We use a field experiment conducted in a North American hospital to study the performance effect of feedback when negative incentives are present. We analyze feedback effects on e-prescribing rates in a context in which the 181 physicians participating in our experiment have an e-prescribing target and a threat of termination for failing to reach that target. This research context is unique in its use of negative incentives, given that prior research shows that people react differently to positive versus negative incentives, but existing feedback-incentive studies commonly use only positive incentives. Our study finds that the effects of feedback differ according to the ex ante level of performance relative to the termination threshold. We find that low performers (those who are ex ante below the termination threshold) who receive feedback improve their e-prescribing rates less and later than low performers in the control group who do not receive direct information about their performance. These differences between the control and treatment groups are also present in a partition of medium performers who underestimate (or overestimate less) their performance, but are not present in either the partition of medium performers who overestimate more, high performers, or performers who exhibit feedback-seeking behavior.

我们使用在北美医院进行的现场实验来研究当存在负面激励措施时反馈的效果。我们在181名参加我们实验的医生具有电子处方目标以及未能达到该目标的终止威胁的情况下，分析了对电子处方率的反馈影响。鉴于先前的研究表明人们对积极激励和消极激励的反应不同，因此该研究背景在使用消极激励方面是独特的，但是现有的反馈激励研究通常仅使用积极激励。我们的研究发现，反馈的影响会根据事前表现水平（相对于终止阈值）而有所不同。我们发现绩效不佳的人（事前低于终止阈值的人员）相比没有收到有关其绩效的直接信息的对照组中表现不佳的人，获得反馈的电子处方率提高的幅度要小，而且要晚一些。对照组和治疗组之间的这些差异也存在于低估（或低估）其表现的中等表现者的分区中，但不存在于高估更多，较高表现者或表现出反馈的中等表现者的分区中寻求行为。