Put the answers beside the questions
Imagine two fictional reviews of a program. One asks whether it improves a short questionnaire score after four weeks. The other asks whether it improves daily function after a year. “Promising” in the first review and “not established” in the second are compatible statements.
Before deciding that experts disagree, match their question, comparison group, outcome, and time period. Sometimes the apparent contradiction disappears at that stage.
Agreement about effects can still leave a choice
Suppose both reviewers agree that a program has a small average benefit, takes several hours a week, and has uncertain long-term effects. One sees the time commitment as acceptable. Another considers it burdensome. That difference is partly about the value assigned to the benefit and the burden.
Calling one position “scientific” and the other “unscientific” would obscure the shared evidence and the actual point of disagreement. It is more precise to say which tradeoff changed the conclusion.
When the evidence itself is disputed
Reviewers may instead differ over whether a study has serious bias or whether several studies are comparable enough to combine. Those are methodological disagreements. They deserve reasons that can be inspected, rather than a vote based on titles or confidence of delivery.
Cochrane's guidance on interpreting results separates effects, applicability, and certainty. That separation helps readers locate a disagreement rather than treating every difference as proof that nothing is knowable.
A question that moves the discussion forward
“What evidence or assumption would need to change for these conclusions to converge?” can reveal whether the unresolved issue is a missing study, a different population, or a different preference. Some disagreements can be reduced with more data. Others remain reasonable choices between outcomes that people value differently.
