Section 41 Upstream (Bayesian) inference

Is it enough to process information down the network? No. True, cause can by definition only flow down it. But we might have information which implies something about antecedents, which means we have to reason backwards. For example, we might know that the student failed to fulfill one of two conditions for entering a college, and we know they entered it, so we know they must have fulfilled the other condition.

When we have coded our causal arrows following the rules set out earlier, we can use this information to calculate specificity and sensitivity, so we can use Bayes’ rule to modify our prior information about the upstream variable(s).

Not implemented, but soon.

This way it will be possible to do Process Tracing with causal maps Tansey (2009),Befani and Stedman-Bryce (2017),Law (2005)