Credible sources genuinely disagree about whether this link holds.
The definition the product enforces, read from the same code the methodology page renders.
Credible sources genuinely disagree about whether the link holds at all. It is a finding, not a weaker inference: the dispute over the mechanism is often the thing a decision turns on.
How a cause-and-effect link is KNOWN, kept apart from how much it matters. A documented link can be weak; a strong link can be inferred.
Every grade here is assigned by the same rules on every scan, and the methodology page renders those rules from the code that enforces them.