由於受保護管束人更生復歸不易，復以我國以橫斷性探討犯罪原因的研究為主，鮮少採縱貫性研究方法來探討犯罪者之再犯情形，為此，本研究彙整近年來國內外相關犯罪學理論及實證研究，並立基於對成年假釋人回到社會後受保護管束期間之問卷調查，藉由長達15年之追蹤，期經由客觀的實證分析，探求成年假釋人再犯及其風險因子，並擬定預防策略。本研究資料來源為國科會於2000年至2002年委託陳玉書、林健陽博士主持「再犯預測：生活壓力、社會控制與社會學習對成人再犯之影響及其預防對策」專題研究之2次問卷調查資料，以研究中之假釋受保護管束人共959人為研究對象，經比對發現計66人死亡，為避免影響分析結果，爰將死亡人數扣除，有效樣本為893人。由於問卷係在假釋人受保護管束期間進行施測，更能貼近其真實生活，復再結合2015年底以前之官方犯罪紀錄，進行為期15年的縱貫性調查，得以觀察樣本動態的變化。此外，使用SPSS 統計軟體，採用次數分配、獨立樣本t檢定、卡方（χ2）獨立性檢定及二元Logistic 迴歸分析等統計考驗比較。研究發現：(1) 追蹤15年後，樣本中再犯者528名，再犯率為59.1%。(2) 性別、年齡、教育程度、初犯年齡、過去藥物濫用經驗、自陳偏差及犯罪經驗、低自我控制、家庭附著與衝突、偏差友儕和遊樂生活型態等，與有/無再犯有顯著關聯或差異性。(3) 性別、年齡、教育程度、入監服刑次數、藥物濫用種類、遊樂生活型態及偏差友儕為有/無再犯之風險因子。最後， 根據研究結果提出成年犯罪者更生復歸和再犯預防等相關建議。

Because it is not easy to rehabilitate with people under custody, and most study about the causes of crimes are through cross-sectional method, rarely research on the recidivism of the offenders through longitudinal research method in Taiwan. Therefore, this study has gathered domestic and foreign relevant criminological theory and empirical research in recent years, based on a questionnaire survey on the period of adult parolee after returning to society, through a 15-year follow-up survey and objective empirical analysis, to explore the recidivism of adult parolees and their risks factors, then develop prevention strategies. The research materials are two questionnaire survey data of " the prediction of recidivism: the impact of life stress, social control and social learning on adult recidivism and its preventive measures", the project Chen Yushu, Lin Jianyang, etc. commissioned by the National Science Council from 2000 to 2002. A total of 959 adult people under custody in the study, 66 were found to have died. The number of deaths was deducted to avoid it’s impact on the analysis results. The effective sample was 893 people. Since the questionnaire was tested during the period of paroles under custody, it is closer to theirs real life situation. Combined with the official criminal record before the end of 2015, a 15-year longitudinal survey was conducted to observe dynamic changes within these samples. In addition, the SPSS statistical software is used, and statistical tests such as frequency distribution, independent sample t-test, Chi-Squared test of independence, and binary logistic regression analysis are used. The study found that: (1) After 15 years of follow-up, 528 re-offenders in the sample, and the recidivism rate is 59.1%. (2) Gender, age, education level, age of first offense, past drug abuse experience, self-reported delinquent and criminal experience, low self-control, family attachment and conflict, deviant peers, and recreational lifestyle, have significant correlation or difference with the recidivism. (3) Gender, age, education level, number of imprisonment, type of drug abuse, recreational lifestyle, and deviant peers are risk factors of recidivism. Finally, based on the research results, the rehabilitation and prevention recidivism to adult offenders are proposed.