Sue Ellen Christian, the author of Overcoming Bias and one of three recipients of the 2016 Michigan Professor of the Year award, is interviewed in Encore Magazine and talks about how teaching and journalism have changed over the years.
by Sue Ellen Christian
This Nieman Storyboard piece is another take on bad journalism due to the confirmation bias. It includes the idea that editors may also suffer from confirmation bias. Sometimes they stick to the original story idea (often their own) at all costs.
by Sue Ellen Christian
The latest storytelling tool for journalists is 360-degree video. The medium allows a viewer to see all sides of a story, literally. The aptness of this metaphor for “Overcoming Bias” is too good to let pass without a brief reflection.
In November 2015, The New York Times made a virtual reality (VR) news splash with its documentary entitled “The Displaced,” about children driven from their homes due to war and persecution. Home subscribers like me received a free Google cardboard viewer with their paper that day.
More than a year earlier, in 2014, The Des Moines Register (with Gannett Digital) became the first mainstream news organization in the U.S. to use the technique of 360-degree video in its “Harvest of Change” series, which examined how Iowa farm families were dealing with demographic, economic and technological changes in America.
My capstone reporting class has begun to use the RICOH Theta 360-degree camera to shoot a series we’re calling “Work a Minute in Their Shoes.” It’s capturing 60 seconds of spherical video of various Kalamazoo workers on the job. What we are discovering is that only certain types of stories lend themselves to this form of immersive presentation. Breaking news of a public protest, for example, is a perfect event for this form of video, because there are many people to record from many vantage points. But a profile of a baker at work making bread in the early morning requires the camera to stay on her and follow her around the otherwise desolate bakery. In short, not all events require that you literally see all sides of the story.
This realization got me thinking about the bias of professional newsgathering to seek out both sides of the story. Sometimes, there isn’t a legitimate “other” point of view. For example, some partisan outlets still insist that President Obama is not legitimately a U.S. citizen. This question has been long settled by factual evidence in the form of Obama’s birth certificate. But the so-called “birthers movement” continues to champion its non-factual side of the story. Does that mean a reporter must trot it out each time he writes about Obama’s family background and/or ethnic origins?
In other instances, the story at hand is truly worth a 360-degree look. It may go beyond just two sides of the story, with multiple perspectives to capture. Chapter 5 of Overcoming Bias, Understanding Culture, Understanding Sources, discusses how social groups serve as lenses for looking at the world. Covering cultures different than one’s own requires a grasp of that culture’s history and customs, the context in which news events are considered because of the culture, and the unique perspectives of members within the culture.
Were we able to bust through the snowglobe effect of VR, were we able to bust through the artificial ceiling created by the camera’s many lenses, we would find yet more versions of truth that are influenced by culture, from religion to geography to language to income.
Immersive video has been called the future of journalistic storytelling. I hope journalists of every medium will keep this wonderful metaphor of capturing many perspectives at the fore of their thinking.
by Sue Ellen Christian
The Rolling Stone story “A Rape on Campus,” about an alleged gang rape at the University of Virginia’s Phi Kappa Psi fraternity house, is another journalistic example of confirmation bias at work. This piece offers much to discuss in a classroom regarding ethical and reporting errors.
The people involved in the story – reporter Sabrina Rubin Erdely, the fact checker, and the editors – dismissed important reporting steps, apparently because the story was just too good to challenge. Later, at the request of Rolling Stone, a team led by Steve Coll, a former Pulitzer Prize-winning reporter and now dean of the Columbia School of Journalism, undertook an analysis of the reporting and editing missteps.
Al Tompkins of Poynter wrote that confirmation bias seems to have been a factor in this story. “Erdely believed the university was obstructing justice. She felt she had been blocked. Like many other universities, UVA had a flawed record of managing sexual assault cases. Jackie’s experience seemed to confirm this larger pattern. Her story seemed well established on campus, repeated and accepted.”
Chapter 7 of Overcoming Bias includes a list of what reporters succumbing to the confirmation bias might do. Here, I apply the list to the reporting on the U-V case as explained by the Columbia report:
Confirmation bias will cause a reporter to remember information that confirms the hypothesis or story angle better than he/she remembers information that disconfirms it.
Erdely didn’t do enough reporting in order to find information that disconfirmed her angle. But it is important to remember that sources also have this cognitive bias – they remember best the information that fits into their mental picture of how things were supposed to play out or how they wanted them to play out. So reporters should seek corroboration with sources’ memories of events.
We all remember selectively and in a biased way, so plan for that.
Confirmation bias causes reporters to seek out only sources that they believe will support the hypothesis.
Erdely was concerned about protecting her primary source, Jackie, so she did not vigorously pursue the sources that may have countered Jackie’s version of events – specifically, the alleged perpetrators. The alleged ringleader, “Drew,” whom Jackie said invited her to the fraternity party, for example, was a critical source to actually interview and name (the story used a pseudonym).
As Poynter’s Tompkins rightly noted, reporting on sensitive topics doesn’t mean that source accounts shouldn’t be fully vetted. Tompkins wrote, “Don’t let sensitivity stop the reporting in rape cases.”
Confirmation bias causes reporters to treat sources critical of the hypothesis in an aggressive or argumentative way.
