One of the most well-regarded science publishers in the world has walked back the main takeaway from a study that claimed smartphones are causing millennials to grow horns.
That claim was parroted by scores of news outlets and likely consumed by millions of people when it became a viral headline in June. Wednesday’s correction from Nature Research — which started with the words, “This Article contained errors” — completely reverses the authors’ main takeaway about so-called neck horns.
As the PBS NewsHour previously reported, the term “horns” was itself a misnomer for what the researchers had actually studied: bone spurs growing on the back of the neck called enlarged external occipital protuberances — EEOPs or EOPs. Under the new correction, the study authors no longer claim that mobile technology of any kind causes these neck bone spurs.
Instead, the report points to more data to back up its assertion that posture and age are linked to the bone spurs. Nature Research provided this statement to the PBS NewsHour regarding the correction in its journal Scientific Reports:
When we became aware of criticisms of this paper, we carefully investigated the concerns raised following an established process. This further assessment of the manuscript, which took additional information into account, revealed that the methodology and data remained valid. It was determined, however, that the paper should be corrected to more accurately represent the study and the conclusions that could be drawn from it. We have also updated the competing interests statement for the paper.
When we first covered this study, which was published by chiropractor David Shahar, who also sells posture devices, and sports biomechanicist Mark Sayers of the University of the Sunshine Coast in Australia, it wasn’t just the smartphone angle that seemed problematic. We outlined six basic flaws in how the researchers reached their conclusions that cast doubt on the whole study.
“We are satisfied with the changes. We wrote and approved them to clarify a few points, so that they are not subject to misinterpretation,” Sayers and Shahar said in a statement. “Our novel findings were not changed, and these could have implications for people around the world.”
While the correction attempts to clarify the record on smartphones, it does not do anything to address the fact that the study’s main finding — that poor posture and age are connected to neck bone spurs — still isn’t supported by the underlying data, which Shahar and Sayers now present publicly for the first time. That’s according to researchers in bone development and statistics who reviewed the original study and the correction.
“I actually think Nature should remove the original article as the correction has not proved their point,” said Sara Becker, a bioarchaeologist at the University of California Riverside.
Another catch: The misinformation in the original study and subsequent news coverage already went viral. And often, “false reports get much more attention than later fact checks of those claims,” said Brendan Nyhan, a political scientist at Dartmouth College who studies how misperceptions in health care and politics spread.
In any attempt to reverse the misinformation of the horns study, Nyhan predicts that the challenge “will not be convincing people to change their minds. It will be providing accurate information to all the people who are exposed in the first place.”
We asked experts what this correction means from a health perspective, and whether anything can be done to fix the public’s perception after a misleading study has gone viral.
Why correcting viral misinformation is so difficult
Shahar and Sayers first published this research in 2018, but it didn’t rise to international fame until June of this year, after BBC Future featured Shahar and Sayers’ study in a broad review about how modern life is transforming the human skeleton. A week later, The Washington Post published a story focused specifically on Shahar and Sayers’ work, with the headline “Horns are growing on young people’s skulls. Phone use is to blame, research suggests.”
On social media, the story spread widely. Data from CrowdTangle, a social media monitoring platform, shows that the Washington Post story alone has garnered 300,000 interactions on Facebook, including 95,000 shares, 145,000 reactions (a.k.a. the number of times people hit the “like” button) and 69,000 comments.
Among the Facebook pages, public Facebook groups, Twitter accounts and subreddits tracked by CrowdTangle, more than 30 million followers could have been exposed to the Washington Post story. Search for the story link on Twitter, and you’ll find some of the most popular journalists and editors in the biz shared it.
And those numbers don’t account for the exposure generated by secondary news coverage. The study’s original premise landed everywhere from morning talk shows to public radio quiz shows. Local TV news and city papers covered it. Fox News and NBC had doctors explain what to do about it — the latter even published a guide on how to prevent the neck spurs from happening.
In their statement, Shahar and Sayers said “because of the international coverage, the study was accessed more than 107,000 times at its source.”
All of this coverage, conceived in the spirit of warning people about a health risk, was based on a false premise.
The PBS NewsHour contacted the BBC and The Washington Post for comment on Nature Research’s correction. The Post declined to comment, but the BBC shared the following: “The BBC used a study published in the journal Scientific Reports as a jumping-off point for a wider discussion osteobiography, of which Dr. Shahar’s work was one example of many. If Scientific Reports and Dr. Shahar have now decided to re-evaluate those findings, we will as well.”
