Pool effect sizes across studies only when the studies are genuinely comparable; otherwise plot the individual estimates and state why they cannot be combined. Use whenever an agent wants to produce a single summary number from multiple studies.
