Why does interpersonal violence erupt? A key source is conflict over resources and a little-studied mechanism for solving such cases concerns the availability of credible information about individuals’ abilities and willingness to use force. Its core prediction is that when such credible information abounds, the outcome of a potential struggle can be anticipated and thus there is no need to fight. I test this prediction using a nationally representative survey of male prisoners (Ninmates = 10,768, Nfacilities = 207) incarcerated in US state correctional facilities. I also make two methodologically orientated contributions. First, recent research highlights the important consequences that data analysis choices can have in determining quantitative findings. To overcome this, I draw on multiverse analysis and specification curve analysis to map out the key analytical decisions I make and run 262,144 plausible regression models. By doing so, I classify associations that are robust, mixed, or entirely frail according to analytic decisions and find partial support for the informational mechanism of interpersonal violence. Second, in contrast to much quantitative prison research, which uses only official or institutionally punished assaults data, I compare self-recognized fights to self-reported institutionally punished fights and highlight key differences.

中文翻译：

---

信息可以减少人际暴力吗？来自监狱的证据

为何会爆发人际暴力？一个关键的根源是资源冲突，而解决此类案件的一个很少被研究的机制涉及有关个人使用武力的能力和意愿的可靠信息的可用性。其核心预测是，当这种可信的信息大量存在时，潜在斗争的结果是可以预见的，因此没有必要进行战斗。我通过对美国州立惩教所中监禁的男性囚犯（Ninmates = 10,768，Nfacilities = 207）进行的全国代表性调查来检验这一预测。我还做出了两项以方法论为导向的贡献。首先，最近的研究强调了数据分析选择在确定定量结果方面可能产生的重要后果。为了克服这个问题，我利用多元宇宙分析和规范曲线分析来制定我做出的关键分析决策，并运行 262,144 个合理的回归模型。通过这样做，我根据分析决策对稳健、混合或完全脆弱的关联进行分类，并找到对人际暴力信息机制的部分支持。其次，与许多定量监狱研究（仅使用官方或制度惩罚的攻击数据）相反，我将自我认可的打架行为与自我报告的制度惩罚的打架行为进行比较，并强调关键差异。