SIMPSON'S PARADOX IN ARTIFICIAL INTELLIGENCE AND IN REAL LIFE

“Simpson's paradox,” first described nearly a century ago, is an anomaly that sometimes arises from pooling data. Dramatic instances of the paradox have occurred in real life in the domains of epidemiology and admissions policies. Many writers have recently described hypothetical examples of the paradox arising in other areas of life and it seems possible that the paradox may occur frequently in mundane domains but with less serious implications. Thus, it is not surprising that the paradox should arise in commonsense reasoning, that subarea of artificial intelligence that seeks to axiomatize reasoning in such mundane domains. It arises as the problem “approximate proof by cases” and the question of whether to accept it may well depend on whether we wish to construct performance or competence models of reasoning. This article gives a brief history of the paradox and discusses its occurrence in our own discipline. It argues that if the paradox occurs frequently but undramatically in real life, every uncertain reasoning system will have to deal with the problem in some way.

[1]  C. Blyth.  On Simpson's Paradox and the Sure-Thing Principle , 1972 .

[2]  Raymond Reiter,et al.  A Logic for Default Reasoning , 1987, Artif. Intell..

[3]  John K. Ryan,et al.  An Introduction to Logic and Scientific Method , 1935 .

[4]  Randy Goebel,et al.  Theorist: A Logical Reasoning System for Defaults and Diagnosis , 1987 .

[5]  D. L. Hintzman.  On variability, Simpson's paradox, and the relation between recognition and recall: reply to Tulving and Flexser. , 1993, Psychological review.

[6]  David W. Etherington.  Formalizing Nonmonotonic Reasoning Systems , 1987, Artif. Intell..

[7]  Y. Mittal.  Homogeneity of Subpopulations and Simpson's Paradox , 1991 .

[8]  Ernest Nagel,et al.  An Introduction to Logic and Scientific Method , 1934, Nature.

[9]  S. Sunder.  Simpsons Reversal Paradox and Cost Allocation , 1983 .

[10]  Henry E. Kyburg,,et al.  The Reference Class , 1983, Philosophy of Science.

[11]  James P. Delgrande,et al.  A First-Order Conditional Logic for Prototypical Properties , 1987, Artif. Intell..

[12]  Eric Neufeld,et al.  A probabilistic commonsense reasoner , 1990, Int. J. Intell. Syst..

[13]  Scott Goodwin.  Statistically motivated defaults , 1991 .

[14]  C. E. Thomas,et al.  Comparing Basal Area Growth Rates in Repeated Inventories: Simpson's Paradox in Forestry , 1989 .

[15]  David Poole,et al.  A Logical Framework for Default Reasoning , 1988, Artif. Intell..

[16]  E. H. Simpson,et al.  The Interpretation of Interaction in Contingency Tables , 1951 .

[17]  Clifford H. Wagner.  Simpson's Paradox in Real Life , 1982 .

[18]  A. Lilienfeld.  Foundations of Epidemiology , 1980 .

[19]  Joseph Y. Halpern,et al.  From Statistics to Beliefs , 1992, AAAI.

[20]  Eric Neufeld,et al.  Conditioning on Disjunctive Knowledge: Simpson's Paradox in Default Logic , 1989, UAI.

[21]  John McCarthy,et al.  Applications of Circumscription to Formalizing Common Sense Knowledge , 1987, NMR.

[22]  Hector Geffner,et al.  On the Logic of Defaults , 1988, AAAI.

[23]  J. R. Brown,et al.  Aggregate efficiency measures and Simpson's Paradox* , 1992 .

[24]  P. Bickel,et al.  Sex Bias in Graduate Admissions: Data from Berkeley , 1975, Science.

[25]  Judea Pearl,et al.  An Algorithm for Deciding if a Set of Observed Independencies Has a Causal Explanation , 1992, UAI.

[26]  N. Cartwright.  Causal Laws and Effective Strategies , 1979 .

[27]  Judea Pearl,et al.  Probabilistic reasoning in intelligent systems , 1988 .

[28]  Fahiem Bacchus.  A Modest, but Semantically Well Founded, Inheritance Reasoner , 1989, IJCAI.

[29]  Graham Newson,et al.  Simpson's paradox revisited , 1991, The Mathematical Gazette.