Aller au contenu principal

Eugenio Verrina

GATE-Lyon
Social Identity and Selection Neglect
Co-écrit avec Zvonimir Bašić, Stefan Schmidt
Lieu
Îlot Bernard du Bois - Salle 17

AMU - AMSE
5-9 boulevard Maurice Bourdet
13001 Marseille

Date(s)
Vendredi 20 mars 2026
12:00 à 13:00
Contact(s)

Jiakun Zheng : jiakun.zheng[at]univ-amu.fr

Résumé

Selection neglect, i.e., the failure to account for selection in the information one receives, is one of the key drivers behind political polarization. Social identity is often embedded in information structure where selection neglect may arise. For example, relying on information coming from one’s echo chamber and neglecting information outside of it may lead to distorted views. In this context, people belonging to one’s echo chamber could be viewed as in-group members, e.g., because they share one’s political affiliation, while people who do not belong to one’s echo chamber could be perceived as out-group members. In this study, we investigate how social identity and selection neglect together affect belief formation. We design a novel experimental paradigm where subjects guess a computer-generated number. For each guess, receivers observe the estimate of multiple senders who have privately received signals about the correct number. We manipulate i) whether the signal structure can induce selection neglect, and ii) whether social identity is present. We find that receivers suffer from selection neglect, replicating the results from previous studies. Importantly, selection neglect is exacerbated when the observed signals predominantly come from in-group senders. In contrast, if the observed signals come from out-group senders, subjects become much better at accounting for selection. Moreover, we show that social identity alone causes no bias, but it is the combination of selection neglect and social identity that drives our results. Finally, we dig into the potential mechanisms that could drive our results. Our findings offer insights into the determinants of polarization and have important implications for the design of information structure and of policies aimed at reducing it.