Social media often seems to be riddled with angst, but research from Northwestern University’s Kellogg School of Management suggests it might not be as angry as we think.

The research suggests that we tend to perceive others as angrier than they actually are when we engage with them online. What’s more, we’re much more likely to engage with content that triggers emotion, which encourages the algorithms behind social networks to amplify it.

Perceiving anger

Employing a novel method that encompassed both content creators and observers on Twitter, the study aimed to examine the extent to which online anger was perceived differently by these two groups and whether such perceptions were accurate.

Leveraging machine learning, the researchers identified Twitter users who expressed high or low levels of indignation while discussing American politics. They then solicited feedback from the content creators, asking them to report their level of happiness or anger when composing each tweet within 15 minutes of posting.

Additionally, the authors presented the tweets to a separate cohort of 650 individuals and requested them to evaluate the level of happiness or anger they thought the content creators had felt.

Overestimating anger

The study revealed that, overall, observers tended to overestimate the level of anger present in the tweets, whereas their estimation of happiness was more precise. This tendency reflects the human inclination to pay greater attention to, and sometimes overanalyze, negative information.

Subsequent studies by the authors involved presenting various participants with artificial newsfeeds consisting of political tweets that were either more outraged or more neutral. They were then asked to assess the level of outrage they believed the average user of the platform would display.

As anticipated, those who were exposed to the more outraged newsfeed concluded that users on Twitter were generally more prone to outrage, and they relied on a select few tweets that were filled with outrage to draw this conclusion.

In the final stage, the authors instructed a third group, who had also viewed the same newsfeeds, to evaluate the appropriateness of new fabricated tweets, which were either outrage-filled or neutral. The participants who had viewed the more outraged-filled feeds rated outrage-filled posts as more socially acceptable on the platform, indicating the presence of a vicious cycle where increasingly outrageous behavior is viewed as normal.

A polarized world

Such interactions can have serious ramifications. The researchers point out that in the United States, there is a growing sense of affective polarization, where political groups are becoming more hostile toward each other. One factor driving this is that members of one group tend to overestimate the extremity of average members of opposing groups.

However, social media does not have to be a mere spiral of escalating toxicity. Much of the content on these platforms is already positive: it can be heartwarming, amusing, or engaging simply because it is interesting, rather than provocative.

“If social media companies sought to find and regulate some of the most extreme content, that might not actually harm engagement much, because a lot of that content is pushed by a small minority of users who are the most extreme and most politically active,” the researchers explain.

A better future

In the coming years, algorithms may be reimagined to avoid taking advantage of our inclination toward negative content. This, coupled with greater transparency about how algorithms function, could contribute toward more informed users.

From an individual perspective, we would do well to pause and reflect on the fact that divisive or inflammatory content does not always reflect the feelings of the average person in our social network, according to Brady.

Moreover, we should remember that online spaces lack the context, cues, and subtleties of in-person conversations. A seemingly blunt or indifferent message from someone we know well may not be so negative in reality, and this same logic can be applied to strangers as well. There is no need to interpret ambiguous content negatively as a matter of course.