善恶在中国文化中有着悠久的历史和渊源，在心理学领域研究者也不断地开展与善恶相关的研究。基于对以往研究的综述，本研究将善恶人格定义为在遗传与环境的交互作用下，个体形成的具有社会道德评价意义的内在心理品质。研究从人格心理学的视角出发，探讨了善恶人格的基础结构、特质推断特征及善恶人格的功能，以期多角度、深层次理解善恶心理。

本论文通过3个大研究共16个子研究，拟从理论和实践方面解决三个主要问题：第一，善恶人格的结构：善、恶人格是单维还是双维？其内在心理结构的构成要素是什么？第二，善恶人格的层次：善/恶人格的不同特质对人们进行善恶判断的影响力或核心程度是否存在差异？第三，善恶人格的功能：善/恶作为人格因素具有什么作用？

第一部分，善/恶人格的结构。本部分研究聚焦于人格的结构，共包含4个研究。研究1采用词汇学方法，从《现代汉语词典（第7版）》中获取3,240个有关善恶的词汇，并基于开放问卷调查获取374个有关善恶的词汇，进行全面分析，建立了表征善/恶人格的基础词库。研究2通过2个子研究探讨了善、恶人格的单双维结构。研究2a在词库的基础上，采用平行分析、探索性因素分析确定合适的因子数，结果发现抽取2个因子更合适；进而，采用验证性因素分析比较了善、恶人格不同模型的拟合度，结果发现：与善、恶人格作为一个维度的两极相比，善、恶人格作为两个维度模型的拟合度更好。研究2b通过分析善人格、恶人格和其他人格效标之间的关系，证明了善、恶人格与效标相关模式存在不同，善人格与利他人格显著正相关，但恶人格与利他人格相关不显著，再次验证善、恶人格的双维结构。研究3基于特质词汇，通过2个子研究分别探究了善人格的结构和恶人格的结构。研究3a从善人格中提取出4个因子：尽责诚信、利他奉献、仁爱友善、包容大度；研究3b从恶人格中提取出4个因子：凶恶残忍、虚假伪善、污蔑陷害、背信弃义。此外，两个子研究所获得的善/恶人格词汇评定量表皆具有良好的信效度。研究4基于行为表现，通过4个子研究分别探究了善人格与恶人格的结构，并考察了善、恶人格的关系和人口统计学差异及该测量工具的效度。研究4a和研究4b针对每个善/恶人格词汇编写6~9个相应描述具体外在表现的句子，通过多轮筛选、整合，获得初步测量善/恶人格的条目。随后将这些条目编制成善/恶人格问卷施测，通过因素分析，研究4a发现善人格包含4个因子：诚信、利他、友善、大度；研究4b发现恶人格包含4个因子：自私、虚假、陷害、背叛，两个研究获得的自陈量表均有较好的信效度。研究4c对731名被试进行施测，再次验证了善/恶人格的四因子结构，并发现人们对善、恶的自评得分并不是简单的此高彼低，还存在自评得分皆高或皆低的模式，这也从另一个角度支持了善、恶人格的双维结构。对不同人口学特征个体进行善/恶人格的差异分析，结果发现年龄对善/恶人格的影响效应更大，但性别对于恶人格的“背叛”维度影响更大。研究4d通过考察善/恶人格与一般人格模型（HEXACO）对亲社会行为和精神病态预测作用的差异，进一步证明了善/恶人格测量工具有较好的效度，反映出使用该工具测量善/恶人格结构的有效性。研究发现，善/恶人格与HEXACO对预测指标的线性回归模式并不对等，善人格对个体的亲社会行为的预测能力更强，HEXACO的宜人性对共情的预测能力更强；相比HEXACO的各人格维度，恶人格对精神病态的预测作用更大。研究证明了善恶人格和一般人格模型存在相关，但仍有其独特的成分。

