Political coherence as a driver of interpersonal liking and ideological sorting

ZIMMERMAN, FEDERICO; NAVAJAS, JOAQUIN

Congreso; 9th International Conference on Computational Social Science IC2S2; 2023

The increase in political polarization has become a worrying concern in different countries as it is a threat to society and democracy itself. Furthermore, polarization encompasses multiple interrelated phenomena such as attitudinal polarization, outgroup animosity, segregation, and ideological sorting. One standard explanation for most of these dimensions is that people prefer politically like-minded others, i.e., political homophily, as this preference biases judgments across multiple domains. However, while the liking-by-similarity effect stands as a largely accepted truth, other factors could also play an important role in the political domain.In the present work, we performed three studies in different countries; and found that, beyond the effect of similarity, people are more attracted to politically coherent ingroups rather than to those who hold ambivalent opinions. For example, a coherent Democrat would be pro-choice regarding abortion and also favor stricter gun control. First, we performed two large-scale social experiments, in which more than three thousand participants were arranged in dyads, discussed five controversial topics, and completed an interpersonal attraction questionnaire. We observed that interpersonal liking increases as a function of similarity but also of ingroup-coherence. These results were validated by performing an online pre-registered experiment in which political coherence was experimentally manipulated.Moreover, we were interested in addressing how this driver of interpersonal attraction relates to the macro-level patterns of political polarization and partisan-ideological sorting. In this direction, we developed a multidimensional agent-based model that incorporates homophilic interactions, in which agents who share similar opinions are more likely to interact, and also ingroup-coherence favoritism, in which agents are more attracted to coherent ingroups. We derived the model’s master equation and fitted its parameters using multiple opinion datasets (e.g., American National Election Studies) which include more than 20,000 standpoints on different controversial issues. By this procedure, we were able to replicate the opinions’ distribution of every dataset and note that homophilic interactions alone could not explain the ideologically sorted states observed in the political domain. However, this gap could be addressed by taking into consideration ingroup-coherence favoritism.In summary, we found empirical evidence that ingroup-coherence drives interpersonal attraction in the political domain, and, by developing a multidimensional agent-based model, we observed that this phenomenon could explain the current levels of ideological sorting.