What spreads more efficiently on Twitter: political love or hate? The answer may surprise you

A recent study published in Scientific Reports has provided new insights into how partisan content spreads on social media during election campaigns. Focusing on Twitter activity during four consecutive national elections in Spain, the study suggests that positive messages about users’ preferred political parties spread more efficiently than negative messages about opposing parties.

In other words, “in-party love” is shared and propagated more successfully than “out-party hate,” even among radical groups. This challenges common perceptions about how social media contributes to political polarization.

Many commentators have argued that social media plays a significant role in fostering political polarization by allowing like-minded individuals to form tight-knit communities that reinforce negative views about opposing political parties. This has been particularly linked to the rise of populist right-wing parties. However, the evidence from prior research has been inconsistent. Some studies suggest that users are exposed to more diverse opinions than expected, while others claim that social media amplifies extreme partisan attitudes.

The researchers behind this study aimed to clarify these conflicting findings by focusing on Twitter behavior during electoral campaigns. Election periods are key moments when political competition is at its most visible, and social media platforms are heavily used by political parties and their supporters. The Spanish context, marked by the rise of several new political parties and significant ideological fragmentation, provided an ideal case for studying these dynamics.

“I am interested in polarization from a general point of view because it seems widespread across multiple countries and because of its potential harms to democracy,” said study author Samuel Martin-Gutierrez, a postdoctoral researcher at the Complexity Science Hub, who conducted this research while a PhD student at the Technical University of Madrid.

Free daily newsletter

“In this paper, my collaborators and I focused on online communication because both experts and laypeople tend to believe that political communities online are organized in echo chambers and that negative polarizing messages spread further and wider than positive ones in social networks, so we wanted to verify whether that was the case with a systematic study.”

For their study, the researchers examined Twitter activity during four national elections in Spain, which took place in 2015, 2016, and twice in 2019. To gather data, the researchers used Twitter’s public data stream, focusing on tweets that mentioned the main political parties and their leaders. These tweets were identified through specific keywords such as party names, slogans, and candidate names.

The researchers then concentrated on retweets as a key measure of influence and message dissemination. Retweeting, they argued, represents more than just an acknowledgment; it reflects a user’s active decision to share a message with their followers, thereby amplifying its reach.

To analyze the data, the researchers focused on three main variables. The first was user ideology, where Twitter users were placed on a left-right ideological spectrum based on their retweeting patterns. Users who frequently retweeted content from left-wing parties were categorized as left-leaning, while those who retweeted right-wing parties were classified as right-leaning.

The second variable was retweeting efficiency, which measured how effectively users’ content spread on the platform. This was calculated by averaging the number of retweets each user received for their tweets. The higher the average number of retweets, the more efficiently that user’s content was spreading.

Finally, the researchers examined in-party vs. out-party content. This involved distinguishing between positive messages that expressed support for the user’s own party (in-party content) and negative messages that targeted other political parties (out-party content).

The study found that Twitter users were much more likely to retweet positive messages about their own political party than negative messages about opposing parties. This trend was consistent across all four elections analyzed. Even among users who supported radical parties, such as the far-left Podemos and the far-right Vox, positive in-party messages were more likely to be shared than hostile out-party messages.

Interestingly, the study also revealed that while negative out-party messages were common on Twitter, they were not very effective in terms of spreading widely. In fact, tweets attacking opposing parties tended to receive fewer retweets, even when posted by users affiliated with radical parties. This suggests that while users may express hostility toward other parties, they are less inclined to share such messages with others.

The researchers observed that areas of high retweet efficiency—where content spread the fastest—were dominated by positive in-party messages. These areas, which they described as “epistemic bubbles,” were ideologically homogeneous, meaning they primarily involved users sharing messages within their own political group. However, these bubbles were more likely to amplify messages of in-party support rather than out-party hostility.

“Finding such a clear difference between the kind of messages posted by highly and lowly retweeted users was surprising,” Martin-Gutierrez told PsyPost. “Usually in these kind of real-world social studies statistics are not so clear-cut. Our results however show that while users with a low average retweet count post positive in-party messages and negative out-party messages at a similar rate, users with a high average retweet count post almost exclusively positive in-party messages.”

Another key finding was that the rise of new political parties, such as the far-right Vox, contributed to the creation of more ideologically homogeneous online communities. As these parties gained prominence, Twitter users became more likely to cluster into groups that primarily engaged with like-minded individuals. However, even within these more radical groups, positive in-party messages were more successful at spreading than negative out-party messages.

“The most relevant result is that this notion that hateful messages spread further seems to be misguided. Instead, we found that users with high average retweet counts posted mostly positive messages reinforcing their party affiliation,” Martin-Gutierrez said.

“But beyond that, we realized that the concept of an echo chamber does not fully capture the way online political communities are structured. Users in an echo chamber are not only isolated from outside information, but also actively discredit and exclude outside sources. Considering our results, we need to consider the concept of an epistemic bubble, where people are only exposed to information that aligns with their existing beliefs, without necessarily attacking out-group members. In a nutshell, we find that in a political environment epistemic bubbles are efficient in spreading messages online but echo chambers are not.”

While the study provides valuable insights into how partisan content spreads on social media, it also has several limitations. First, the study focused solely on Twitter, which is just one of many social media platforms. The dynamics observed on Twitter may not be the same on other platforms like Facebook or Instagram, where users may engage with political content differently.

Another limitation is the focus on retweets as a measure of influence. While retweeting is an important way that information spreads on Twitter, it is not the only form of engagement. Users may also like, comment on, or share tweets privately through direct messages, and these forms of engagement were not captured in the study.

“We can’t establish a clear causal direction between message popularity and message content: do popular users post more positive messages, or are the positive messages making users popular (and vice versa for negative messages)?” Martin-Gutierrez said.

The study also simplified the ideological landscape by grouping political parties into left- and right-wing blocks. This approach may have overlooked important nuances in how users engage with specific political ideologies, especially in a multiparty system like Spain’s.

“Although this simplification is widely used, multiparty democracies are typically multidimensional, such that additional axes are required to fully characterize this ideological space and enable the distinction between parties of a similar ideology,” Martin-Gutierrez noted. “We addressed this question in another paper.”

Looking ahead, future research could also explore how the findings from this study apply in other countries with different political systems and social media usage patterns. While the Spanish context provided a useful case for studying political polarization, it is not clear whether these patterns would hold in other contexts, such as the United States or the United Kingdom.

“My long-term goal for this research line is to go beyond the diagnosis of polarization to understand its structural causes using data and quantitative methods,” Martin-Gutierrez said. “I’m interested in understanding how an initially unpolarized society may transition into a polarized one.”

The study, “In-party love spreads more efficiently than out-party hate in online communities,” was authored by Samuel Martin-Gutierrez, José Manuel Robles Morales, Mariano Torcal, Juan Carlos Losada, and Rosa María Benito.