2026-10-14T14:41:10Z https://www.scidb.cn/oai

10.57760/sciencedb.46333 2026-08-21T17:13:17Z

2026-08-21 A Survey Dataset on Multidimensional Risk Assessment for Juvenile Delinquency in China doi:10.57760/sciencedb.46333 en This dataset originates from a survey study on risk factors for juvenile delinquency in China. Data were collected from March 2025 to December 2025 across five educational institutions in Chongqing—including three specialized correctional schools for adolescents with serious misbehavior, one regular public middle school, and one secondary vocational school—as well as through community-level police stations that referred adolescents with documented delinquency records.The survey employed a self-developed instrument, the Juvenile Delinquency Risk Information Questionnaire, which was developed through three-level grounded theory coding of semi-structured interviews with 15 students from specialized correctional schools, and was finalized following expert content validity review. The questionnaire comprises 80 binary-scored items (0 = no, 1 = yes), covering 8 demographic variables, 16 environmental risk items, and 56 individual psychological and behavioral risk items . Data were collected through group-administered, anonymous paper-and-pencil surveys, with standardized instructions delivered by trained researchers in classroom or designated settings, and questionnaires were collected on-site upon completion.The data are stored in Microsoft Excel (.xlsx) format, consisting of 5 worksheets with a total file size of approximately 3.2 MB:(1) The first three worksheets contain raw data from the three participant sources, totaling 929 responses (305 from the regular middle school, 307 from the vocational school, and 317 from the specialized correctional schools; these participants are distinct from the qualitative interview sample). These data remain unprocessed—invalid responses have not been excluded, and reverse-coded items have not been recoded. In each worksheet, rows represent individual cases, and columns sequentially represent case number, ID, demographic variables (8 columns), environmental risk items (16 columns), and psychological risk items (56 columns).(2) The fourth worksheet contains the cleaned analytical dataset prepared by the authors for machine learning modeling. Based on the raw data, this sheet includes an additional column, "Risk Type," serving as the outcome variable. Data cleaning has been completed: invalid responses have been excluded (yielding 840 valid responses, with a validity rate of 90.4%), and reverse-coded items have been recoded. The data contain a small proportion of missing values (less than 5% of total items), and no imputation has been performed on the uploaded data. Among the valid sample, the high-risk group consists of 305 adolescents who have committed delinquent acts or serious misbehavior (recruited from specialized schools and community channels), while the low-risk group comprises 535 students from regular middle schools and secondary vocational schools.(3) The fifth worksheet provides a complete codebook, including variable names, coding schemes, and indications of reverse-coded items. juvenile delinquency; risk assessment; questionnaire survey; machine learning Zhang Xuan RESTRICTED dataset Science Data Bank