In Broadcast News, the movie that made me want to be a journalist, the protagonist Jane Craig (Holly Hunter) has a routine every morning. At an appointed time, she sits and weeps profusely for a minute. When the minute’s done, she wipes her face and goes about her day without showing outward signs of sadness.

Journalists have always been a fairly morose bunch, and the news they produce reflects that. Communications scholars have found that across many years and countries, coverage of political topics tends to more often be conveyed in a negative or cynical tone rather than a positive one; one study in the mid-2000s found that about half of US, German, Italian, and Austrian campaign coverage conveyed bad news, while as little as 6 percent conveyed good news. By some measures, the situation is deteriorating; a recent study found that the “proportion of headlines denoting anger, fear, disgust and sadness” grew markedly in the US between 2000 and 2019.

Some news consumers have surrendered to the phenomenon and find themselves hooked on “doomscrolling,” in journalist Karen Ho’s memorable term, proceeding between articles asking if the war in Ukraine could be World War III, or whether another world-destabilizing pandemic could be on the way, or if we’ve already passed key climate tipping points. At least some news consumers aren’t too happy about the situation. An international survey from Oxford’s Reuters Institute for the Study of Journalism last year found that in almost every country surveyed, trust in media is falling, and more people are saying they’re avoiding news. Why? Because, respondents say, it “has a negative effect on their mood.”

So … why are we like this? Are we journalists just a miserable lot who insist on spreading our neuroses to the rest of the world? Are readers, despite their protestations to the contrary, likelier to click on news that’s negative or dire?

It’s, of course, both, and the supply- and demand-side reasons might come from the same source. Humans, it turns out, have what social psychologists call a “negativity bias”: We tend to pay more attention to bad-seeming information than good-seeming information. That could be a root factor for why the news is so goddamned depressing. That’s what we’re looking for.

Negativity bias, explained

One of social psychologists’ greatest passions is scouring human behavior for its many failures of rationality and perception, the systematic biases that push us off track. “Negativity bias,” the tendency for negative information and experiences to overwhelm the positive, kept coming up. As early as 1967, psychologist Marjorie Richey and co-authors concluded that university students, given paragraphs describing a stranger’s personality, were influenced more by negative descriptions than positive ones. In 1982, Teresa Amabile and Ann Glazebrook proposed that there might be a general “bias toward negativity in evaluations of persons or their work,” noting that already by that point, a number of other studies had found the same.

A 2001 review paper put it bluntly: “bad is stronger than good.” And all this research was conducted before the dawn of the doomscrolling Instagram era. It points to something deep in human cognition, rather than the effects of social media.

Now, in the year of our Lord 2023, your first reaction to someone telling you “social psychologists say X” should be “why in the world would I believe social psychologists, given that so many of their fanciest results keep getting overturned?” It’s true: This field was ground zero for the “replication crisis,” and many social psych concepts that were once widely touted (like “ego depletion,” devised by the main authors of the “bad is stronger than good” paper) have crumbled when subjected to repeated tests.

Researchers I spoke with, however, said that the general existence of a negativity bias is so widely validated that it has thus far survived the replication crisis unscathed. A huge “diversity of labs and traditions and backgrounds have found evidence for negativity bias, in memory and attention across all kinds of stimuli,” Carey Morewedge, professor of marketing and Everett W. Lord Distinguished Faculty Scholar at Boston University, explained. Morewedge’s work has found a negativity bias in “external agency”: When something bad happens, people are likelier to blame another person for a bad event than give them credit for a good one. Subsequent work on infants replicated that finding.

“This is probably one of the most robust findings in the psychology literature,” Stuart Soroka, a professor in the communications and political science departments at UCLA, agreed. Soroka specifically studies what this bias means for news. With Patrick Fournier and Lilach Nir, he conducted a massive 17-country study with over 1,100 participants measuring how consumers reacted to positive- or negative-seeming news. “Respondents watched 7 randomly ordered BBC World News stories on a laptop computer while wearing noise-cancelling headphones and sensors on their fingers to capture skin conductance and blood volume pulse,” Soroka et al wrote.

Examples of positive news included a ballet company in Brazil that employs blind dancers, a young child recovering from a liver disease, and a Swedish news story about “the ongoing popularity of ABBA.” Examples of negative news included the Peruvian town of Chimbote burning down, UN investigations into war crimes in Sri Lanka, and ultra-Orthodox activists in Israel blocking girls from going to school. By looking at physiological responses like “skin conductance,” or how easily electricity passes through the skin, a measure widely viewed as reliable that has been in use in various forms since the turn of the 20th century, the authors attempted to get around the limits of self-reporting and identify readers’ immediate, sometimes subconscious reactions. Given that people sometimes say they prefer good news even when their behavior suggests the opposite, relying on something other than self-reports makes sense.

Across their sample, they found that negative news provoked stronger physiological reactions and garnered more attention than positive or neutral news on average — though individual people’s reactions varied quite a bit, with a minority of people responding more to positive news.

This speaks to the demand side of the bad news dilemma. People who watch and consume news seem to be drawn to negative, dour stories more than positive ones. But it speaks to the supply side too. Journalists have some leeway in deciding what stories to cover, and if we, too, have a negativity bias, we could be facing the same impulses pushing us toward more negative stories that our readers do.

