They conflict only if those attributes match. Suppose a diet trial estimates assignment regardless of later adherence, while another estimates the effect under continued adherence: those are different treatment-effect questions, even with the same outcome and population.
0 karmaView comment