Yes, journalists show more (cognitive) bias on Twitter
Editor’s note: Longtime Nieman Lab readers know the bylines of Mark Coddington and Seth Lewis. Mark wrote the weekly This Week in Review column for us from 2010 to 2014; Seth’s written for us off and on since 2010. Together they’ve launched a monthly newsletter on recent academic research around journalism. It’s called RQ1 and we’re happy to bring each issue to you here at Nieman Lab.
Campaign reporters’ cognitive shortcuts on Twitter
One of the most fruitful areas of psychological research over the past several decades has been in the heuristics and biases that serve as cognitive shortcuts for people attempting to evaluate situations and make decisions. The concepts in this area have taken on several different terms — dual processing, heuristics, central and peripheral processing, System 1 and System 2 — but they all illustrate a similar idea.
Our brains have two different modes of thought: One is quick, relatively low-effort, and relies heavily on emotions, habits, and simple shortcuts to make judgments. The other is slower, more deliberate, and relies on more complex analytical processes. (Following Daniel Kahneman, we’ll call the former System 1 and the latter System 2.) Both are prone to cognitive biases like the desire to confirm pre-existing beliefs, but System 1 thinking is much more susceptible to those biases. And being cognitive misers who need to make countless judgments each day, we rely heavily on System 1 for our everyday cognition.
Journalists aren’t any different, and there are studies going back three decades indicating that their cognitive biases affect the way they determine story angles, sources, headlines, and incorporate contradictory information. A new study by Stanford’s Jihye Lee and James T. Hamilton in the prominent scientific journal PLOS ONE tries to deepen our understanding of cognitive bias in journalism by examining journalists’ tweets, news articles, and broadcast transcripts for evidence of it in the language they use.
Looking at journalists covering the 2016 U.S. presidential campaign, Lee and Hamilton posit that these journalists’ propensity to use System 1 thinking (thanks to tight deadlines and hectic campaigns) is heightened on Twitter, with its relentless speed and emphasis on top-of-mind thoughts. To test this, they compared the Twitter output of 73 journalists covering the 2016 campaign to their professional output in newspaper, magazine, and online articles and broadcast transcripts. In the year leading up to Election Day, they collected 220,000 text samples, encompassing more than 12 million words.
Through automated analysis, they found that journalists’ tweets displayed significantly more linguistic evidence of System 1 thinking than their articles. That is, journalists’ language on Twitter included more emotion, more certainty, more emphasis on the present, and fewer analytic words and numerical terms. Some of these were true of the broadcast sample as well, though broadcast transcripts had even more focus on the present and less analytic language than tweets, and roughly equal levels of certainty to tweets.
Lee and Hamilton also tested for one System 1 cognitive bias in particular: anchoring, or the tendency to estimate uncertainties based on an “anchor” of prior information. They found that journalists who covered previous presidential elections were significantly more likely to refer to those elections in their news reports, even after controlling for age, gender, and type of media. That difference disappeared for references to the 2016 primaries which all journalists had covered, indicating that the increase in references to prior elections could be specifically tied to experience with those elections.
So what does all that mean? We see evidence that campaign reporters rely on cognitive bias in particular — using prior elections as an anchor to influence their perception of current campaigns, potentially distorting their interpretation. But beyond that, we see more evidence that Twitter is where journalists go to process their information heuristically, and their stories (especially for print and online journalists) are where they go to process more systematically.
That shouldn’t surprise anyone who has spent much time following journalists on Twitter. But it’s an important empirical reminder that journalists are quite susceptible to the heightened cognitive biases found on Twitter as they process information in public with unprecedented speed.
An RQ1 read: What’s the Point of News? by Tony Harcup
This is part of an occasional series of summaries by RQ1 readers of notable recent books on news and journalism. This month’s summary is from Zhong (Dan) Zhang, a doctoral student in journalism studies at the University of Sheffield. If there’s a recent research-oriented book on news or journalism that you’d like to write about, let us know!
