Validating Causal Estimates In Experimental And Observational Designs