Concerns have been raised that cross-agency differences in the definition and measurement of juvenile recidivism may hamper the generalizability of knowledge and comparisons across jurisdictions. However, it is unclear whether measurement choices do impact the conclusions of studies of juvenile recidivism. This study examined whether the rates and the demographic, risk, and contextual predictors of juvenile recidivism varied by the operationalization of recidivism. The sample included 14,537 terms of probation of youths who completed probation in Florida between 2012 and 2016. Recidivism rates differed depending on the type of system contact and the follow-up length. Rates were comparable when adult system data were and were not included. Three-level multivariate multilevel models showed that the predictors were more strongly associated with commitment than with referral or adjudication. The directions and significance of the predictors’ effects were consistent across types of system contact, follow-up lengths, and data sources. Researchers should use varied measurement strategies, clearly describe their approach, and test for robustness across measures.

有人担心，青少年累犯的定义和衡量方面的跨机构差异可能会妨碍知识的普遍性和跨司法管辖区的比较。然而，目前尚不清楚测量选择是否会影响青少年累犯的研究结论。这项研究考察了青少年累犯的比率以及人口、风险和背景预测因素是否因累犯的实施而变化。该样本包括 2012 年至 2016 年间在佛罗里达州完成缓刑的 14,537 名青少年。累犯率因系统接触类型和后续时间长短而异。当包含和不包含成人系统数据时，比率具有可比性。三水平多元多水平模型表明，预测因素与承诺的相关性比与转诊或裁决的相关性更强。预测变量效应的方向和显着性在不同类型的系统接触、随访长度和数据源中是一致的。研究人员应该使用不同的测量策略，清楚地描述他们的方法，并测试各种测量的稳健性。