Fake news, biased news, news fatigue, diminished public trust in news media and not to mention doing journalism in such a polarized world … for journalists and journalism students, this is absolutely not a golden age for doing journalism. However, this can be a right time for someone to stand out and make this statement: news matters.
Tony Harcup’s 2020 book What’s the Point of News reiterates a fundamental but important argument: The value of news is to serve the public good. This is not some sort of meaningless idealistic call or pipe dream from a journalism scholar; rather, drawing on alternative journalism theories and practices in different social circumstances, Harcup clearly states that serving the public good is possible, and also necessary, for news media, and there are definitely some ways to achieve that.
“Twenty ways of making a difference,” “Six criteria for news values” — all the specific tips like these will be helpful for the news industry to make an improvement and for journalists to examine their news reporting, even when facing such a political and economic world that impose various constraints on news production. For pessimistic people in our society who believe journalism is dying, those stories in this book about how theoretical arguments about journalism can be applied in practical work will act as a cardiotonic, sending an inspiring message: Journalists do not quit yet, and you should not either.
“Who, what, and how: Identifying judicial constructions of journalism.” By Jared Schroeder, in Journalism Studies.
Who is a journalist? It’s a question that seems innocuous at first glance, but it gets all kind of complicated when you think about how it has been thrown up in the air in the digitally networked era. And yet, these definitions are crucially important, not simply to scholars (like us) trying to make sense of the shifting boundaries of news and what they mean for the role and practice of journalism in society, but also and especially to the legal system. It is judicial definitions of journalism in U.S. state and federal courts, for example, that shape who gets to enjoy reporter privileges and other press-specific protections.
Schroeder’s article closely examines judges’ efforts to classify what counts as journalism and who qualifies as a journalist — and how those determinations have become increasingly complicated as courts face “a rising tide of cases that have challenged them to interpret laws that have traditionally been reserved for journalists, but are being called upon by all manner of publishers.” For example, can a poster on a message board get shield-law protections? Can political action groups identify as news organizations to receive FOIA fee exemptions? Are bloggers journalists? (OK, you thought that last one was a debate we abandoned lo these many years ago, but the distinction matters in a defamation case.)
Schroder argues that “in articulating rationales for their decisions, jurists, in a very pragmatic sense, have provided a separate and relatively unexplored discourse about journalism in the networked era.” And what does that discourse suggest?
Through close readings of U.S.-based judicial proceedings, Schroeder finds that jurists “constructed a discourse that communicates their understanding of journalism as being defined by the processes and practices newsgatherers follow in creating reports, the nature and public-good intent of the publisher, and the journalistic credentials of those who seek newsgatherer’s protections” (emphasis original). Judges, Schroeder notes, constructed this way of defining journalism not out of a broader worry about its social purpose or its survival in the digital era; rather, theirs was a practical desire to “rationalize decisions in which they had to conclude whether a publisher qualified for protections that have historically been associated with traditional journalism.”
In all, the article contends that these judicial evaluations matter because they reinforce the need to hold fast to journalism’s intent and processes: that is, its public-service orientation (who is it for?) and its basic practices (what does it do?) such as accuracy, fairness in reporting, original content, and sourcing. Importantly, too, the judicial discourse aligned with Bill Kovach and Tom Rosenstiel’s famous book in arguing that journalism is “not defined by technology” — that matters of intent and process are agnostic of the tools involved.
“Forgiving the news: The effects of error corrections on news users’ reactions and the influence of individual characteristics and perceptions.” By Jakob Henke, Stefanie Holtrup, and Wiebke Moehring, in Journalism Studies.
All news organizations make mistakes from time to time, and respectable ones will own up to them and publish corrections accordingly. While there has been research on news accuracy vs. errors for many decades, there has been comparatively less study of news users’ reactions to such errors. This is surprising given the generally low level of trust that people have in news across many countries and the presumably cynical feelings that audiences may develop after spotting errors in news coverage.
