Using field and laboratory data, we show that leader charisma can affect COVID-related mitigating behaviors. We coded a panel of U.S. governor speeches for charisma signaling using a deep neural network algorithm. The model explains variation in stay-at-home behavior of citizens based on their smart phone data movements, showing a robust effect of charisma signaling: stay-at-home behavior increased irrespective of state-level citizen political ideology or governor party allegiance. Republican governors with a particularly high charisma signaling score impacted the outcome more relative to Democratic governors in comparable conditions. Our results also suggest that one standard deviation higher charisma signaling in governor speeches could potentially have saved 5,350 lives during the study period (02/28/2020–05/14/2020). Next, in an incentivized laboratory experiment we found that politically conservative individuals are particularly prone to believe that their co-citizens will follow governor appeals to distance or stay at home when exposed to a speech that is high in charisma; these beliefs in turn drive their preference to engage in those behaviors. These results suggest that political leaders should consider additional “soft-power” levers like charisma—which can be learned—to complement policy interventions for pandemics or other public heath crises, especially with certain populations who may need a “nudge.”

利用现场和实验室数据，我们表明领导者魅力可以影响与新冠病毒相关的缓解行为。我们使用深度神经网络算法对美国州长的演讲进行了编码，以传递魅力信号。该模型根据智能手机数据移动解释了公民留在家里的行为的变化，显示了魅力信号的强大影响：无论州级公民的政治意识形态或州长政党的忠诚度如何，留在家里的行为都会增加。在类似条件下，与民主党州长相比，魅力信号得分特别高的共和党州长对结果的影响更大。我们的结果还表明，在研究期间（2020 年 2 月 28 日至 2020 年 5 月 14 日），州长演讲中的魅力信号提高一个标准差可能会挽救 5,350 条生命。接下来，在一项激励实验室实验中，我们发现，政治上保守的个人特别容易相信，当他们听到极具魅力的演讲时，他们的同胞会听从州长的呼吁，保持距离或呆在家里；这些信念反过来又促使他们倾向于参与这些行为。这些结果表明，政治领导人应考虑额外的“软实力”杠杆，例如魅力（这是可以学习的），以补充针对流行病或其他公共卫生危机的政策干预措施，特别是针对可能需要“推动”的某些人群。