This week at Scientific American’s Observations, reporter Katherine Harmon asks: Does science need more compelling stories to foster public trust?
This excellent question came my way, rephrased on Twitter, as this: “Should scientists fight misleading anecdotes w/ their own?”
The @SciAm piece starts like this:
The touching stories that advocacy groups are so good at telling—the 49-year old mother whose breast cancer was detected by an early mammogram before it had spread; the 60-year-old neighbor who had a prostate tumor removed thanks to a routine PSA test—should inspire scientists to use anecdotes of their own, argue two doctors from the University of Pennsylvania.
So we’ve got a battle metaphor on Twitter (scientists would ‘fight’) while doctors at a prestigious medical school advocate the use of anecdotes to “inspire” (i.e. manipulate) peoples’ views on information.
I agree with Katherine Harmon that the question of whether or not to use anecdotes in health and science reporting is crucial; these stories appeal to readers’ sympathy and influence their impressions of what’s true. But the underlying question is, how do you know what’s misleading? In other words, what if the scientists’ data are incorrect, outdated or misrepresented? Using the mammography example, what if screening is effective for women in their 40s?
The Observations piece is illustrated with a stock photo. It shows a truncated individual, wearing a white coat and a stethoscope around the neck, who holds up an unlabeled graph for display. Presumably this image is intended to represent science, the truth, or “facts.” But many sets of published data are seriously flawed, as any reader of Retraction Watch might tell you.
Anecdotes can work either way. So can data, depending on who interprets and frames the analysis, and who are the editors who publish new findings, or don’t.
Scientists and doctors wield significant authority. If they’re overconfident in their findings or ideas, they may cause harm to individuals or the public at large. If journalists are not sufficiently questioning of labels on a graph or other details of a reported study they may cause harm, just the same.
—-