历史遗产与原苏东国家的民主转型——基于26个国家的模糊集与多值QCA的双重检测

Historical Legacies and Democratic Transition of Post-Communist Regimes:A FSQCA and MVQCA Analysis

历史遗产是解释原苏东国家民主转型的一个重要维度。作者以格里戈雷·波普-伊莱切斯运用历史遗产来解释原苏东国家民主转型的不同结果的研究为基础,比较回归分析和定性比较分析两种技术。伊莱切斯用回归分析甄别出对转型影响最为显著的7个历史遗产因素,但回归分析方法在变量间的自相关和共线性的影响下,对变量的作用存在错估的可能。定性比较分析的结果则表明,"非伊斯兰教"是民主巩固的必要条件,"非二战前苏联加盟共和国"接近于必要条件。除伊斯兰国家之外的其余20个国家则有5条通往民主巩固的路径,独立国家经历和非苏联加盟共和国是其中覆盖率最高的条件组合形式。基督教、东正教等宗教变量对民主是否巩固几乎没有影响,这一结果挑战了伊莱切斯以回归分析所得出的基督教有显著正向作用的结论。伊斯兰国家同时具备二战前苏联加盟共和国、缺乏独立国家经历、经济水平较低等不利于民主巩固的历史遗产,因而,伊斯兰教的作用也可能被高估。通过两种方法的比较可以得出,在变量主要由二分、定类和定序等形式组成且中小规模样本的研究中,定性比较分析相比于回归分析具有一定的优势。

Historical legacies have been an important set of explanatory variables for explaining post-communist regime transitions. Using regression analysis, Pop-Eleches has identified seven factors of historical legacies that significantly influence post-communist regime change. However, there are possibilities of misestimation because of multicollinearity and interaction among variables. This article deploys QCA to examine the same causal relation between historical legacies and democratic transition. Fuzzy-sets QCA analysis shows that non-Islam is a necessary condition of becoming a democratic state, and non-Republics of the Soviet Union before the Second World War is close to necessary condition for the same outcome. For the 20 countries after excluding all 6 Islamic countries, there are five configurations that facilitate democracy consolidation. Among the five configurations, the configuration of statehood and non-Republics of the Soviet Union possesses the most explanatory power, with democratic experience and economy developing level being additional causes. In contrast, religious factors such as Christianity and Orthodox Christianity have no significant impact over outcome. In addition, the influence of Islam may have been over-determined, because the six Islamic countries had been Republics of the Soviet Union before the Second World War, with lower socioeconomic levels and lacking statehood experience. Our results show that with small-N and moderately large-N samples, categorical and ordinal variables, QCA holds some important advantages over conventional regression analysis.