In this research paper, the authors not only conducted two experiments to investigate how users respond to errors in the news, but they did so with a twist: They brought into play, among other factors, concepts from the psychology of forgiveness to explore if a person’s willingness to accept apologies and forgive “transgressions” (in this case, a transgression of expectations about news accuracy) might be related to their willingness to forgive journalistic mistakes.
Indeed, the authors found that trait forgiveness — a person’s general ability to forgive transgressors — can serve as a useful predictor of how people engage with news and the errors that appear in the news. However, the paper also notes with some concern that “these results indicate that journalists are, at least partially, at the mercy of their audience: With every error published in the news, there is a non-zero chance that less forgiving news users will lose faith in the news to provide them with an accurate account of current events.”
It’s not all bad news, though. This research discovered that “correcting errors and apologizing for them has positive effects on accepting and excusing errors.” This suggests that journalists could accomplish something just by being more transparent and contrite in the way they approach their audience (e.g., see Jacob L. Nelson’s argument for “journalistic humility”).
“Injecting disinfectants to kill the virus: Media literacy, information gathering sources, and the moderating role of political ideology on misperceptions about Covid-19.” By Porismita Borah, Erica Austin, and Yan Su, in Mass Communication and Society.
Two years into the pandemic, there is no shortage of concern about false information — about fake causes and cures, for example — that continues to circulate widely online. Setting aside for the moment whether “infodemic” is the right term to characterize the situation, it’s nevertheless true that Covid-related fakery has been a key focal point of ongoing research about misinformation and social media.
This study sought to bring together what the authors see as a uniquely important combination of factors that may be connected to people’s susceptibility to fake information about the virus: (1) a person’s methods of gathering information, (2) their media literacy, and (3) their political ideology and its moderating influence on the first two factors.
The authors find that “conservatives, younger individuals, information gathering from social media, conservative media use, and information gathering from Trump were positively associated with Covid-19 misperceptions. On the other hand, information gathering from government organizations such as the CDC and scientists were negatively related to Covid-19 misperceptions.” Perhaps that’s not terribly surprising, given the politicization of the pandemic, and yet it’s worth noting that political ideology couldn’t explain everything: disinformation, the authors note, appeared to “work” on liberals who were not well equipped with media literacy abilities.
“Social media metrics in the digital marketplace of attention: Does journalistic capital matter for social media capital?” By Jieun Shin and Katherine Ognyanova, in Digital Journalism.
This study opens with an important question for news organizations trying to strategize about social media and the unique opportunities and frustrations that it provides: “Is social media a mere popularity contest where clickbait websites have more advantage over reputable news sites? Or is it a marketplace of ideas in which high-quality content rises to the top?”
Taking a cue from Pierre Bourdieu’s field theory and its emphasis on cultural capital, the authors looked at how two different types of “journalistic capital” — brand reputation and site quality — might be connected to “social media capital” (defined here in terms of audience size as well as audience reach through sharing on Twitter).
By analyzing social media metrics drawn from the entire population of Twitter users and a representative sample of U.S. Twitter users, the authors find that a news organization’s journalistic reputation is, indeed, a reasonable predictor of its social media capital.
That’s the good news.
Now for the not-so-good news: “News site quality, however, was not significantly associated with social media metrics. In fact, the quality of news sites was at times related negatively to social media capital” — which meant that, all things being equal, news from low-quality sites tended to get more retweets than news from high-quality competitors. And this pattern was especially evident among politically conservative users.
What does this mean? For one thing, the authors say, “it implies that, like other brands, news media heavily rely on reputation as an important intangible asset” — and such reputations matter when moving to the social media realm.
But reputation also isn’t everything. As the authors sum up, “the finding that news quality does not correspond to social media capital on Twitter is worrisome because it can be a signal that underlying dynamics of social media engagement are polluting the news market.”
Leave a Reply