第二部分，善/恶人格推断的特质差序。本部分聚焦于善恶推断时特质的核心性，即人格结构的层次，共包含4个研究。研究1探讨了善、恶是否存在差序。研究结果发现，当需要选择属于道德范畴类的词汇时，被试选择善人格特质的词汇数量显著高于恶人格特质的词汇数量，善、恶人格特质被选择的次数都大于非道德类型特质。研究2通过2个子研究分别探讨了善/恶人格不同的特质对判断个体是“善”或“恶”的代表性（即有用程度）差异。研究2a结果发现，当判断人是“善”时，各维度的代表性从高到低分别为：尽责诚信 > 仁爱友善 ≈ 包容大度 > 利他奉献。研究2b结果发现，当判断人是“恶”时，各维度的代表性从高到低分别为：凶恶残忍 > 背信弃义 ≈ 污蔑陷害 > 虚假伪善。研究3以好恶度和特质度作为衡量标准，再次验证当人们进行善/恶判断时，善/恶人格不同维度的核心程度存在差异。其中，特质度是指人们确认一种特质所需要的最小行为频率，特质度高说明判断该特质所需要的行为频率高，反之，则低。研究结果发现，人们对整体善、恶人格的好恶度和特质度评价结果一致，善人格显著高于恶人格；对善人格各维度的好恶度和特质度评价结果一致，从高到低分别为：尽责诚信 > 仁爱友善 ≈ 包容大度 > 利他奉献；对恶人格各维度的好恶度和特质度评价结果也一致，从低到高分别为凶恶残忍 < 背信弃义 ≈ 污蔑陷害 < 虚假伪善。研究4以重要性为指标，采用对特质重要性排序的方式验证善/恶人格判断的差序性，结果发现，当需要判断一个人有多善时，尽责诚信的重要性显著高于利他奉献；当需要判断一个人有多恶时，凶恶残忍的重要性显著高于虚假伪善。

第三部分，善/恶人格的功能。本部分聚焦于人格的应用，从人格对他人影响（与他人相关的行为及行为倾向，如助人行为）和自我影响（自我的身心健康，如幸福水平）两个角度探究善/恶人格的功能，共包括2个研究。研究1采用追踪研究的方法，使用交叉滞后模型，探究从新冠疫情爆发、被初步控制到被稳定控制3个时间段善人格与幸福水平、积极情绪、消极情绪和助人行为的纵向关系。结果发现：善人格可以显著正向预测同时间段及下一时间段的幸福水平、积极情绪和助人行为，善人格显著预测同时间段的消极情绪，在疫情初步得到控制至得到稳定控制之间，显著负向预测下一时间段的消极情绪。研究2探索了恶人格对攻击行为的影响，结果发现恶人格可以显著预测个体的攻击行为。

善恶人格定位于人格心理学领域的基础性研究，作为善恶或道德研究这一“大厦”的“根基”，发挥着重要作用。在理论方面，本研究突破以往只关注积极或消极道德品质的单维视角，采用双维视角研究善、恶心理，并在此基础上深入地探讨了善/恶人格的结构和层次，对人的善恶品性进行了较为细致地刻画，丰富、完善了与善恶相关的人格理论。本研究与传统的道德心理学、人格心理学等领域研究内容相呼应，为探究人性善恶问题提供了新的依据和思路。在实践方面，本研究建立了善、恶人格的词库与测量工具，一定程度上有助于测量个体在善恶问题上的基本差异性倾向，为探讨善恶人格的后续理论与实际研究奠定了扎实的基础。此外，本研究初步探讨了善/恶人格的功能，并发现善/恶人格对道德相关的行为及倾向具有独特的预测作用。研究从真实的灾难事件（新冠疫情）中发现了善良人格的价值，使得研究更加具有生态效度，对个体与社会的健康发展具有重要的应用价值。

Good and evil characteristics are two aspects of Chinese culture which have a long history and origins. Abundant descriptions of good and evil characteristics can be found in human nature. Researchers in the field of psychology have also carried out research on good and evil, in order to understand good and evil from varied aspects and perspectives. This study defined good and evil personalities as two internal psychological qualities with meanings of social moral evaluations, and they are formed in the process of socialization. This study integrated the research perspectives of personality psychology and social psychology, discussed the basic structure, measurement methods, trait inference characteristics, and psychological application of good and evil personalities, so as to understand the good and evil personalities in a more detailed and in-depth way.

This study aimed to answer three questions via four studies with a total of 17 sub studies. First, are good and evil personalities two sides of one dimension, or two separate dimensions? And what is the inner psychological structure of good and evil personalities? Second, are there differences in the influence of dimensions of good and evil personalities on people’s judgment of good and evil? Third, how does good personalities affect the public’s mental health in the period of the COVID-19 epidemic?

