I wonder what this model would suggest if used to guide interventions aiming reduce anxiety/violence. Would it be more effective to e.g. introduce more helpful similarity-based influence into each group, or reduce cross-group incursions? (There are obvious connections there to political debates.)
Was a bit concerned about the probability of those clusters in Figure 1 occurring by chance, particularly with the middle one looking so weak: it's about a 0.8% chance of getting three and a 7% chance of getting two over a thirty year period...<p>Just about seems convincing.
This model is pretty far from reality. It treats every religion identically. In the real world there are some religions in which it isn't acceptable to harm any living thing... and others that take a very very different approach to violence. The model ignores this.
I don't trust a paper whose authors are so invested in burying their ideas in pretentious vocabulary that they are forced to turn off their spellchecker and can't spell "xenophobic" consistently.<p>Also, the authors definition of "religion" is quite broad -- it's any belief that some unseen force affects the world. For example, "illegal immigrants" or "Russian bots" or any other "bogeyman" fit the authors; definition of a "supernatural" religious target
The importance of the contagion threat should be noted when people talk about immigrants or minorities bringing disease or calling them vermin or rats. It's been part of every genocide or pogrom.