Scroll. Like. Divide. The filter bubble effect on electoral perceptions

This study presents a validated psychometric instrument for assessing filter bubble effects and underscores the need for targeted, system-level interventions that extend beyond individual user-focused solutions. By situating filter bubbles within electoral dynamics, this study contributes to a more nuanced and policy-relevant understanding of how social media platforms shape democratic discourse. This study examines the impact of algorithmic personalisation on electoral perceptions through the lens of the filter bubble phenomenon. Drawing on a multidimensional instrument, the Bubblemetrix Index, we assess four core components: algorithmic curation awareness, perceived source diversity, ideological homogeneity and confirmation bias, and repeated interaction and engagement. Data were collected through an online survey of 451 Romanian Facebook users during the pre-electoral period and analysed using latent profile and latent class modelling. The findings identify three distinct user profiles, with the largest group consisting of younger individuals who demonstrate high algorithmic awareness but also exhibit strong confirmation bias and repeated engagement, which suggest that awareness alone is insufficient to counteract ideological encapsulation. Older users, by contrast, exhibited lower susceptibility to ideological homogeneity and engagement-driven reinforcement. The results challenge the assumption that media literacy and perceived diversity constitute adequate safeguards against polarisation and suggest that algorithmic systems may create an illusion of diversity while structurally amplifying partisan exposure.