Long-term evaluation of a bimodal universal prevention program: Effects on antisocial development from kindergarten to adolescence

Objectives: Only a handful of studies on developmental crime prevention contain very long-term evaluations and all these addressed high-risk groups in English-speaking countries. In contrast, this article investigates long-term outcomes of a bimodal universal prevention program within the Erlangen-Nuremberg Development and Prevention Study (ENDPS) in Germany. Methods: The ENDPS is a combined prospective longitudinal and experimental project that originally consisted of 675 kindergarten children from 609 families who were nearly representative for the local area. In the prevention part of the project, a group-wise randomization and matched pairs design was used to evaluate a training of children's social problem solving skills, a parent training on positive parenting behavior, and a combination of both programs. Originally, 239 children were each in the program group and control group. Outcomes were measured after ca. 3 months, and 2, 5, and 10 years. The outcome measures varied over time and contained, inter alia, reports on child behavior in the Social Behavior Questionnaire (SBQ) from kindergarten teachers, school teachers, mothers, and the youngsters themselves. The overall retention rates in the ENDPS were satisfactory (e.g., 90 % after 10 years), but missing data from various informants further reduced the groups over time. The outcome evaluation was mainly carried out by causal regression models. Results: There were various desirable effects of the program not only in the short and medium term but also after 5 and 10 years, i.e. on externalizing behavior, property offences, and total behavioral problems. The significant effects were mostly small (d = .23-.59) and significances became rare over time. As a trend, the combined parent and child training and the child training alone were more effective; however, this was not consistent across all follow-ups. The outcomes did not only vary with regard to time but also between different measures and informants. The youth self-reports and teacher reports were more suitable to detect effects than the mothers' reports. Children at higher risk seemed to benefit most from the intervention; however, this was also not fully consistent across measures and times. Conclusions: The various desirable effects of a relatively short and inexpensive universal program are in accordance with a public health approach in developmental crime prevention. However, it should not be seen as an alternative to selective and indicated approaches, but as a 'foot in the door' for high-risk children and families that need more intensive and costly programs. The variations in results across time, outcome measures, informants, and program components confirm the heterogeneity of meta-analytic findings in the field. Therefore, one must be aware of 'fishing for significances' and research on 'what works' should go beyond the content of programs. More studies should investigate how characteristics of program delivery, contexts, participants, and evaluation methods contribute to effectiveness.