Public Opinion and Polarisation in the 2021 Israel–Palestine Conflict: A Twitter-Based Study

The 2021 conflict between Israel and Hamas, marked by an eleven-day escalation of violence beginning in May, generated a global wave of reaction on social media. Twitter, in particular, emerged as one of the most prominent digital platforms for the public to express views, share content, and organise around narratives related to the conflict. The academic article “Taking Sides: Public Opinion Over the Israel–Palestine Conflict in 2021” by Daniel Ben Harush presents a thorough analysis of these online behaviours, examining the patterns of engagement, sentiment polarity, and network formations using large-scale computational methods. The study is notable not only for its technical robustness but also for its attempt to provide a neutral and data-driven understanding of how people discuss complex geopolitical issues online. Using a dataset of over 2.7 million English-language tweets collected during the height of the conflict, Ben Harush focuses on mapping public opinion along three axes: stance detection (i.e., determining whether tweets are pro-Israel, pro-Palestinian, or neutral), sentiment analysis (positive, negative, or neutral emotion), and network interaction patterns (who retweets whom, which groups are isolated or central, etc.). This approach provides a layered view of public opinion and helps disentangle the noisy, often emotionally charged discourse on Twitter.

One of the core findings is the significant imbalance in online support: the dataset revealed that pro-Palestinian tweets outnumbered pro-Israel tweets by a considerable margin. However, this numerical dominance does not translate into network dominance. Ben Harush finds that pro-Israel and pro-Palestinian users formed distinct “echo chambers”—networks of users who primarily interacted within their ideological group. In this environment, users rarely encountered or engaged with opposing viewpoints. The study shows that the pro-Israel network was relatively smaller but more internally cohesive and centralised, while the pro-Palestinian network was larger and more fragmented. This network polarisation was intensified by the nature of content shared within each ideological cluster. Pro-Palestinian tweets tended to focus on humanitarian narratives, casualties, and visual content such as images and videos depicting destruction in Gaza.

These posts were widely retweeted and amplified by both grassroots activists and high-profile influencers. In contrast, pro-Israel tweets leaned toward political framing, emphasising Israel’s right to self-defence and the role of Hamas as a terrorist organisation. These were often shared by verified accounts, official state representatives, and media professionals. Neutral tweets were the minority and tended to gain less engagement, reflecting the high degree of polarisation in this digital environment. A particularly interesting dimension of the study is the dynamic analysis of tweet volume over time. The author demonstrates that public engagement follows a “spike and decay” model: tweets surge in response to major events (e.g., airstrikes, civilian deaths, ceasefire announcements) and then quickly decline.

This finding reflects a pattern of reactive discourse, where sustained attention is difficult to maintain unless continually triggered by dramatic developments. The study also employs stance detection using machine learning models to identify each user’s ideological leaning. While technical, this aspect of the research is crucial for quantifying polarisation. The machine learning models achieved high accuracy, allowing the author to track not only what people were saying but also how the discourse evolved across time and geography. Notably, tweets from Western countries (particularly the United States and United Kingdom) were predominantly pro-Palestinian, while those from certain European countries and Israel skewed pro-Israel. This geo-political dimension underlines how national politics influence online opinion. Ben Harush also explores the interplay between sentiment and stance.

Unsurprisingly, both pro-Israel and pro-Palestinian tweets were predominantly negative in tone, expressing outrage, grief, and anger. However, the sentiment intensity differed: pro-Palestinian tweets frequently used emotionally charged language tied to images of children and civilians under attack, while pro-Israel tweets used assertive language focused on justification and security. This divergence in emotional framing helps explain why different messages resonate more with certain audiences and why dialogue between camps is so limited. The paper situates its findings within the broader literature on online polarisation, activism, and information warfare. It acknowledges that Twitter is not representative of the general population but serves as a barometer of elite and activist opinion.

Moreover, it raises the question of whether Twitter activity influences real-world policymaking or merely reflects it. While the article stops short of normative conclusions, it implicitly warns against equating Twitter trends with democratic consensus. From a methodological standpoint, the study stands out for its combination of qualitative interpretation and quantitative rigour. It draws from disciplines including computational linguistics, network theory, and political science to paint a multi-dimensional picture. The tools used—stance detection algorithms, sentiment analysis, and community detection models—are not just technological novelties but essential instruments for parsing complex digital debates. Importantly, the author provides transparency in methodology, including validation metrics and error margins, which enhances the study’s credibility and reproducibility. One of the broader implications of this research is its contribution to the understanding of how public narratives are constructed and contested online during international crises. In particular, the 2021 Israel–Hamas conflict highlighted how non-state actors, influencers, and ordinary citizens play a central role in shaping international opinion. The digital battleground is not just a reflection of reality but part of the conflict itself—amplifying voices, distorting facts, and entrenching divisions.

Ben Harush’s work also serves as a cautionary note for media consumers and policymakers. In highly emotional, fast-moving crises, online discourse is prone to misinformation, manipulation, and performative activism. The study notes several instances of false images and recycled videos gaining viral traction, reinforcing the need for media literacy and critical engagement with digital content. In conclusion, “Taking Sides” is a compelling and timely study that sheds light on the mechanics of digital polarisation in the context of one of the world’s most enduring and contested conflicts. It offers a model for future research on digital public opinion and sets the groundwork for more nuanced discussions about media, identity, and politics in the digital age. While the 2021 conflict may have faded from the headlines, its digital traces—and the patterns of engagement it revealed—remain deeply relevant for understanding future conflicts and the contested narratives that surround them.

Reference: Ben Harush, Daniel. “Taking Sides: Public Opinion Over the Israel–Palestine Conflict in 2021.” Master's thesis, George Mason University, 2020. https://mars.gmu.edu/handle/1920/13178.