In this case, the aggression was passive: When seeking comment from the Phi Kappa Psi fraternity, Erdely didn’t fully disclose to her source the details of what she would write about them. The fraternity didn’t have enough information to fully respond to the story she was writing. Additionally, the three friends quoted in the story were not contacted by Erdely, who relied on Jackie’s account of their unsupportive responses to her attack.
Confirmation bias causes reporters to interpret unanswered questions in a way that infers hypothesis confirmation.
When Erdely asked Jackie to give her Drew’s last name and perhaps a way to contact him, Jackie became upset, according to Erdely. Rather than considering whether or not this reaction might be hiding a nonexistent aggressor, Erdely took it as in keeping with the emotions expected of a victim.
The Rolling Stone story presents the fraternity members as planning and conspiring to execute the attack: Seven men waiting in a dark room to which Jackie was led by her date. That, noted Washington Post’s Erik Wemple, was an unanswered question of enormous magnitude: “If this allegation alone hadn’t triggered an all-out scramble at Rolling Stone for more corroboration, nothing would have . . . The ‘grooming’ anecdote indicates not only that Erdely believed whatever diabolical things about these frat guys [that were] told her, she wanted to believe them.”
Confirmation bias causes reporters to give the comments of sources supporting the hypothesis more emphasis, space or more prominence in a story.
Jackie’s account of the attack is all readers know. Corroborating evidence such as a police report, hospital documentation or documents from a university investigation (or lack of any of these) are not part of the story. Jackie is the prominent voice throughout.
Confirmation bias causes reporters to evaluate sources and information confirming the hypothesis less critically.
Sean Woods, the principle editor on the story, was quoted in the Columbia report as saying: “Ultimately, we were too deferential to our rape victim; we honored too many of her requests in our reporting,” Woods said. “We should have been much tougher, and in not doing that, we maybe did her a disservice.”
Erdely told the Columbia Journalism School team that, “If this story was going to be about Jackie, I can’t think of many things that we would have been able to do differently. … Maybe the discussion should not have been so much about how to accommodate her but should have been about whether she would be in this story at all.”
Confirmation bias causes reporters to regard disconfirming information as unusual, superficial or of poor quality.
Jackie didn’t produce the bloodstained red dress she said she had worn on the night she said she was attacked. The messages that Erdely left at the number that Jackie gave her to reach her mother received no response. Those are just two examples of information that was missing that should have signaled a problem to the Rolling Stone reporter and her editors.
Perhaps the best lessons out of this failed piece, which raises real concerns about its consequences on the willingness of rape victims to publicly share their stories, are the classroom lessons. Let’s hope that by reviewing the missteps in this reporting, all future such missteps remain purely academic.
A shooting happens in a neighborhood. Many things happen right after that initial event, including an adult version of the child’s game of telephone―the happening gets told and retold, and the retelling gets retold.
At some point in this oral storytelling cycle, a journalist arrives and starts interviewing witnesses.
While the journalist tries hard to get as close to the eyewitnesses of the shooting as he or she can, direct access is not always possible. And, as new research suggests, and as good journalists know, once the actual facts have been whispered around the neighborhood, they are likely to emerge quite different from the original version.
Researchers call this phenomenon “third party contagion.” A study by Tiane L. Lee, Michele J. Gelfand and Yoshihisa Kashima documented “the role played by third party observers as drivers of conflict escalation through their biased communication.” Put more simply: People who have nothing directly to do with a conflict―in my hypothetical, the shooting―often muddy everyone’s understanding of the event by describing what happened in loaded terms. They offer evaluations not based on facts. Utter judgments that suit their purposes. Apply moral dimensions that aren’t there in the first place. And, of course, assign cause.
In the Lee et al. study using about 200 undergraduates, the first person in a four-person chain learned about a made-up conflict between two groups. (The original version of the conflict assigned equal fault to the disputing parties.) The first person then retold the narrative by typing up what he or she remembered about the event, and that version was passed along to the next participant in the chain. In the end, participants completed a survey about the conflict and its involved parties.
The results? “Partisan spectators to others’ disputes not only become involved in, but escalate, the dispute to a multitude of others,” states the resulting article, “The serial reproduction of conflict: Third parties escalate conflict through communication biases,” in the Journal of Experimental Social Psychology (vol. 54, Sept. 2014).
Even people without direct experience in a conflict tend to fuel the conflict by framing it in a morally biased way, giving causes for the conflict, providing judgments of the disputing parties, and seeking revenge. They finesse the tale to fit their purposes. Everyone seeks to add his or her two cents’ worth to the tale.
The researchers sum up their findings this way: “Using the serial reproduction method, we demonstrate the role of third parties’ communication biases in conflict escalation, revealing that successive generations of partisan observers share and reproduce conflict narratives that become increasingly biased in their moral framing, attributions for the conflict, evaluations of the disputing parties, and quest for revenge. Despite equal fault between the disputing parties at the beginning, these communication biases increased, rather than subsided, with each iteration throughout communication chains, cumulating in distortions and group biases far above and beyond initial ingroup favoritism.” (See Chapter 5, “Understanding Culture, Understanding Sources,” of Overcoming Bias for more on how ingroup favoritism influences interviewing.)
Why do people do this? The research suggests that people pass along distorted information because they want to create a common narrative, “a collective identity.”
Journalists, beware. Such striving for a “shared reality” involves more imagination and opinion than fact.