## An Anytime Algorithm for Causal Inference (2001)

Venue: | in the Presence of Latent Variables and Selection Bias in Computation, Causation and Discovery |

Citations: | 10 - 1 self |

### BibTeX

@INPROCEEDINGS{Spirtes01ananytime,

author = {Peter Spirtes},

title = {An Anytime Algorithm for Causal Inference},

booktitle = {in the Presence of Latent Variables and Selection Bias in Computation, Causation and Discovery},

year = {2001},

pages = {121--128},

publisher = {MIT Press}

}

### OpenURL

### Abstract

The Fast Casual Inference (FCI) algorithm searches for features common to observationally equivalent sets of causal directed acyclic graphs. It is correct in the large sample limit with probability one even if there is a possibility of hidden variables and selection bias. In the worst case, the number of conditional independence tests performed by the algorithm grows exponentially with the number of variables in the data set. This affects both the speed of the algorithm and the accuracy of the algorithm on small samples, because tests of independence conditional on large numbers of variables have very low power. In this paper, I prove that the FCI algorithm can be interrupted at any stage and asked for output. The output from the interrupted algorithm is still correct with probability one in the large sample limit, although possibly less informative (in the sense that it answers "Can't tell" for a larger number of questions) than if the FCI algorithm had been allow...

### Citations

7414 |
Probabilistic reasoning in intelligent systems: Networks of plausible inference
- Pearl
- 1988
(Show Context)
Citation Context ...elationship, named d-separation, among three disjoint sets of vertices, which allows all of the conditional independence relations entailed by the Causal Markov Principle to be read off of the graph (=-=Pearl, 1988-=-, Lauritzen et al. 1990). The definition of d-separation is contained in the Appendix. For the purposes of this article, the important point is that there is a purely graphical relation “d-separation”... |

1235 | Causality: Models, reasoning, and inference - Pearl - 2000 |

1174 | Structural equations with latent variables - Bollen - 1989 |

1151 | Graphical Models - Lauritzen - 1996 |

994 | Multivariate Analysis - Mardia, Kent, et al. - 1979 |

891 | A tutorial on learning with Bayesian networks
- Heckerman
- 1998
(Show Context)
Citation Context ...omputationally much more expensive than calculating the ratio of posterior probabilities for two causal DAGs without latent variables. Various approximations can be used to simplify the calculations (=-=Heckerman 1998-=-). Heckerman (1998) describes a Bayesian search over a number of different latent variable models. The best model that they found for the College Plans data is shown in Figure 11, where H is a latent ... |

662 | Probabilistic Networks and Expert Systems - Cowell, Dawid, et al. - 1999 |

652 | Learning in graphical models - Jordan - 1998 |

634 | Bayesian Networks and Decision Graphs - Jensen - 2001 |

633 | Structural equation modeling in practice: A review and recommended two-step approach - Anderson, Gerbing - 1988 |

530 |
Causation, Prediction and Search
- Spirtes, Glymour, et al.
- 2001
(Show Context)
Citation Context ...onds roughly to changing a probability distribution in response to changing the state of the world in a specified way (or doing). (For more on the difference between conditioning and manipulating see =-=Spirtes et al. 2000-=- and Pearl 2000). To illustrate the difference, we will consider a very simple example in which our pre-theoretic intuitions about causation are quite strong and uncontroversial. Consider a population... |

481 | Machine learning - Mitchell - 1997 |

461 | Graphical Models in Applied Multivariate Statistics - Whittaker - 1990 |

275 | Latent variable models and factor analysis - Bartholomew - 1987 |

255 | General Intelligence": Objectively determined and measured - Spearman - 1904 |

223 |
Equivalence and synthesis of causal models
- Verma, Pearl
- 1990
(Show Context)
Citation Context ... same BIC scores. Such models are guaranteed to exist if there are models that have the same number of degrees of freedom and contain graphs that are distribution equivalent to each other. Theorem 1 (=-=Verma and Pearl 1990-=-, Spirtes et al. 2000) shows how distribution equivalence can be calculated in O(E 2 ) time, where E is the number of edges in a path diagram. X is an unshielded collider in directed acyclic graph G i... |

