People often use social media to express outrage about injustice and support calls for social change. But a new study published in Social Psychological and Personality Science suggests that outrage may drive online attention without always translating into meaningful action. Analyzing over a million posts linking to petitions on Change.org, researchers found that posts expressing moral outrage were more likely to go viral—but those same posts were not associated with more signatures.
The researchers were motivated by longstanding concerns about “clicktivism,” the idea that social media enables people to feel as though they’ve contributed to a cause without taking more impactful steps. Petitions offer a useful case study for testing this, as they are widely shared online and require some level of active engagement to sign. The team wanted to know: Does the language used in social media posts influence not just how many people see the message, but whether they’re moved to act?
“I share the concern expressed by many about the potential negative impacts of social media,” said study author Stefan Leach, an incoming lecturer in psychology at the University of Southampton. “Calls for increased regulation and accountability are becoming louder, but social media companies are limiting their responsibilities and curtailing access to data. Independent research has a key role to play by providing an empirically-grounded understanding of how these platforms are shaping society.”
For their study, the researchers collected data on 24,785 unique petitions hosted on Change.org, one of the world’s largest online petition platforms. They used the X (formerly Twitter) academic application programming interface to identify all English-language posts that shared a link to one of these petitions. Their final dataset included over 1.2 million such posts, dating from 2006 to 2023. For each post, they recorded the number of likes and reposts it received, along with the total number of signatures the linked petition had accumulated.
To assess the emotional and linguistic tone of the posts, the researchers used a combination of machine learning models and validated word lists. They used the Digital Outrage Classifier, a tool trained to detect moral outrage in text, as well as BERTAgent, which identifies agentic (goal-oriented) language. They also measured how often posts included words that signal group identity—like “we” and “us”—and prosocial intent, such as “generous” or “selfless.” Each post was scored along these four dimensions.
The research team then used statistical models to examine how these types of language related to three outcomes: how often posts were liked, how often they were reposted, and how many signatures the linked petitions received. To avoid inflating statistical errors, they aggregated data at the petition level, calculating the average level of each language type across all posts linking to a given petition.
The researchers found that posts with higher levels of moral outrage were more likely to be liked and reposted. In other words, outrage helped petitions spread. But outrage did not lead to more signatures. When controlling for virality—how widely a post was shared—outrage was actually associated with slightly fewer signatures. This suggests that while outrage can help get attention, it may not inspire people to take the extra step of signing.
By contrast, agentic language, group identity words, and prosocial expressions showed the opposite trend. Posts using these types of language were not more likely to go viral, but they were associated with more signatures. For example, petitions shared using agentic phrases like “we act now” tended to get more signers, even if the posts didn’t spread as widely. The same was true for posts that signaled shared group identity or emphasized compassion and care.
The researchers also conducted mediation analyses to test whether moral outrage might still have an indirect effect on petition signatures by helping content go viral. They found that while outrage did boost sharing, which in turn predicted more signatures, the direct effect of outrage—independent of sharing—was negative. This indirect pathway means that outrage can still help petitions succeed by spreading them widely, but that benefit may be offset if outrage reduces people’s motivation to engage further.
“I was surprised that petitions posted with moral outrage did not gather more signatures,” Leach told PsyPost. “Moral outrage tends to draw attention, and getting more attention is a surefire way of gathering signatures. That we did not observe this link suggests a more complicated process.”
These findings highlight the potential misalignment between what social media platforms reward and what helps movements succeed. Engagement-based algorithms tend to elevate content that triggers strong emotions like anger or disgust. But this type of content may not encourage the deeper, effortful behaviors required to bring about change.
“Platforms are calibrated to capture attention by amplifying moralized and emotional content,” Leach said. “This can have benefits, such as raising awareness of injustices by propagating expressions of outrage. We saw this in response to the murder of George Floyd. At the same time though, the findings suggest that online moral outrage may sometimes fail to translate into other types of collective responses, which could have greater influence on stakeholders and policymakers, such as petition signing. I hope this encourages a healthy skepticism about the role of social media in activism and collective action.”
As with all research, there are some caveats to consider. The data came exclusively from X, which may not represent how people behave on other platforms with different user bases or engagement features. The researchers also focused on just one type of collective action—signing petitions—and their findings may not apply to other forms of engagement, such as donating, protesting, or contacting elected officials. Since the data are observational, the study cannot determine whether moral outrage causes people to avoid signing petitions, or whether some other factor explains the relationship.
“We cannot say with certainty why online petitions that express moral outrage often go viral yet fail to attract more signatures,” Leach said. “One possibility is that expressing moral outrage is psychologically satisfying—so much so that it may reduce the perceived need for further action. In other words, strongly condemning injustice might feel like enough, in situations where signing a petition would have otherwise seemed necessary.”
Future research could explore these questions using controlled experiments that track how people respond to different types of messaging. For example, researchers could test whether expressing outrage on social media makes people feel as though they’ve already “done their part,” reducing the likelihood that they will take further steps. It would also be useful to examine how moral outrage functions in the context of more sustained or offline forms of activism.
“The possibility that expressing moral outrage online might reduce the impetus for further action is an interesting one,” Leach said. “Confirming it will require careful experimental work to isolate the causal effects of expressing online moral outrage on further actions.”
“I would not want readers to conclude that expressions of moral outrage lack value or should be discouraged,” he added. “In many cases, it is both appropriate and necessary to condemn behavior in the strongest terms and to call attention to injustice. The research is intended to help understand how moral outrage operates online, and to invite further work on its impacts.”
The study, “Moral Outrage Predicts the Virality of Petitions for Change on Social Media, But Not the Number of Signatures They Receive,” was authored by Stefan Leach, Magdalena Formanowicz, Jan Nikadon, and Aleksandra Cichocka.
