Those possible explanations translate straightforwardly into Bayes net fragments. If we had data, it would be pretty straightforward to actually infer the strength of each explanation.
How effective are those techniques in practice? Have any similar sociology/economic problems been convincingly resolved with Bayesian networks in the absence of randomization or instrumental variables?
How effective are those techniques in practice? Have any similar sociology/economic problems been convincingly resolved with Bayesian networks in the absence of randomization or instrumental variables?