What is CAT?
CAT (the Causal Attribution Tool) is a tool for exploring different causal scenarios incorporating uncertainties and allowing you to entertain and test different hypotheses about what is causing what in those scenarios. CAT will also allow you to compare alternative ideas about the nature of causal attribution itself; that is, distinct criteria for judging causal attribution can (and will) be implemented – although we have implemented our own preferred criterion first, naturally, along with a couple of popular alternatives.
Load a Causal BN
My Causal BNs
Public Library of Causal BNs
Tenure
The Tenure network is a very simple BN that illustrates the basic concepts of CAT. This causal model induces a positive correlation between having White Hair and getting Tenure: the common cause of Age/Time sets up such a probabilistic dependency.
Coronary Risk
A modified version of the model developed by Assessment Technologies, Inc. Estimates the 10 year likelihood of developing coronary artery disease [CAD], as a function of accepted coronary disease risk factors. Modified to remove non-causal factor. Original model at norsys.com.
Simpson's Paradox
Simpson's Paradox is named after a statistical oddity described in Simpson, E. H., 1951, “The Interpretation of Interaction in Contingency Tables”, Journal of the Royal Statistical Society: Series B (Methodological), 13(2): 238–241. Namely, he demonstrated with the example of this CBN that an overall positive, negative or neutral association between two variables (here Treatment and Recovery) could be reversed given an association with a third variable (Sex). In consequence, the average effect size of a cause should not in general be trusted, unless the presence of confounders can actually be ruled out. The effects given particular values of the cause are more to the point. In this example, the overall effect of Treatment on Recovery is practically nil, even though the power of Treatment to improve Recovery rates is clear for both men and women taken separately. It is commonly called Simpson's Paradox, although the issue was raised earlier by Karl Pearson and George Yule. For discussion, see the nice overview article by Sprenger and Weinberger: Sprenger, Jan and Naftali Weinberger, "Simpson’s Paradox", The Stanford Encyclopedia of Philosophy (Summer 2021 Edition), Edward N. Zalta (ed.), URL = <https://plato.stanford.edu/archives/sum2021/entries/paradox-simpson/>.