腐败是当今人类社会所共同面临的一个世界性问题。自1978年中国改革开放以来，伴随着迈向市场经济的制度转轨，腐败现象不断滋生并呈现出向社会各行业领域蔓延的态势。腐败犯罪行为造成的危害是多方面的，对政治、经济、社会等领域都会带来巨大的负面影响。面对这种严峻的形势，党和国家对腐败问题越来越重视，反腐工作不断深入，反腐力度不断加大。 与此同时，越来越多和腐败相关的问题得到了学界的重视。目前，已经有相当多的研究对腐败的类型、腐败成因、预防手段和治理对策进行了论述，但是对于腐败的惩处，尤其是司法惩处，一直比较缺少系统的实证研究。因此，基于目前中国腐败司法惩治的现状，本文的研究问题是：影响中国腐败司法惩治的因素有哪些?更具体地说，这一大问题可以分解为两个小问题：在司法框架之内，中国目前对腐败司法惩治(量刑)的决定因素有哪些?在司法框架之外，有哪些因素会影响腐败的司法惩治(量刑)? 本研究的数据来源于以下三个部分：第一部分来自于中国裁判文书网的刑事判决书，主要涵盖个体层面的变量，如被判决者的职务、贪污受贿数额、刑期、犯罪类型等个体层面的统计学变量；第二部分来自于中山大学廉政与治理研究中心和国家治理研究院联合发布的(2015——2017年全国廉情评估蓝皮书》，主要涵盖省级层面的腐败数据，包括行贿指数、贪污指数和腐败容忍指数；第三部分数据来源为各省市自治区的统计年鉴和相关政府网站，主要收集的是省级政府层面的经济指标、政府绩效以及宏观层面的地区指标等。通过科学的抽样程序，共收集到4083份判决文书作为本研究的研究样本。由于考虑到中国各省之间可能存在的差异，且本文数据的时间跨度为2015年至2017年，因此本研究的统计模型应该控制年度之间的变化。基于数据的这些特点，本研究采用最小二乘法二分变量模型(LSDV)作为本研究的统计模型之一。此外，本研究的部分解释变量(省级层次变量)和被解释变量(个体层级)存在着不同层次的问题，为了更好地将个体变量嵌入到省级层次变量中，本研究采用多层次线性模型来解决嵌套性数据的问题。 文章基于对4083起腐败案件刑事判决书的实证研究，分析了法定量刑因素以及部分法外因素对量刑结果的影响。研究发现，贪腐案件发生地点对贪污受贿案件刑事量刑的影响比较明显；涉案金额、作案次数和行政级别确实在很大程度上决定了腐败案件的量刑结果；民众对当地腐败现状的认知也会对腐败案件的量刑结果有显著影响。而诸如犯罪人的年龄、工作性质等其他因素对贪污受贿案件刑事量刑的影响有限。 通过判决书分析影响腐败案件刑事判决的影响因素具有重要的理论价值意义。本研究在整合国内外已有研究的基础上，建立了一套“微观+宏观”的嵌入式分析框架。该框架主张，腐败个体不仅受到社会因素的影响，比如其所处城市的经济发展水平、经济增长速度和政府反腐败力度，还受到个体因素和组织层面因素的影响，比如地区社会风气、大众舆论及认知等因素。因此，文章建立了一个宏观与微观相结合、由公众感知相连接的整体性分析框架。此外，文章在客观数据的使用上，比起以往的研究，也有了质的飞跃。文章不仅首次使用了最近刚刚公开的刑事判决文书数据，而且还增加了由中山大学廉政与治理研究中心与国家治理研究院联合调查发布的全国廉情调查数据，并辅之以国家统计局提供的经济、财政和社会发展数据，保证了数据的效度与信度。这样的数据规模和质量在此前的腐败文献中是从未有过的，从而使得文章结论有更强的说服力。 关键词：腐败 腐败案件 司法惩治 刑事量刑

 Corruption is a global problem that all human society is faced with today. Since China's economic reform and open-door policy in 1978, and along with the country's transition to market economy, the phenomena of corruption have been growing constantly and rippled to all industries and areas of the society. The consequences of corruption crimes are multi-faceted, including negative impacts on politics, economy, and society. Facing the serious threats of corruption, the Party and the country have both laid more and more emphasis on this problem, and their anti-corruption efforts have escalated to higher and higher level and expanded to more areas. In the meantime, corruption has also attracted wide attention from the academics. At present, many studies have investigated the types of corruption, the causes of corruption, preventive means and countermeasures. However, there are very few studies that have systematically explored the punishment, especially the judicial punishment of corruption. Therefore, on the basis of the current status of judicial punishment in China, this dissertation raises the following research question: what factors determine the judicial punishment of corruption in China? More concretely, this research question can be divided into two sub-questions: within the judicial framework, what are the determinants of punishment (sentence term) for corruption? And outside the judicial framework, what are the determinants of punishment for corruption? Data for this study came from three sources: the first source was the court documents from the “China Judgements Online”, covering individual level statistical variables such as the rank of the convict, total amount of bribery and embezzlement, sentence terms, type of crime and so forth; the second source was the provincial-level bribery index, corruption index and tolerance index compiled from the 2015-2017 Integrity Assessment Blue Book jointly released by the Research Center of Anti-corruption and Governance and the Research Institute of Chinese Governance at Sun Yat-sen University; the third source was the statistical yearbooks of all provinces in China and their relevant portal websites, which provided provincial-level data on economic development, performance measures and macro-level social data, etc. By conducting a scientific sampling, I obtained a total sample of 4,083 court documents on criminal corruption cases for this research. Given the notable differences among provinces and across time (from 2015 to 2017), the statistical model of this research controlled for both province and time fixed effects. Based on these characteristics of the data, the study employed a least squared dummy variable model (LSDV) as the base model. In addition, the dependent variable of this study (individual level) and some of the explanatory variables (provincial level) are of different data levels. Therefore, I also employed a multilevel regression model to better explain the individual-level dependent variable which is embedded in the provincial-level independent variables. Based on 4,083 corruption criminal cases, this dissertation analyzes the judicial and nonjudicial determinants of final sentencing results. The findings show that the sentencing results of corruption cases vary among different places; the total amount involved, number of criminal accounts and the rank of the defendant all have great impact on sentencing result; the public attitude towards corruption also has significant impact on sentencing results. Nonetheless, the age of the defendant, types of their employment, etc., have little effect on the sentencing results of corruption cases. Analyzing the determinants of sentencing results has great theoretical and practical meanings. This dissertation builds on existing studies and proposes an analytical framework embedded with both micro- and macro-level factors. This framework advocates that corrupted individuals are not only affected by the general social factors such as development level, economic growth and intensity of anti-corruption effort, but are also influenced by the individual and organizational-level social capital including local culture, public media and general cognition, etc. For this purpose, this dissertation establishes a comprehensive framework combining micro and macro views, as well as objective and subjective information. In addition, compared to previous studies, this research has made significant contribution in data collection. First, this dissertation utilizes the most recent judicial sentence reports just made available to the public. This dissertation further utilizes the large-scale national survey data from the Center of Anti-corruption and Governance Studies at Sun Yat-sen University. Last, this dissertation also includes object economic, fiscal and social data from the National Bureau of Statistics. Combining data from various credible sources not only improves the quality of the data but also increases the validity of the data analysis results. To date, no previous corruption study has used datasets of such scale and quality. Therefore, the conclusions of this dissertation are more reliable and convincing than most previous studies of similar nature. Key words: corruption, corruption cases, judicial punishment, criminal sentence