The Case for Evaluating Causal Models Using Interventional Measures and Empirical Data