本研究有四項目的：尋求隨機殺人案定義、比較各國防治政策、研究我國類似案件殺人者特徵，以及提出預防對策。經國內外文獻比較與本研究實證資料分析後，本研究認為，就本案之核心研究對象而言，「無差別殺/傷人」一詞，較「隨機殺/傷人」、「大規模殺人」或「陌生者間殺人」更為適當。無差別殺/傷人案件又可分為典型與非典型兩種，典型無差別殺/傷人案件可定義為：「非因情財仇性等動機，不選擇被害人，或犯罪的時、地之殺人案件」。非典型的無差別殺/傷人案件則可定義為：「因情財仇性等動機，但未選擇特定被害人，也不選擇犯罪時、地的殺人案件。」本研究比較挪威、美國、日本與我國的防治隨機殺人對策發現，挪威認為防治事件再度發生，需重視領導者指揮能力、正確辨識現有風險人口、警方動員能力、應變機制，以及機關間的情報交流與溝通。日本與我國較為類似，更傾向歸因於犯罪人個人之社會、心理、精神疾病因素，而且我國又特別突出殺人犯之毒品使用問題，故政策上傾向於高風險人口的預防、輔導與追蹤。美國因為發生大規模攻擊事件數量大，故發展一套風險評估、管理、分類分級、轉介指導原則，重視科學分析與程序。本研究收集了50份一般民眾問卷為對照組，又在我國八所監獄的467名殺人犯，抽取其中221名收容人施測，其中9%拒訪，共收回211份問卷，2份無效問卷。研究將殺人者分為四組，分析後發現親密殺人組佔30%、熟識殺人組37%、陌生人非隨機殺人組佔28%、陌生人隨機殺人組佔5%。此外，15位無差別殺/傷人受刑人接受本團隊深度訪談、精神評估與心理衡鑑，本研究也取得其歷審法院精神鑑定報告進行比對分析。比較五組暴力犯罪因子，研究發現：（一）各組自編量表中自尊、暴力態度、憤世嫉俗等人格特性，社會疏離感、精神病等均無顯著差異。（二）陌生人隨機殺人者比對照組同理心顯著更低，較多生長在多重高風險家庭、親密關係冷淡、國中小中輟率高，但與對照組的反社會人格、孤獨感、憂鬱症、藥酒癮、少年與成年前科則無顯著差異。（三）同樣是殺人者中，不論哪一種類型，他們的人格特性中有六項相似（自尊、對暴力態度、憤世嫉俗、憤怒、憂鬱、社交焦慮感），其他相似的社會關係與行為，尚有親密關係均淡薄、藥酒成癮性高、中輟率高、出身高風險家庭機率高、少年及成人前科多。根據分析結果，本研究建議建置一套以二、三級預防為主的整合系統，除建置次系統之雲端資料庫，也可設置「社會安全網團隊執行長」一職監督受案、評估、及執行追蹤等處遇流程與成效。本研究也以研究收集到的15位個案將之比對目前已有之教育、社政、少年司法、成人司法、醫療衛生等社會安全網絡，測試一旦系統整合後，這些個案可能在被通報或接受服務機率（亦即命中率）。若以單一網絡命中率而言，成人司法系統中的前科記錄最高（0.73）、精神醫療系統第二（0.6），第三是高風險家庭（.53），少年司法系統中的少年犯或虞犯紀錄與中輟紀錄，各為.47。15名受訪者中，完全未曾出現在前述五種網絡中者約1.4%（漏接率）。若再加上自殺通報網絡，漏接率可降低至0.8%。

The aims of this study are to clarify the definitional issues of indiscriminate murder, analyze relevant policies in different countries, collect evidence in Taiwan and make recommendations. Based on our empirical research findings, ＂indiscriminate murder＂ is our preferred proxy term for ＂random killing＂ or ＂mass/stranger killing＂. There are two sub-categories of indiscriminate murders - classical and non-classicial. Classical indiscriminate murders are cases involving offenders with randomly chosen times, places, and victims to kill without any definite motivation. Non-classical indiscriminate murders are similar however, but with an element of personal motivation. Our literature search and review found that countries defined indiscriminate murder mainly based on their own previous societal experiences and developed policies thereafter. For example, Norway, after the Breivik case, focused on improving the leadership response, identifying risk populations, active police operations, information exchange and communication. Japan, with more than fifty indiscriminate murder cases over the past 10 years, emphasizes the social, psychological and mental status and policies in relation to potential risk populations, while in Taiwan, while mostly sharing a similar approach, also attempts to link drug abuse issues with such crimes. The USA, on the other hand, focuses on mass killing and develops a full set of risk assessment, management, classification, and diversion protocols. In terms of evidence-based policies, this study uses a multi-method research method. A specially designed questionnaire was distributed to a control group (n=50) and four murder inmate groups (n=211, two invalid), namely, indiscriminate murder (5%), stranger murder with motivation (28%), domestic murder (30%), and acquaintance murder (37%). The response rate was 91%. There were fifteen in-depth interviews with indiscriminate murder inmates, with full psychiatric examinations and psychological assessment by the research team. Their previous court psychiatric forensic reports, if any, were provided as supplementary documents for analysis. The study's main findings are, first, there were no significant differences among the five groups on self-esteem, violent attitudes, cynicism, social alienation and mental health; secondly, compared with the control group, indiscriminate murderers had much lower empathy, raised in multiple high risk family situations, failed to form intimate relationships and with high school dropout rates, but no significant differences with other murder groups in terms of anti-social personality, feelings of loneliness, depression, substance abuse, and prior criminal record; thirdly and finally, there are similar personality traits within the four murder groups, in terms of self-esteem, violent attitudes, cynicism, feelings of anger, depression, social anxiety, lack of intimate relationships, substance abuse, drop out, high risk family, and prior criminal records. This finding would suggest there is little sense in developing specific prevention strategies in respect of indiscriminate murder. Our study suggests that an integrated secondary prevention and tertiary prevention network is much more important than primary prevention. Apart from sharing the e-high risk database, we recommend appointment of a high-rank social safety team manager to be in charge of supervising the whole process including in-take, classification, resources allocation and follow-up assessment. The study also tested an assumed integrated model which would have merged the existing six social safety databases in Taiwan with our 15 individual cases. We found that the highest predictor hit rate would be in the adult prior record judicial system (0.73), followed by national mental health records (0.60), high risk family welfare reporting system (0.53), drop out educational reporting system (0.47) as well as juvenile prior record system (0.47). Among the 15 cases, less than 2 percent were unaccounted for in the five social safely systems if all systems work as they are designed. Less than one percent was missing if the national suicide reporting system is included.