light music
Narrator: Having reliable and well-researched news is crucial for a democratic society to function effectively. Scholars have long debated the role of news organizations in determining what’s newsworthy and what isn’t. How do TV networks like CBS, NBC, and Fox decide what to report? Chicago Booth’s Emir Kamenica and his research collaborators set out to answer that question.
Emir Kamenica: If you consider the nightly news, which is precisely the context that we’re studying empirically in this paper, they have the fixed amount of time that they’re going to dedicate to telling you what happened in the world and they’re going to have to make decisions about whether some particular thing—like the stock market is up 1.2 percent—whether that is newsworthy enough to talk about relative to the opportunity cost of talking about other things that happened in the world. And in order to determine whether some particular event—the stock market’s up 1.2 percent—how newsworthy that is, we need some kind of a benchmark. And a very natural benchmark is built through this idea of a scoring rule as a measure of newsworthiness of a particular realization of an unknown state.
Narrator: The researchers develop a scoring rule to test how newsworthy a specific event would be based on how much the development differs from what was previously expected—the surprise factor.
Emir Kamenica: In this project, we try to do, first, sort of systematic analysis of certain types of data on what gets covered by the news, and complement that with a sort of a theoretical analysis of what might we expect, what sort of patterns in coverage should we see under an off-the-shelf, benevolent, nonmanipulative model of news coverage?
Narrator: Researchers explored how the news media cover various topics, examining supposed biases in reporting. Two notable biases are the negativity bias, where negative news is more frequently reported than positive news, and the fatigue bias, which occurs when audiences lose interest in a topic, leading to decreased coverage.
Emir Kamenica: But what’s going to look a lot like fatigue bias is something just stops being newsworthy. Something that initially was surprising and worth knowing because it was unexpected eventually becomes relatively common. And even if it’s sad and even if it’s very costly to a lot of people, it just becomes less newsworthy because it is something that we might have expected to happen even if we didn’t tune into the news. So the perspective we try to bring is that what news ought to be telling us are precisely those things we might not expect, we might not think happened if we did not tune into the news.
Narrator: The researchers applied their model to actual data, analyzing news coverage patterns from nightly news transcripts, focusing on stock market fluctuations, unemployment rates, weather conditions, and US military casualties in Iraq and Afghanistan. This approach not only highlights the underlying biases in media reporting, like the negativity bias, but also provides a structured method to analyze how news topics are selected and reported based on their unexpectedness.
Emir Kamenica: So we could ask: Is it the case, in fact, that negative movements to the stock market are more likely to be talked about than positive movements? We do not find any such pattern, and moreover, we do not find that pattern because what we do find is that what really drives whether a stock market movement has been reported, is likely to be reported, is whether it was of the magnitude that was surprising. What really is likely to get covered is a large movement at what was anticipated to be a quiet day. We also conduct similar analyses of the coverage of unemployment rate, of precipitation, of military casualties in Iraq and Afghanistan. And in all of these domains, we find that the model of the reporting decision based on this view that we want to report whatever is newsworthy actually performs extremely well.
Narrator: The researchers then explored the role of the surprise factor in news coverage.
Emir Kamenica: If you were to simply look at whether negative news about unemployment and the unemployment rate went up instead of staying the same or going down, whether they’re more likely to be covered, you would find that in fact they are. There seems to be about a 5 1/2 percentage point gap in how likely negative news—a jump in the unemployment rate—is likely to be covered. But it turns out that this is partly, probably more than 50 percent, due to the fact that quite often, those negative news are in fact more newsworthy than positive news, even if they have the same sort of raw magnitude. The reason for that is that the distribution of changes in the unemployment rate is skewed. The unemployment rate tends to rise more quickly and then sort of fall down more slowly. Because of that, these negative movements can often be much more of a surprise than the—when I say negative movements, I mean negative news—than the positive news that the unemployment rate has decreased.
Narrator: The research highlights the importance of considering not just the positive or negative nature of news, but whether it’s a surprise. This approach provides a clearer perspective on potential biases in news coverage. But can it reveal when news agencies are themselves politically biased?
Emir Kamenica: The goal here is not to prove that there is no bias in media coverage. The goal is rather to think more carefully about what steps one would want to take in order to establish bias, and to take into account the fact there are forces such as, say, different volatility of the stock market in some period versus another that’s going to govern what reporting looks like even in the absence of bias. So what we really are proposing is a framework through which one can think about testing for the presence of bias rather than trying to establish its absence in all aspects of media coverage.