Why this correction is unlikely to fix public misperception
Considering this deep penetration across the internet, is it possible to reverse and replace the misperceptions created by the original study and its media coverage?
“No. Nope. Unfortunately no,” said Jennifer Stromer-Galley, a professor of information studies and director at the Center for Computational and Data Science at Syracuse University.
Stromer-Galley said the shortfalls of issuing corrections are as old as print media, but are especially apparent in the modern era of journalism. Research on the topic stretches back decades, and the emergence of the internet and social media offers stark new evidence of how difficult it is to replace misperceptions.
She added while most journalists value a factual recording of a day’s events, they can be hesitant, except in the most egregious cases, to take down a story with a major mistake — and instead rely on a correction that most people don’t see.
A 2014 study of New York Times stories, for example, found 99 percent of the paper’s corrections involved objective errors — with people’s names, titles, numbers and dates — rather than subjective changes, which may arise when a journalist recounts an event out of context.
There are many challenges in combating the spread of false information. But two big, consistent ones are a story’s reach and people’s emotions.
How far falsehoods spread
This may sound obvious, but lies spread faster and farther than the truth. A 2018 study showed the most popular false news on Twitter routinely reaches between 1,000 to 100,000 people — while truthful news rarely if ever lands in 1,000 feeds.
“You get people in the social network who reshare the misinformation, and then resharing cascades until it has touched such a large number of people that we could say a ‘misinformation event’ has happened,” Stromer-Galley said.
Don’t be too quick to blame this entirely on social media algorithms. Part of the problem lies with how news is produced and consumed. Stromer-Galley said news outlets don’t typically promote a correction with the same energy as an initial story. For example, as of early Wednesday, the Washington Post’s original posts on Facebook and Twitter (and a follow-up op-ed) about the horns story don’t mention any kind of correction, despite the newspaper adding corrections to the original article months ago.
Adam Marcus — co-author of Retraction Watch blog — said the same rule applies to academic journals, which issue about 1,000 retractions per year. He added that there are 10 times as many corrections published per year, but very few are promoted by journals. Nature Research only shared its horns study correction with Retraction Watch and the PBS NewsHour before it was released, the publication said, even though many more news organizations covered the original story.
“Journals historically have not done a very good job of promoting a retraction,” said Marcus, whose blog serves as the premier outlet for reports about retractions and corrections in research papers. “They might issue a press release about a study, and then when the retraction comes, they don’t issue a press release about the retraction.”
Even after a correction happens, will we accept it?
Nyhan has conducted multiple studies about how misinformation spread during the 2016 election, and he has found “people who were exposed to fake news were very unlikely to be exposed to a corresponding fact check of the dubious information.” He calls this phenomenon “a mismatch problem.”
While this is easy to observe in everyday life, it’s hard to explain exactly why it happens. For example, Nyhan has evidence showing only one in four Americans consumed a reputable fact check in the lead-up to the 2016 election, suggesting debunks are not automatically sought by the general public. And even then, only 63 percent of that small group had a favorable view of those debunks. Moreover, most people who consistently consume fake news say they do not like fact checks in general.
This distaste for corrections likely speaks to the basic way our brains work, and our emotional ties to what we want to believe.
“When we encounter information that is contrary to a story that we have about the world, we tend to discount it. That’s confirmation bias,” Stromer-Galley said. “It’s quite hard to dislodge a belief that someone holds.”
Nyhan said this emotional defensiveness tends to form around people’s strongest beliefs, such as politics and religion. Stromer-Galley has found negative, evocative messages about a political opponent are substantially more likely to spread than messages that are positive or neutral.
These forces could also act as barriers to correcting misperceptions about something like the horns study, Stromer-Galley said.
“[The original study] trades on our assumptions and fears about cellphone use and our worries about ‘what is this technology doing to us?’” she said. “It feeds into those places [in our minds] where we have some worry or concern but not a lot of information. That is a good place for misinformation to spread.”
What the “horns” correction fixed — and what it didn’t
In addition to the problems of virality and bias, and the limited scope of the correction issued Wednesday by Nature Research raises questions.
One of the most obvious blemishes about the research — “the study doesn’t actually measure cellphone usage” and “the study isn’t studying horns” — would be extremely difficult for the scientists to fix. It would mean going back to recruit the same study subjects, pulled from a database of chiropractic patients, and assessing their cellphone usage. That’s a massive endeavor. (We contacted the authors for comment to see if they tried this, but did not receive a reply). With regards to the other biggest failing, “horns” aren’t made of bone, so such a label is simply inaccurate.
