詐騙集團車手、機手與對應被害人關聯性之質化與量化分析 -以真人假愛情交友詐騙類型為例

Abstract

依據「111年中華民國刑案統計」，詐騙類型計有38種，以當年度破獲嫌疑人數前5名為例，分別為假投資、解除分期付款、假網路拍賣、購物詐欺及假愛情交友，本論文是研究假愛情交友類型中，有真人車手佯裝戀人，第一線與被害人約會、交往，甚至性交易之類型，而非純網路戀愛，使被害人在網路上陷入虛擬熱戀，但卻從未見面，無真人現身類型。 本研究資料是從警政165反詐騙系統平台取得2016-2024年間偵破真人假愛情交友詐騙集團案件資料，總計14團詐騙集團，被害人275名，機手 83名，車手 67名，從刑事案件報告書、筆錄與附件資料中擷取被害人及對應之車手、機手相關資料、被害及犯罪特徵，而找到對應被害人之機手是建立資料庫中最重要也最難一環，因為此類詐騙集團常是多名機手搭配同1名車手詐騙，事後釐清當時究是哪名機手詐騙某位被害人頗有難度，須不斷爬梳案件文書、相互交叉比對始能找出對應機手，被害人部分包含星座、年齡、現住地、現住地至機房距離、職業、教育程度及經濟水準等基本資料，及被害人主動報案意願、遭詐金額、遭詐時間、遭詐頻率、第1次遭詐金額與第1次遭詐時間等被害紀錄；詐騙集團部分包含機房所在地與機手、車手之星座、年齡、教育程度、經濟水準、薪資與犯罪年資等資料，逐一判讀與建檔完成資料庫，據此編碼、分組並定義成各個變數，設計實證模型，分析上述相關變數與被害人詐騙金額、詐騙時間及主動報案意願之相關性，另分析假感情詐騙為何無法根斷?詐騙得逞關鍵?詐騙集團被檢警偵破關鍵?從經濟學視角切入真人假愛情交友詐騙犯罪，初探此類犯罪全貌，目標是從中得到對警政、學術有貢獻之發現與結論。 實證結果發現，以被害人遭詐金額、遭詐時間與主動報案意願為應變數，經控制被害人、機手與車手等多項變數後，許多變數具顯著性，例如雙魚座遭詐金額與遭詐時間分列全星座第1及第2，學生遭詐時間最久，是須受高度重視、關懷及犯罪預防宣導，且須家長及時介入之族群，而車手年齡越大，被害人遭詐騙時間越短、犯嫌薪資與犯罪年資可提高詐騙時間及降低被害人主動報案意願……等也具有10%顯著水準可供參考。 本論文也採質化分析，訪問目前已知之臺灣真人假愛情交友之最早型態-剝皮酒店之早期高階幹部，得知剝皮酒店可能由來、早期運作模式、經營成功關鍵，有助於瞭解臺灣真人假愛情交友詐騙歷史演進過程，並分析破獲真人假愛情交友詐騙集團資料，釐清集團間差異處、共同點及特徵，並進而推演現今真人假愛情交友可能處境，進而研判未來發展趨勢。

According to the 2022 Criminal Statistics of the Republic of China (Taiwan), there are 38 recognized types of fraud. Among them, the top five categories with the highest number of suspects apprehended that year were: fake investment scams, installment payment cancellation scams, fake online auctions, shopping frauds, and romance scams. This thesis focuses specifically on the real-person romance scam subtype, in which members of fraud syndicates—particularly money mules—impersonate romantic partners in person. These individuals meet, date, and even engage in sexual activity with victims, distinguishing this form of scam from typical online romance scams, where victims fall in love in a purely virtual setting without ever meeting the scammer face-to-face. The dataset for this research was obtained from Taiwan’s National Police Agency’s 165 Anti-Fraud Reporting System, covering cases of dismantled real-person romance scam groups from 2016 to 2024. A total of 14 fraud syndicates were identified, involving 275 victims, 83 scammers, and 67 money mules. Data were extracted from official criminal investigation reports, interrogation transcripts, and case file attachments. The most difficult yet critical part of building this database was correctly identifying the corresponding scammer for each victim, as multiple scammers often collaborate with a single mule to defraud multiple victims. This required extensive document review and cross-referencing of case files. Victim information included basic demographics such as zodiac sign, age, current residence, distance from residence to the scam group’s base of operation, occupation, education level, and economic status. Victim-specific behavioral data were also collected, including willingness to report the crime, total defrauded amount, duration and frequency of scams, as well as details about the first defrauded transaction. Fraud syndicate data included the scam group’s location and the zodiac sign, age, education level, economic status, monthly earnings, and criminal experience of both scammers and money mules. These variables were carefully coded and compiled into a comprehensive dataset for empirical modeling. The analysis examined how the above variables relate to three main outcomes: the amount of money defrauded, the duration of the scam, and the victim’s willingness to report the crime. In addition to quantitative analysis, the study conducted qualitative interviews with former senior operatives of Taiwan’s earliest form of real-person romance fraud, commonly known as “Romance Scam Hostess Bars”. These interviews provided insights into the origins, early operations, and key success factors of this industry, helping to trace the historical evolution of real-person romance scams in Taiwan. Comparative analysis of dismantled scam syndicates further revealed similarities, differences, and shared characteristics, which were used to infer current developments and future trends in this form of fraud. Empirical findings suggest that several variables have statistically significant associations with scam outcomes at the 10% level. For instance, Pisces victims experienced the highest total defrauded amounts and ranked second in scam duration among all zodiac signs. Occupation also showed strong significance: students were identified as a high-risk group requiring focused prevention efforts and timely parental intervention. Additionally, older mules were linked to shorter scam durations, while higher scammer earnings and longer criminal experience were associated with longer scam durations and lower victim reporting willingness—results all significant at the 10% level. From an economics perspective, this study provides an in-depth exploration of real-person romance scams as a criminal phenomenon. It sheds light on why such scams persist, the key mechanisms through which they succeed, and the factors that contribute to their eventual dismantling by law enforcement. The goal is to offer meaningful insights for both academic research and law enforcement policy development.