Here we go again. A study finds an association between A and B, but some news reports say that A causes B. The public is misled, and trust is further eroded—not just in science, but also in journalism.
The study, “The ‘Twinkie Defense’: the relationship between carbonated non-diet soft drinks and violence perpetration among Boston high school students,” certainly merits coverage. Published in the journal Injury Prevention, it found an association between violent behavior and the amount of non-diet soda Boston teens had consumed More sugared soda was associated with a noticeably higher amount of violence. But an association between A and B doesn’t mean that A causes B. The researchers say it right in the abstract of their paper, the “lead” that many reporters don’t read beyond:
There may be a direct cause-and-effect relationship, perhaps due to the sugar or caffeine content of soft drinks, or there may be other factors, unaccounted for in our analyses, that cause both high soft drink consumption and aggression.
Somehow reporters ignored that crucial detail, so directly stated, in numerous stories, under headlines like, “Soda Boosts Violence Among Teens, Study Finds,” at The Washington Post’s Check Up blog; “Fizzy drinks make teenagers violent; Drinking five fizzy drinks a week makes teenagers more likely to act violently and carry a weapon, a study has found,” at The Telegraph’s Health News column; “Soda is Making Teens Violent and Aggressive: Study,” at The International Business Times; and my favorite, “Soda Totally Turns Teens Into Killers,” on a “news” website named Jezebel which describes itself as reporting on celebrity sex and fashion for women.
It wasn’t just the headlines. The stories under them all emphasized the worrying (news-making) connection between non-diet sodas and violence, and either didn’t include or buried the caveat that the study found an association that’s worth looking into, but did not find that sugary sodas cause violence.
GlobalPost ran a story under the headline, “Teen violence linked to soda consumption: study,” and, midway down, included a link to a YouTube video, “Girls arrested after fight,” enhancing the case for a connection between soda and violence, and diminishing the caveat that there may be no direct link at all. GlobalPost’s story did report mention the caveat about association and causation—in the last paragraph.
Likewise, in The Telegraph, the caveat was in the last paragraph of an eighteen-paragraph story. In the Washington Post piece, it was in the sixth of a seven-paragraph story. In the International Business Times it was in paragraph nine of eleven. To Jezebel’s credit, it came up in paragraph two of a four-graph story.
Why do journalists continue to confuse association and causation? Surely, some reporters or editors were not familiar with the distinction, but that’s no excuse, because the researchers explained it in the abstract. Maybe some journalists didn’t even read the abstract, let alone the whole paper. That gets you an F in any freshman journalism class.
More likely, reporters played up the dramatic association between soda and violence, and played down the distinction between association and causation, because it felt like a stronger, more dramatic, attention-getting story. But we don’t have to buy that excuse, either, because the study got a lot of coverage that made the distinction perfectly clear. Under the headline, “Soft Drinks Linked to Violent Tendencies in Teens: Study,” Newsday’s Amanda Gardner wrote:
Teens who drink lots of soda seem to be prone to violence, new research suggests. But the study authors concede that sodas are probably not the direct cause of the aggression.
While there’s a chance that the sugar and caffeine from carbonated drinks contributes to violent behavior, the study shows an association, not a cause-and-effect. Soda consumption, for example, may be a marker of heightened violent tendencies already present in the teen, or of poor parenting, the researchers said.
“Soda [could be] a red flag that is indicating something else is wrong,” said study co-author Sara Solnick, an associate professor of economics at the University of Vermont in Burlington.
AFP’s wire piece included the association-causation caveat in paragraph four, before describing details of the association found by the researchers. The Australian Broadcasting Corporation ran the caveat in paragraph five and repeated it in several later paragraphs. Compare Discovery’s cautiously open-ended headline, “Does Soda Cause Violence?” with the inflammatory ones inaccurately claiming causation.
The discrepancy in the coverage makes this a great teaching moment for health, science, and risk reporting. There’s an obvious tendency to play up the dramatic aspects of stories and play down or omit the tempering or balancing aspects, which does an injustice to accuracy.