Part one, the structure of good and evil personalities. This part contains four studies. The lexical approach was used to determine the constitution of good and evil personalities in Chinese culture. A set of 3,614 good and evil moral character descriptors (i.e., moral and immoral character traits) were selected from The Contemporary Chinese Dictionary (7 th) and accompanied by daily life expressions and were ultimately condensed into a basic lexicon of good and evil personalities (Study 1). Study 2 tested for the dimensionality of good and evil personalities. 700 participants were randomly split into two samples, exploratory factor analysis (EFA) and parallel analysis (PA) with the first split sample favored a two-factor over a one-factor solution. In confirmatory factor analysis (CFA) with the second split sample, the two-factor structure showed satisfactory goodness-of-fit while the one-factor structure showed poor data fit (study 2a). Additionally, evidence of the two-dimensional structure based on correlations with Honesty-Humility, Dirty Dozen and altruistic personality was demonstrated (study 2b). In Study 3, the dimensions of a good personality and an evil personality explored using EFA and CFA. Results revealed that there were four dimensions of the good personality: conscientiousness and integrity, altruism and dedication, benevolence and amicability, and tolerance and magnanimity. And there also were four dimensions of the evil personality: atrociousness and mercilessness, mendacity and hypocrisy, calumniation and circumvention, and faithlessness and treacherousness. Moreover, the internal consistency, reliability, and criterion-related validity of the good and evil personality lexical rating scales were verified. In Study 4, three sub studies were used to confirm the dimensions of a good personality and an evil personality by developing the self-report scales of good and evil personalities, and to explore the relationship between good and evil personality and their demographic differences. 15 good behaviors and 12 evil behaviors were reserved after analysis and integration. Results revealed that good personality contained four dimensions: conscientiousness, altruism, amicability, and magnanimity. The evil personality contained four dimensions: selfishness, hypocrisy, circumvention, and treacherousness. Both good and evil personality scales had good reliability and validity, they were effective tools that measure the good and evil personalities of Chinese people (Study 4a and 4b). Study 4c verified the structures of good and evil personalities, and showed that people’s self-evaluation of good and evil personality was not completely opposite, which also verified the two-dimensional structure of good and evil personalities. Results showed that age had a greater impact on good and evil personalities, and gender has a more significant impact on the betrayal dimension of evil. Study 4d explored the function of good and evil personality via examining the relationship between good and evil personality and general personality model (i.e., HEXACO). It was found that the linear regression models of good/evil personality and HEXACO on the prosocial behavior and antisocial tendencies were not completely equivalent. Good and evil personalities were related to HEXACO, but they still contain the unique moral components.

Part two, the differential patterns of good and evil trait inference. This part contains four studies. Study 1 investigated whether there was a differential order between good and evil personalities. The results showed that when asked to select which traits belonging to morality, the frequency of good personality traits that participants selected was more than evil personality traits. Both good and evil personality traits were selected more often than non-moral traits. In Study 2, two sub studies were conducted to explore the differences in the representativeness (i.e., the degree of usefulness) of different dimensions of good or evil personality in judging whether an individual is “good” or “evil”. The results of study 2a showed that when people judged a person as “good”, the representativeness of each dimension from high to low was: conscientiousness and integrity > benevolence and amicability ≈ tolerance and magnanimity > altruism and dedication. The results of study 2b showed that when people judged a person to be “evil”, the representativeness of each dimension from high to low was: atrociousness and mercilessness > faithlessness and treacherousness ≈ calumniation and circumvention > mendacity and hypocrisy. In Study 3, preference and the scope of trait were used as the measurement standards to verify the differential patterns of good and evil trait inference. The scope of trait describes the minimum behavior frequency that people need to identify a trait. Results showed that preference and the scope of trait of each dimension were consistent to Study 2. The differential patterns were also demonstrated by ranking method in Study 4. When people need to judge the good and evil degrees of one person, they thought that the conscientiousness and integrity dimension was more important than the altruism and dedication; and the atrociousness and mercilessness dimension was more important than the mendacity and hypocrisy.

Part three, the function of good and evil personalities. This part includes two studies. A structural equation modelling (SEM) approach was used to investigate the longitudinal cross-lagged relationships between good personality and happiness, positive affect, negative affect, and prosocial behavior. Results showed that the higher levels of good personality positively predicted an increased happiness, positive affect and prosocial behavior, and the good personality at peak stage could negatively predicted the negative affect at steadily controlled stage (Study 1). Study 2 explored the influence of evil personality on aggressive behaviour. The results verify that evil personality can positively predict aggressive behaviour.

The construction of good and evil personalities is positioned as the basic research in the field of personality psychology, plays a role as the “foundation” of the “mansion” of morality, enriches previous morality-related personality theories, improves traditional personality theories, and makes a detailed description of the good and evil personalities of human beings. This not only provides a new basis for exploring the problems of good and evil in human nature, but also can be related to the psychology of morality and virtue. At the same time, this study established two sets of measurement tools of good and evil personalities, which can help to measure individual's basic difference tendency on the issue of good and evil. Based on the basic structure of good and evil personalities, this study also explored the differential patterns of good and evil trait inference with multiple measurement indicators, which points a new direction for the theoretical research of good and evil personalities in trait inference. In addition, this study found the value of good personality from the real disaster event (COVID-19), which made this research more ecological validity. This result has important application value for the healthy development of individuals and society.