Subjective Organizational Performance and Measurement Error: Common Source Bias and Spurious Relationships

Abstract

In any design science such as public management, the importance of measurement is central to both scholarship and practice. Research built on measures that are not valid or reliable can generate misleading results and produces little value in the world of practice. This article applies measurement theory to administrators’ self-perceptions of organizational performance, measures commonly used in the literature. Such measures can be prone to common source bias whereby spurious results are highly likely. This article uses measurement theory to show why common source bias can be a problem and then introduces an empirical test for bias based on a data set that includes both perceptual measures of performance and archival measures. The empirical test shows that spurious results are not only common but that they might include as many as 50% of all statistical tests. The article further examines the specific types of questions and measures most likely to generate measurement bias and provides guidelines for scholars conducting research on public organization performance. 《组织绩效的主观评定和测量误差：同源偏差及伪关系》 作者：肯尼思J. 迈尔，小劳伦斯 J. 奥图尔 与其他设计科学(design science)一样，(准确合理的)量度方法 (measurement) 对于公共管理学的学术研究和实际应用至关重要。基于不合理、不准确的量度方法的科研项目通常会作出令人误解的结论，而对实际应用毫无意义可言。我们将计测理论(measurement theory) 应用到与政府官员对于组织绩效 (organizational performance) 的主观评定相关的实证研究中。主观绩效评定数据在公共管理学的实证研究中广为应用。 然而，这样的绩效量度方法容易引起同源偏差(common source bias)，因而极可能令研究者得出虚假的结论(spurious results)。在本文中，我们引用计测理论来证明同源偏差的危害性。同时，我们通过比较主观和客观的绩效评定数据介绍一种估算计量偏差的实证研究方法。 这种比较研究方法表明， 同源偏差普遍存在，并且有近一半的的数据模型都有此问题。 我们进一步讨论了比较容易造成同源偏差的问卷和量度设计。 在文末，我们总结了相关的研究准则，用以帮助学者们在研究公共组织绩效的过程中减少或避免同源偏差的问题。