Replication Data for: Conceptual and Measurement Issues in Assessing Democratic Backsliding

This paper addresses three interrelated questions. First, how strong is the evidence that democracy has declined globally over the last decade? Second, how should we best measure (change in) democracy? Third, is there evidence that expert coders have become harsher judges of democratic quality in recent years? We begin our analysis by discussing how to conceptualize democracy and democratic backsliding. Next, focusing on V-Dem's methodology, we show---both through theoretical considerations and empirical tests---that it is highly unlikely that time-varying coder biases drive the recent observed global democratic decline. We then dissect the distinction between \"subjective\" and \"objective\" measures, examine how measurement error can affect even seemingly objective indicators, and highlight how subjectivity pervades all measurement enterprises. Finally, we evaluate Little and Meng's (2023) proposed objective democracy measures. We demonstrate multiple issues that make their measurement strategy ill-suited to study trends in global democracy or to serve as a benchmark for very different democracy measures, such as those from V-Dem.

本文探讨了三个相互关联的研究问题。其一，近十年来全球民主出现衰退的证据强度如何？其二，我们应当如何最优地衡量民主（及其变化）？其三，是否有证据表明，近年来专家编码员对民主质量的评判标准愈发严苛？
我们首先从民主与民主倒退的概念化路径展开分析。随后，聚焦于民主多样性项目（Varieties of Democracy，下称V-Dem）的研究方法，我们通过理论推演与实证检验双重维度证明：由编码员偏倚随时间变化所引发的近期全球民主衰退观测结果，在概率上极难成立。
紧接着，我们剖析“主观”与“客观”衡量方式之间的区别，探讨测量误差如何对看似客观的指标产生影响，并阐明主观性如何渗透于所有测量实践之中。最后，我们对利特尔与孟（Little and Meng，2023）提出的客观民主衡量方案进行评估。研究表明，该衡量方案存在多处缺陷，既不适用于全球民主趋势研究，亦无法作为与V-Dem等其他民主衡量体系差异显著的各类指标的参照基准。