From Clicks to Clusters: A Data-Driven Analysis of Non-Electoral Political Engagement on the e-petition Platform

The proliferation of digital participation platforms has reshaped the modes and meanings of civic participation, including in post-communist democracies. This study examines the Latvian citizen-initiated petition platform MyVoice (ManaBalss.lv) as a case of non-electoral, online political engagement that both challenges and complements traditional democratic governance. The research attempts to establish (1) what patterns of engagement emerge in digital petitioning? (2) who participates? (3) how concentrated or diffuse is this form of political expression? And crucially (4) if latent political communities can be detected through shared patterns of action? The research includes computational analysis of over 900,000 voting records linked to nearly 1,000 citizen-submitted initiatives. The method combines descriptive statistics, demographic inference, and network similarity modeling to identify clusters of voters who exhibit shared behavioral profiles. First, normalized metadata on voters and citizen initiatives, inferring age and gender where possible were grouped into engagement tiers (low, moderate, high) based on the number of initiatives supported. Second, a distributional analysis to uncover the degree of participation concentration was conducted. Third, Jaccard similarity indexing was performed over a bipartite user-initiative matrix to uncover latent communities—users who systematically vote on the same petitions. The research finds that engagement on the platform follows a long-tail distribution, with just a few of initiatives accounting for half of all votes. Participation skews toward women and the 30–49 age group, especially on social and health-related initiatives. High-engagement users—those who vote on more than ten initiatives—form distinct behavioral clusters, often aligning with issue-specific campaigns such as COVID-19 policy, tax reform, or animal rights. Importantly, our similarity network analysis highlights how political participation on digital platforms is not only manifest in visible vote counts but also latent in behavioral patterns that suggest emergent publics with shared concerns, preferences, and possibly ideologies. By grounding the empirical analysis in the Latvian context—a parliamentary democracy with strong traditions of mostly representative not direct democratic elements—the research contribute to the literature on democratic innovation, citizen-led governance, and digital modes of participation in transition democracies. The findings support the view that online petitioning can serve as a meaningful complement to electoral politics, particularly in societies where institutional trust is fragile, and opportunities for deliberation are limited. The research offers granular, data-driven perspective on how citizens engage outside formal electoral channels and how such engagement reflects broader shifts in democratic practice. In doing so, the researcher sheds light on how governance is being reshaped not only by voices cast in ballots, but by signatures clustered around shared demands.