We analyzed a data set containing 7,216 hostage and barricade incidents that had been reported to the Federal Bureau of Investigation over a 35-year period. From two subsamples of the data set, we identified potential predictors of important outcomes—resolution by negotiation or surrender and violence after onset. In a third subsample, we combined and weighted the potential predictors to form two actuarial tools. We used three additional subsamples in the data set to cross-validate and calibrate the scores of each tool. Predictive validity was acceptable across all subsamples.

中文翻译：

---

人质和路障事件中谈判和暴力的情景预测因素

我们分析了一个数据集，其中包含 35 年来向联邦调查局报告的 7,216 起人质和路障事件。从数据集的两个子样本中，我们确定了重要结果的潜在预测因素——协商解决或投降以及发病后的暴力。在第三个子样本中，我们将潜在的预测因子组合并加权以形成两个精算工具。我们在数据集中使用了三个额外的子样本来交叉验证和校准每个工具的分数。所有子样本的预测有效性都是可以接受的。