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

183 | A new approach to causal inference in mortality studies with sustained exposure periods - Application to control of the healthy worker survivor effect - Robins - 1986 |

171 | Assignment to treatment group on the basis of a covariate - Rubin - 1977 |

151 | Nature’s Capacities and Their Measurements - Cartwright - 2002 |

151 | Probabilistic Reasoning in Expert Systems - Neapolitan - 1990 |

149 | Factor analysis a a statistical method - Lawley, Maxwell - 1971 |

147 | Independence properties of directed Markov fields - Lauritzen, Dawid, et al. - 1990 |

111 |
Testing structural equation models
- Bollen, Long
- 1993
(Show Context)
Citation Context ...Equivalence Consider the College Plans example. There are a variety of ways of scoring such a discrete model, which include p(χ 2 ), and the BIC or Bayes Information Criterion (see section 5.2.7, and =-=Bollen and Long 1993-=-.) In order to evaluate how well the data supports this causal model, it is necessary to know whether or not there are other causal models compatible with background knowledge that fit the data equall... |

103 | Mathematical statistics: basic ideas and selected topics. Holden-Day - Bickel, Doksum - 1977 |

95 | A characterization of Markov equivalence classes for acyclic digraphs - Andersson, Madigan, et al. - 1997 |

94 | A transformational characterization of equivalent Bayesian network structures - Chickering - 1995 |

88 | The Art of Causal Conjecture - SHAFER - 1996 |

84 | A bayesian approach to causal discovery - Heckerman, Meek, et al. - 1997 |

83 | Causal inference and causal explanation with background knowledge - Meek - 1995 |

79 | Causal ordering and identifiability - Simon - 1953 |

79 | The methods of path coefficients - Wright - 1934 |

76 |
Discovering Causal Structure
- Glymour, Scheines, et al.
- 1987
(Show Context)
Citation Context ... in Spearman (1904). It is easy to calculate which vanishing tetrad constraints are entailed by a given DAG Hence these constraints can be useful in searching for causal models with latent variables (=-=Glymour, et al, 1987-=-, Spirtes et al., 2000) as described further in section 6.2.7.4. However, the existence of many kinds of non-conditional independence constraints entailed by a DAG (possibly together with a distributi... |

70 | The dappled world: A study of the boundaries of science - Cartwright - 1999 |

69 | Structural Equation Modeling: Foundations and Extensions, number 10 - Kaplan - 2000 |

67 | Probabilistic Causality - EELLS - 1991 |

63 | Probabilistic Evaluation of Sequential Plans from Causal Models with Hidden Variables.” Pp. 444–53 - Pearl, Robins - 1995 |

61 | The foundations of factor analysis - Mulaik - 1972 |

60 | Causal inference from graphical models - Lauritzen - 2001 |

59 | Impulse response functions based on a causal approach to residual orthogonalization in vector autoregressions - Swanson, Granger - 1997 |

52 | Chain graph models and their causal interpretations - Lauritzen, Richardson - 2002 |

50 | Some methods for respecifying measurement models to obtain unidimensional construct measurement - Anderson, Gerbing - 1982 |

50 | Spurious Correlation: A Causal Interpretation - Simon - 1954 |

47 | Causal Asymmetries - Hausman - 1998 |

47 | The problem of equivalent models in applications of covariance structure analysis - MacCallum, Wegener, et al. - 1993 |

44 | Computation, causation, and discovery - Glymour, Cooper - 1999 |

44 | Causal analysis: Assumptions, models, and data - James, Mulaik, et al. - 1982 |

43 | Recursive versus nonrecursive systems: An attempt at synthesis - Strotz, Wold - 1985 |

42 | Directed cyclic graphical representation of feedback models - Spirtes - 1995 |

37 | Causality in Macroeconomics - Hoover - 2001 |