Good news for people who like good news

Soroka is not a doomer. He thinks the conditions for good-vibes journalism are actually improving. For one thing, he doesn’t think the overall mood of reporting is getting worse.

About a decade ago, he and colleague Lori Young developed a “sentiment analysis” tool meant to assess how positive or negative different political messages are, known as the Lexicoder Sentiment Dictionary. The LSD is simple: It contains a list of positively coded words (“decency,” “priceless”), and negatively coded words (“brazen,” “psychotic”), and counts how often each occurs in the test, adjusting for the use of negations (so “not good” codes, accurately, as negative and “not bad” codes as positive). Soroka argues that the dictionary’s classifications of texts into “positive” versus “negative” tend on average to line up with how humans classify them, and by automating the process, the LSD enables researchers to analyze much vaster texts.

Stuart Soroka and Yanna Krupnikov, The Increasing Viability of Good News, 2021.

He and another co-author, University of Michigan’s Yanna Krupnikov, used the LSD to analyze the sentiment of nightly news segments on NBC, ABC, and CBS from 1990 to 2018. They found no downward trend — but a lot of variability. Sometimes (after terrorist attacks, notably) coverage is unusually negative; sometimes (like after Barack Obama’s election) it’s glowing. But, on average, there’s not much of a trend line.

More importantly, Soroka and Krupnikov argue, people vary in their receptiveness to good versus bad news. Good news that seems “novel” or like an “outlier” tends to get more coverage, as people separately have attention biases toward novelty that can mitigate their negativity bias. (Bad may be stronger than good, but new may be stronger than everything.) They note that on nightly TV news, the final segment is almost always “good news.” “You don’t want to leave the audience on a total downer before you say good night,” Frederica Freyberg, a Wisconsin anchor for PBS, told the authors. In other words, people crave something different from the dour news that came before; they desire novelty.

Some people are also just overall more interested in good news than others. With his colleague Marc Trussler, Soroka once conducted a lab study to see if consumers who say they prefer good news stories actually click more on them in practice. They didn’t; whether you say you want to read good news doesn’t predict your actual consumption habits. But they nonetheless found heterogeneity: There was a substantial minority of people, both those who said they preferred good news and those who didn’t, who really did click more on good news.

In a world of media monopolies, where most people were dependent on one or two local newspapers and three national news networks, this happy-go-lucky minority was … out of luck. The majority preference prevailed, and the majority was biased toward the negative.

We don’t live in that world anymore. The nightly news has collapsed in popularity, newspapers from halfway around the world are as easy to access as local ones, and just about everyone has access to thousands of rival news outlets. For Soroka and Krupnikov, that suggests that the market for good news is getting stronger. Outlets can carve out niches offering less negatively valenced articles in a way they couldn’t 30 or 40 years ago.

“The end result is probably that people are better able to find their ideal balance of content now than they ever were before,” Soroka told me.

And they find evidence that this is happening, or at least happened for a brief moment. They document that some highly successful news posters on Facebook — specifically Upworthy and Occupy Democrats on the left, and the “Barracuda Brigade” on the right — posted more positive than negative news on average. Diehard partisans are open to positive news if it’s mobilizing, affirms their beliefs, etc. That said, because access to more recent data is lacking, their data here all comes before 2017 and the Trump era. Stuff … got dark then.

As any reporter can tell you, though, the new media landscape is hardly all roses. For one thing, the algorithmic structure of platforms like Facebook can magnify our own negativity biases. From 2016 to 2019, Facebook gave “anger” emoji reactions to posts five times as much weight as “likes” in deciding which posts to show other users because their machine learning algorithms found posts that angered people fueled more engagement than posts that pleased them. That partly reflected that humans do, in fact, prefer to share news that enrages them, but it also magnified that tendency, which has costs for both the site and its users.

News consumers, then, are caught between two competing forces. On the one hand, they enjoy a vastly larger and more diverse news ecosystem than has ever existed before in history, as well as social media networks that serve them up exactly the news they demonstrate they want through their posts, likes, and other interactions. This should in principle make it easier for people who want good news to access it. But it also places consumers at the mercy of their own impulses. While at a higher level they may want to want news that makes them less miserable, in the moment they might prefer doomy news — and the media and the platforms they depend on are only too happy to serve up the bad.

“We may, right now, be more motivated to attend to a negative story,” Morewedge explains. “If there’s enough negative information, that may reduce my incentive to come back to that site. It may have a negative effect on my well-being. In the long-term, these sites may be better suited by providing more of a mix of positive info. Just thinking about what people do in the present may not capture their full preferences.”

He analogizes the current situation to an algorithmically run airline, which decides to only serve the meals people most want in the moment. That airline would start by offering people either, say, potato chips or baby carrots; when almost everyone chose the potato chips, maybe they’d move on to asking “potato chips or brownies,” then “brownies or ice cream,” and before long the whole menu is sugar. That satisfies people’s immediate preferences, but in the long run it makes them miserable.

That’s the tricky task, for news outlets as well as social networks, in thinking about negativity bias. We can give the people what they want right now. But in doing so, we might be feeding them empty calories that will only make them sick in the future.