Abstract

Self-reported measures of leadership are widely used in public management research, but nonresponse bias poses a threat to the validity of these data. Although this measurement problem is acknowledged, it has received limited empirical attention because nonresponse bias is inherently challenging to study. To address this issue, we examine nonresponse bias among public managers by analyzing multilevel surveys of managers and employees in which we can compare employee ratings of leadership for both responding and nonresponding managers. Using 16,531 employee responses spread over six data sets from three countries, we find only limited evidence of nonresponse bias in managers’ self-reported leadership. Additional Bayesian analyses indicate that—overall—the data are indicative of the absence of substantial nonresponse bias. However, the results vary between data sets and call for more research on nonresponse bias in leadership research.

摘要

自我报告的领导力测量广泛用于公共管理研究，但无反应偏见对这些数据的有效性构成了威胁。尽管这个测量问题得到了承认，但它受到的实证关注有限，因为无响应偏差本质上是难以研究的。为了解决这个问题，我们通过分析经理和员工的多层次调查来检查公共经理的不回应偏见，在这些调查中，我们可以比较回应和不回应经理的员工领导力评级。使用分布在来自三个国家的六个数据集的 16,531 名员工回应，我们发现只有有限的证据表明管理者自我报告的领导力存在不回应偏见。额外的贝叶斯分析表明，总体而言，数据表明不存在实质性的无响应偏差。然而，