Three conditions under which experiments and observational studies produce comparable causal estimates: New findings from within‐study comparisons | Synapse