## Markov Chain Monte Carlo Model Determination for Hierarchical and Graphical Log-linear Models (1996)

Venue: | Biometrika |

Citations: | 55 - 8 self |

### BibTeX

@ARTICLE{Dellaportas96markovchain,

author = {Petros Dellaportas and Jonathan J. Forster},

title = {Markov Chain Monte Carlo Model Determination for Hierarchical and Graphical Log-linear Models},

journal = {Biometrika},

year = {1996},

volume = {86},

pages = {615--633}

}

### Years of Citing Articles

### OpenURL

### Abstract

this paper, we will only consider undirected graphical models. For details of Bayesian model selection for directed graphical models see Madigan et al (1995). An (undirected) graphical model is determined by a set of conditional independence constraints of the form `fl 1 is independent of fl 2 conditional on all other fl i 2 C'. Graphical models are so called because they can each be represented as a graph with vertex set C and an edge between each pair fl 1 and fl 2 unless fl 1 and fl 2 are conditionally independent as described above. Darroch, Lauritzen and Speed (1980) show that each graphical log-linear model is hierarchical, with generators given by the cliques (complete subgraphs) of the graph. The total number of possible graphical models is clearly given by 2 (

### Citations

1144 | Bayes factors - Kass, Raftery - 1995 |

921 | Reversible jump Markov chain Monte Carlo computation and Bayesian model determination
- Green
- 1995
(Show Context)
Citation Context ... the posterior probability of each plausible model. For high-dimensional contingency tables, the set of plausible models is very large. We focus attention on reversible jump Markov chain Monte Carlo (=-=Green, 1995-=-) and develop strategies for calculating posterior probabilities of hierarchical, graphical or decomposable log-linear models. Even for tables of moderate size, these sets of models may be very large.... |

319 |
Adaptive rejection sampling for Gibbs sampling
- Gilks, Wild
- 1992
(Show Context)
Citation Context ...he model parameters, discussed in Section 3, is a special case. They showed that the univariate conditional distributions required for sampling are log-concave, and hence adaptive rejection sampling (=-=Gilks and Wild, 1992-=-) is straightforward to implement. Forster and Skene (1994) used a Gibbs sampler to generate from posterior distributions arising from multinomial likelihoods and logistic normal prior distributions, ... |

292 | Model Selection and Accounting for Model Uncertainty in Graphical Models Using Occam’s Window
- Madigan, Raftery
- 1994
(Show Context)
Citation Context ...ingle edge. Hence, in principle, all posterior model probabilities (1) are straightforward to calculate. However, for large tables, the number of models may prohibit this calculation. Occam's window (=-=Madigan and Raftery, 1994-=-) and Markov Chain Monte Carlo model composition (MC 3 , Madigan and York, 1995) are strategies for overcoming this problem. A comparison of the two approaches is provided by Madigan et al (1994). The... |

255 | Bayesian Graphical Models for Discrete Data
- Madigan, York
- 1995
(Show Context)
Citation Context ...tforward to calculate. However, for large tables, the number of models may prohibit this calculation. Occam's window (Madigan and Raftery, 1994) and Markov Chain Monte Carlo model composition (MC 3 , =-=Madigan and York, 1995-=-) are strategies for overcoming this problem. A comparison of the two approaches is provided by Madigan et al (1994). There are, however, interesting models which fall outside the class of decomposabl... |

129 |
Assessment and propagation of model uncertainty [with discussion
- Draper
- 1995
(Show Context)
Citation Context ...culation of posterior probabilities of all competing models, regardless of their relative size or structure, and this model uncertainty can be incorporated into any decisions or predictions required (=-=Draper, 1995-=-, gives examples of this). The problems with this approach are associated with the computation of f(mjn) using (1), which requires calculation of the marginal likelihood f(njm) = R \Theta m f(njm; `m ... |

123 | Hyper markov laws and statistical analysis of decomposable graphical models - Dawid, Lauritzen - 1993 |

117 |
Fractional Bayes factors for model comparisons
- O’Hagan
- 1995
(Show Context)
Citation Context ...tion for the model parameters for each model. Such prior distributions must be proper, otherwise (1) is undefined, although possible solutions to this problem do exist (Spiegelhalter and Smith, 1982; =-=O'Hagan, 1995-=-). However, for contingency tables, it is straightforward to construct prior distributions which are both vague and proper. The choice of prior distribution for the model parameters may have a conside... |

108 | Approximate Bayes factors and accounting for model uncertainty in generalised linear models." Biometrika 83:251{266 - Raftery - 1996 |

99 |
Practical Markov chain Monte Carlo (with discussion
- Geyer
- 1992
(Show Context)
Citation Context ...esults displayed in the table are based on a sample of 500 000. The MCMC standard errors of the estimates of the model probabilities are calculated by splitting the MCMC output into 10 `batches' (see =-=Geyer, 1992-=-). The model probabilities did not change substantially, even when the sample size was increased by a factor of ten. Only the model probabilities for the four most probable models are displayed. For a... |

90 | Markov fields and log-linear interaction models for contingency tables. Annals of S~atistics - Darroch, Lauritzen, et al. - 1980 |

50 | Hypothesis testing and model selection - Raftery - 1996 |

49 | A fast procedure for model search in multidimensional contingency tables - EDWARDS, HAVRANEK - 1985 |

44 |
Bayes factors for linear and log-linear models with vague prior information
- Spiegelhalter, Smith
- 1982
(Show Context)
Citation Context ...non-informative) prior distribution for the model parameters for each model. Such prior distributions must be proper, otherwise (1) is undefined, although possible solutions to this problem do exist (=-=Spiegelhalter and Smith, 1982-=-; O'Hagan, 1995). However, for contingency tables, it is straightforward to construct prior distributions which are both vague and proper. The choice of prior distribution for the model parameters may... |

40 | Bayesian inference for generalized linear and proportional hazards models via Gibbs sampling - Dellaportas, Smith - 1993 |

39 | Bayesian model averaging and model selection for Markov equivalence classes of acyclic digraphs - Madigan, Andersson, et al. - 1996 |

25 | Strategies for Graphical Model Selection - Madigan, Raftery, et al. - 1994 |

24 | Some observations on inverse probability including a new indifference rule - PERKS - 1947 |

15 | Incorporating prior information into the analysis of contingency tables - Knuiman, Speed - 1988 |

13 | Bayes factors for independence in contingency tables - Günel, Dickey - 1974 |

6 | A Bayesian test for a two-way contingency table using independence priors. The Canadian - Albert - 1990 |

4 | The robustness and sensitivity of the mixed-Dirichlet Bayesian test for “independence” in contingency tables - Good, Crook - 1987 |

3 | Using Gibbs sampling for Bayesian inference in multidimensional contingency tables - Epstein, Fienberg - 1991 |

3 | Calculation of marginal densities for parameters of multinomial distributions - Forster, Skene - 1994 |

1 | Bayesian selection of log-linear models. Working Paper 95-15 - Albert - 1995 |