Even the esoteric flaws about the study’s statistics are not fixed by the correction.
“I wasn’t able to reproduce their results from the information and data given,” said Regina Nuzzo, senior advisor for statistics communication and media innovation at the American Statistical Association.
Jeff Goldsmith, a biostatistician at Columbia University, ran into the same hiccup when he crunched the numbers offered in the correction.
“I’m close. I’m in the same ballpark but there are some choices made in the data processing that at least right now I don’t completely understand,” Goldsmith said.
Goldsmith said one warning sign centers around the study’s continued claim that bad posture — defined as forward head protraction or FHP in the study — has a relationship to the bone spurs at all. He said among the variables Shahar and Sayers examined, FHP was one of the least predictive for these bone spurs.
“If I took anything away from this, it would not be that posture is bad. It would be that being male and being older increases your risk,” Goldsmith said. “The relationship to FHP was statistically significant, but not substantial enough to be the kind of thing that a person would worry about.”
As we reported in our original piece, the most powerful statistical tools for pulling out trends among multiple variables as they change over time are things like a regression analysis. With that test, a scientist could plug in their data about gender, age and posture, and reveal whether a significant link to neck bone spurs existed.
While the journal’s correction attempts a new logistic regression analysis on the study data, it doesn’t present the results in their entirety, Nuzzo said — only those for gender and FHP. She guesses “that’s because age was not a significant predictor,” even though it is still presented that way.
“They didn’t find it in the most powerful robust analysis, which is the logistic regression,” Nuzzo said. “So you can’t then say, ‘Alright, well, let’s try this other way of slicing and dicing the data and see if we get something.’
But that’s exactly what the research team did, Nuzzo and Goldsmith said. The study still arbitrarily groups subjects by age — 18 to 29, 30s, 40s, 50s and then everyone above 60 — and still relies on a less powerful test — Chi-squared — for making multiple comparisons. Except with the correction, it now argues that both millennials and senior citizens experience more of these neck bone spurs.
Goldsmith spotted another, more basic concern with this method.
“The authors controlled how many people were in these different age groups. And they had more people in the 18 to 29 group and the 60-plus group than they do in their groups for 30s, 40s and 50s,” Goldsmith said.
As a result, seeing a higher number of bone spurs cases in the oldest and youngest age groups is almost guaranteed by the way they sampled, he said.
So … do you need to worry about bone spurs growing on your neck?
The study’s findings still don’t mean anything for the general public, given the investigation looked at a subset of chiropractic patients. Bioarchaeologists and bioanthropologists who study the evolution of bone development expressed their appreciation for the attempt at a correction, but felt its clarifications didn’t go far enough.
John Hawks, a paleoanthropologist who specializes in bone evolution, said you can’t really blame a biological change on a single technology. Craning our necks is a behavior that humans have done for hundreds of thousands of years — not just to look at our phones. He isn’t suggesting that technology has zero effect on humans. Shoes, for instance, changed the way we walk. But he said these biological changes are subtle and typically due to a range of new behaviors.
“The idea that a new technology is going to suddenly cause people to really shift in a noticeable way is more like a ‘Doctor Who’ plot,” said Hawks, who added that the X-rays for this study “weren’t collected for the purpose of demonstrating that humans were changing over time.”
Sara Becker, the bioarchaeologist at the University of California Riverside, wondered: If these neck spurs are a real thing that happen among normal, healthy people, why don’t they exist in South America where she conducts research? There, she sees people hunch over to perform traditional tasks — weaving, sewing, knitting, ceramics production, farming — for hours.
“When we ask about pains, no one really complains about the back of the head or neck – instead, we get a lot of back problems and maybe shoulder pains,” Becker said. “These are people who started performing these tasks as children, as young as 7 or 8.”
If a person still believes that smartphones are causing these so-called “horns,” they might ask themselves if they know any sufferers who sew, embroider, read books or do any other tasks that require constant looking down.
“My guess would be they do not,” Becker said, “because the correlation is not there.”
Dan Cooney and Claire Mufson contributed to reporting this story.
Editor’s note: This story has been updated with responses from The Washington Post, David Shahar and Mark Sayers. It has also changed “Dartmouth University” to “Dartmouth College.”