Digital platforms provide an ample environment to express opinions where individuals engage in conversations on multiple topics simultaneously. These topics often correlate due to the overlap of arguments used to justify positions for or against each issue. Previous research suggests that interpersonal interactions tend to have a persuasive effect, with individuals holding extreme views becoming more moderate when exposed to opposing perspectives. However, an increasing trend of societal polarization is observed both within digital spaces and in broader social contexts. This paradox may be explained by the bias in how individuals assess the validity of others' opinions based on the degree of alignment with their own views.

To investigate this phenomenon, we propose an agent-based model where opinions are represented in a multidimensional space. In this model, the influence exerted by one agent on its neighbors is determined by the similarity between their opinions. By simulating persuasive interactions between agents with opposing views and reinforcement among those with similar opinions, the model successfully reproduces a range of polarized and ideologically aligned states. We further validate the model by analyzing political survey data, classifying opinion distributions as polarized, ideological, or non-polarized. These results provide insights into the mechanisms driving polarization and ideological coherence in social systems.