region of the domain-specific corpus (97th percentile for distinctiveness): few abstractions share its structure, so a faithful description tends to retrieve it precisely.
Family — Unclustered & Miscellaneous (1565 abstractions)
Nearest neighbors
- Probability of direction — 0.76
- Widely applicable information criterion — 0.76
- Information matrix test — 0.75
- Prior Probability — 0.75
- Universal Hypothesis Testing — 0.75
Computed from structural-signature embeddings · 2026-09-08