## Causal Inference in the Presence of Latent Variables and Selection Bias (0)

Venue: | In Proceedings of Eleventh Conference on Uncertainty in Artificial Intelligence |

Citations: | 28 - 4 self |

### BibTeX

@INPROCEEDINGS{Spirtes_causalinference,

author = {Peter Spirtes and Christopher Meek and Thomas Richardson},

title = {Causal Inference in the Presence of Latent Variables and Selection Bias},

booktitle = {In Proceedings of Eleventh Conference on Uncertainty in Artificial Intelligence},

year = {},

pages = {499--506},

publisher = {Morgan Kaufmann}

}

### OpenURL

### Abstract

This paper uses Bayesian network models for that investigation. Bayesian networks, or directed acyclic graph (DAG) models have proved very useful in representing both causal and statistical hypotheses. The nodes of the graph represent vertices, directed edges represent direct influences, and the topology of the graph encodes statistical constraints. We will consider features of such models that can be determined from data under assumptions that are related to those routinely applied in experimental situations:

### Citations

7053 | Probabilistic Reasoning in Intelligent Systems - Pearl - 1988 |

1075 | Herskovitz: A Bayesian Method for the Induction
- Cooper, E
- 1992
(Show Context)
Citation Context ...bability distribution conditional on each vector of values of the variables having edges directed into V. This data has been used to test several different discovery algorithms. (Spirtes et al. 1993, =-=Cooper and Herskovits 1992-=-, Chickering 1994) The interpretation of the variables is not relevant to the study described here. 6 5 4 27 11 32 34 35 36 37 19 20 31 15 23 16 10 21 22 13 17 28 29 12 24 25 18 26 7 8 9 1 2 3 30 33 1... |

903 | Learning Bayesian networks: The combination of knowledge and statistical data
- Heckerman, Geiger, et al.
- 1995
(Show Context)
Citation Context ...s been some progress recently in heuristic greedy DAG searches based upon maximizing some model score such as the posterior probability of the minimum description length. (Cooper and Herskovits 1992, =-=Heckerman et al. 1994-=-, Chickering et al. 1995.) Combining an independence test algorithm for DAG search (essentially a special case of FCI) and greedy DAG searches based upon maximizing a score has proved successful in si... |

749 | Structural Equations with Latent Variables - Bollen - 1989 |

239 |
The ALARM monitoring system: A case study with two probabilistic inference techniques for belief networks
- Beinlich, Suermondt, et al.
- 1989
(Show Context)
Citation Context ...ance of the algorithm. We used 10,000 cases generated pseudo-randomly by Cooper(1992) from the Alarm network, shown in Figure 11. The ALARM network was developed to model an emergency medical system (=-=Beinlich, et al. 1989-=-). The 37 variables are all discrete, taking 2, 3 or 4 distinct values. There are 46 edges in the DAG. In most instances a 25 directed arrow indicates that one variable is regarded as a cause of anoth... |

217 | Equivalence and synthesis of causal models - Verma, Pearl - 1990 |

172 | Causal diagrams for empirical research - Pearl - 1995 |

140 | Independence properties of directed markov fields - Lauritzen, Dawid, et al. - 1990 |

81 | An algorithm for fast recovery of sparse causal graphs - Spirtes, Glymour - 1991 |

77 | An algorithm for the construction of Bayesian network structures from data - Singh, Valtorta - 1993 |

73 | D.: Learning Bayesian networks: Search methods and experimental results - Chickering, Geiger, et al. - 1995 |

56 |
Learning bayesian networks with discrete variables from data
- Spirtes, Meek
- 1995
(Show Context)
Citation Context .... 1995.) Combining an independence test algorithm for DAG search (essentially a special case of FCI) and greedy DAG searches based upon maximizing a score has proved successful in simulation studies (=-=Spirtes and Meek 1995-=-, Singh and Valorta 1993). An analogous strategy might improve the accuracy of the FCI algorithm, although the task of calculating scores for a PAG faces a number of obstacles. The FCI algorithm will ... |

49 |
Strong completeness and faithfulness in Bayesian networks
- Meek
- 1995
(Show Context)
Citation Context ...bining an independence test algorithm for DAG search (essentially a special case of FCI) and greedy DAG searches based upon maximizing a score has proved successful in simulation studies (Spirtes and =-=Meek 1995-=-, Singh and Valorta 1993). An analogous strategy might improve the accuracy of the FCI algorithm, although the task of calculating scores for a PAG faces a number of obstacles. The FCI algorithm will ... |

28 |
A discovery algorithm for directed cyclic graphs
- Richardson
- 1996
(Show Context)
Citation Context ...tion of a PAG and the assumed acyclicity of the directed graphs, that there are no edges AsB in a PAG, and no directed cycles in a PAG. (PAGs can also be used to represent directed cyclic graphs. See =-=Richardson 1996). Informa-=-lly, a directed path in a PAG is a path that contains only "" edges pointing in the same direction. Theorem 1: If p is a partial ancestral graph, and there is a directed path U from A to B i... |

24 | Causality from Probability - Spirtes, Glymour, et al. - 1990 |

18 | A History of Econometrics - Epstein - 1987 |

14 | Causal discovery from data in the presence of selection bias - COOPER, F - 1995 |

9 | Equivalence of causal models with latent variables - Spirtes, Verma - 1992 |

8 | Using dseparation to calculate zero partial correlations in linear models with correlated errors
- Spirtes, Richardson, et al.
- 1996
(Show Context)
Citation Context ...esents P(O). (For example in the linear case, the PAG can be given a complete orientation with all "o" ends removed, and interpreted as a linear structural equation models with correlated er=-=rors. See Spirtes, et al. 1996-=- for details.) Hence the PAG can be used to find an unbiased estimator of the population parameters that has lower variance than any unbiased estimator based on a DAG with the same set of variables. E... |

5 | Explanations for multivariate structures derived from univariate recursive regressions - Wermuth, Cox, et al. - 1994 |

2 | Building causal graphs from statistical data in the presence of latent variables - Spirtes - 1994 |