An association is a statistical link or pattern between two variables, but an effect can only result if one is actually having an impact on the other. Effect should only be used when causation is involved.
Deeper dive
Perhaps the first and most important axiom of covering medical studies is that correlation is not causation. Just because two things are linked, that doesn’t mean one of them caused the other or vice versa. Most of us learned this long before becoming journalists, though delving into coverage of medical research requires it to be a constant mantra in our minds. And yet, even journalists who might unconsciously chant “correlation is not causation” in their dreams can occasionally fall into the trap of referring to an outcome as an “effect” of an exposure it’s associated with.
Simply using the word “effect,” however, implies that the exposure was the cause, and whether the reporter intended that meaning or not, it’s inevitably what the reader will take home from the story. It’s an easy trap to fall into because “effect” is a simple, succinct, easy-to-understand word already embedded in the brain as connected to health—which means it often shows up in headlines where it can do the most damage by conveying to readers a message that isn’t supported by the evidence.
An association means only that two (or more) things follow a pattern of interaction such that a change in one results in a proportional change in the other. A positive association means one increases as the other increases (or decreases as the other decreases), and a negative (or inverse) association means one increases as the other decreases or vice versa. The only responsible synonyms for “associated with” are words that similarly imply only an interaction independent of cause and effect, such as “linked,” “correlated with,” “related to,” etc. Unfortunately, even some researchers make this mistake, typically those with an ideological bias in their work that they don’t recognize. They will use the word “effect” when the evidence doesn’t (at least yet) support a cause-and-effect relationship. If it seems necessary, a reporter could ask for clarification when a researcher uses “effect” in an interview by asking whether additional evidence exists that would support a cause/effect relationship. In other cases, the best route may be using quotes that don’t include “effect.”
Occasionally, there may be sufficient mounting evidence that meets criteria for establishing cause and effect from a large evidence base of observational studies, such as the link between smoking and cancer. Using “effect” in studies showing associations may be acceptable there, but it’s walking a fine line unless there is already a solidly established consensus.